Power distribution equipment state visualization platform based on big data

Big data-based distribution equipment status visualization platform

Technical field

[0001] The present invention relates to the field of device management technologies, and in particular, to a power distribution device state visualization platform based on big data.

Background technique

[0002] At present, major power grid companies mainly rely on the production management system (power production management syStem, PMS) for information management of power grid equipment. PMS can provide most of the information needed for equipment management, including equipment ledgers, defects, status evaluation, maintenance tests, online monitoring and other data. However, from the perspective of strengthening the overall control of equipment status, the effective information provided by PMS is still very limited, and the sensitivity to state changes and risks is still far from enough, and the management efficiency and effect are not high.

[0003] In recent years, equipment failures caused by quality problems have occurred from time to time. At the same time, extreme weather such as thunderstorms and squall lines have been affected by global climate change, and the impact on the safe operation of power grid equipment has become increasingly prominent. Therefore, it is urgent to have a management platform that can combine the information provided by real-time systems such as scheduling, weather, video surveillance, etc., and add advanced intelligent diagnosis and auxiliary analysis functions to better meet the operational needs of the inspection and inspection department.

Summary of the invention

The present invention is directed to at least one of the above technical problems.

[0005] Therefore, the object of the present invention is to provide a data visualization device state visualization platform based on big data, which can meet the business requirements of intensive development and lean management, improve the multi-source information interaction and fusion capability of the device, and realize Panoramic real-time sensing and multi-dimensional intelligent monitoring and control of power distribution equipment, prevent equipment operation risks in advance, improve response speed to emergencies, and make equipment life cycle management transparent and efficient.

[0006] In order to achieve the above object, an embodiment of the present invention provides a power distribution device state visualization platform based on big data, including: a data processing module, configured to acquire multi-platform data, and process the multi-platform data. And displaying the processed data; a data analysis module for performing big data integration, storage, retrieval, and data mining analysis; an evaluation module for generating a big data-based power distribution equipment evaluation model, and according to the power distribution device The evaluation model evaluates the power distribution equipment and generates a corresponding processing strategy based on the evaluation results.

In addition, the big data-based power distribution equipment state visualization platform according to the above embodiments of the present invention may further have the following additional technical features:

[0008] In some examples, the multi-platform data includes at least: production management system data, online monitoring system data, spatial geographic information system data, weather system data, and video monitoring platform data.

[0009] In some examples, the data processing module is configured to perform pre-processing and cleaning on the acquired multi-platform data, including: applying a unified specification according to the service system, type, structure, and size of the multi-platform data. The tag is used to identify the source and type of the data. At the same time, according to the preset data rule base, according to the tag of the data, the corresponding rule and data are encapsulated, and the encapsulated data can be identified, controlled and correspondingly Cleaning rules can be sent to the data cleaning stage for cleaning.

[0010] In some examples, the data analysis module includes a sensing layer, a network layer, and an application layer, wherein the sensing layer is configured to perform data collection; the network layer is configured to perform data transmission; and the application layer further The service layer, the business layer, the presentation layer, and a tool set are used to provide mining and analysis capabilities of data, the business layer is used to implement business requirements of a specific product, and the presentation layer is used to provide an interaction interface. The tool set is used to provide installation deployment tools, data mining tools, business modeling tools, and code generation tools.

[0011] In some examples, the sensing layer, the network layer, and the application layer interact, the interaction including a message flow and a data flow, and the processing of the data flow is controlled by the message flow.

[0012] In some examples, the power distribution equipment evaluation model includes at least: a transformer-type equipment failure prediction model, a development trend of a switch and a combined electrical equipment state, and a failure probability dynamic prediction model, and a transmission line fault based on a complex correlation relationship. Forecast model.

[0013] In some examples, the evaluation module is configured to adopt a fusion multi-factor state evaluation analysis algorithm, including: 1) analyzing a decision problem, constructing a system's proposition set, that is, the system identification frame Ω = {A1, A2 ,......, Ak}; 2) For the target information system, construct the evidence body Ei (i = l, 2, ..., m) based on the recognition framework; 3) According to the collected data of each evidence body, a global total The data, combined with the characteristics of each proposition set in the recognition framework, determines the basic credibility distribution of each evidence body 11^ "), "=1, 2, ..., K, indicating the ability of different state information to respond to the state of the device. 4) According to the basic credibility distribution mi (Aj), calculate the reliability interval of each proposition in the recognition framework under the action of the single evidence body [Beli, Pli]; 5) use the DS synthesis rule to calculate the joint action of all evidence bodies The basic credibility distribution m (Aj) and the reliability interval [Bel, Pl]; 6) construct corresponding decision rules according to specific problems; 7) draw decision conclusions according to the decision rules.

[0014] In some examples, the evaluation module evaluates the power distribution device, including: A) scanning the data one by one according to each state quantity threshold in the corresponding guide according to the relevant requirements in the power distribution device status evaluation guide. When any one of the data exceeds the threshold defined in the guide, the data is marked as an outlier and separated from the original data; B) the data is transformed into a multivariate time series, and the cross-covariance function and mutual of each one-dimensional time series are calculated. Correlation function, which obtains the order and delay parameters of the transfer function numerator and denominator polynomial, then fits the transfer function model, and finally determines the interference time and the generated abnormal data according to the ACF test of the model residual sequence; C) The least squares regression parameter estimation and the generalized likelihood ratio change point detection are used. The incremental mechanism is used to determine the parameters and segmentation points of the data series regression model, and the data trend characteristics are extracted in real time, and the data of the trend change is marked as abnormal data.

[0015] In some examples, wherein the failure rate of the distribution line under different weather conditions is the number of times the time is converted into a unit when the failure occurs, and the average value of the failure rate is expressed in units of one calendar year. for:

[0016]

Where N is the expected duration of normal weather and S is the expected duration of bad weather. Indicates the expected value of the component failure rate in normal weather, 'the expected value of the component failure rate in bad weather;

[0018] The fault rate and cumulative probability distribution function of the transformer can be expressed as:

[0019]

[0020]

[0021] Wherein, the model assumes that the parameters β and C remain unchanged at different temperatures, and the two-state weather model is used to describe the accidental failure mode failure rate of the transformer, the expression of which is:

[0022]

[0023] wherein

For the statistical average of accidental failure of the transformer, N is the duration of normal weather, S is the duration of bad weather, F is the proportion of faults occurring in bad weather, w is the current weather condition of the transformer, normal weather W = 0, bad weather w=l.

[0024] In some examples, the evaluation module is further configured to perform device importance evaluation according to device status and system risk, including: a) using a PHM model calculation system according to a big data status evaluation result, operation information, and micro weather data. The component considers the real-time failure probability of big data; b) uses the enumeration method to select the system state, enumerates to the third-order fault, forms the expected fault event, and calculates the probability of the fault event occurring; c) performs static security analysis on the selected system state Using the optimal power flow to calculate whether the state of the system satisfies the abundance. If the load is required, the system state is an emergency state, and the process proceeds to step d). If the load is not required, the system state is an alert state or a healthy state, and the system is subjected to NI. Verification, if the safety criterion is met, it is healthy, return to step b), otherwise it is in the alert state, go to step d); d) calculate the emergency index or number of police rings in the system state, calculate the state using the risk tracking model The contribution value of each faulty component; e) returning to step b) until traversing all fault events of the expected fault set; f) calculating the system total emergency finger And the total index of vigilance, and calculates the urgent and important index and element index important alert, sorted according to the importance index, determine the system equipment is weak.

[0025] The big data-based power distribution equipment state visualization platform according to the embodiment of the present invention maximizes reuse of various distribution equipment basic data and maintenance management resources, and utilizes data integration through multi-dimensional visual display functions. And data mining technology to achieve comprehensive analysis of equipment, for the transmission, substation equipment to carry out state evaluation, fault diagnosis, risk assessment, life prediction and maintenance decision-making work, assist with technical supervision and intelligent report management, provide a panoramic, real-time for the operation and maintenance department Multi-dimensional, intelligent equipment management and control information platform, so that a large number of new, old and different, different state of the equipment is always under close monitoring, prevent equipment operation risks in advance, improve the response speed to emergencies, and meet intensive development Lean management business needs, improve the multi-source information exchange and fusion capabilities of equipment, realize panoramic real-time perception of power distribution equipment, multi-dimensional intelligent monitoring and control, and make the life cycle management of equipment transparent and efficient.

[0026] Additional aspects and advantages of the invention will be set forth in part in the description which follows in the Description

DRAWINGS

[0027] The above and/or additional aspects and advantages of the present invention will become apparent and readily understood from

1 is a structural block diagram of a big data based power distribution equipment state visualization platform according to an embodiment of the present invention;

2 is a schematic diagram of a data cleaning process in accordance with an embodiment of the present invention;

3 is a schematic diagram of a fault prediction process of a transformer-type device according to an embodiment of the present invention;

4 is a schematic diagram of a fault diagnosis process of a transmission line according to another embodiment of the present invention;

5 is a schematic diagram of a state evaluation analysis algorithm for merging multi-factors according to an embodiment of the present invention;

6 is a schematic diagram of a process for evaluating a differentiated state of a transformer-type device according to an embodiment of the present invention;

7 is a schematic diagram of a rapid evaluation process of a device state according to an embodiment of the present invention.

detailed description

The embodiments of the present invention are described in detail below, and the examples of the embodiments are illustrated in the drawings, wherein the same or similar reference numerals are used to refer to the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the drawings are intended to be illustrative of the invention and are not to be construed as limiting.

[0036] In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right" are understood. The orientation or positional relationship of the indications such as "vertical", "horizontal", "top", "bottom", "inner" and "outside" is based on the orientation or positional relationship shown in the drawings, only for the convenience of description. The invention and the simplification of the description are intended to be illustrative and not to be construed as a limitation of the invention. "For purposes of description only, it is not to be construed as indicating or implying relative importance."

[0037] In the description of the present invention, it should be noted that the terms "installation", "connected", and "connected" are to be understood broadly, and may be fixed connections, for example, or It is a detachable connection, or is integrally connected; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and may be internal communication between the two elements. The specific meaning of the above terms in the present invention can be understood in a specific case by those skilled in the art.

[0038] A big data based power distribution equipment state visualization platform according to an embodiment of the present invention will be described below with reference to the accompanying drawings.

1 is a structural block diagram of a big data based power distribution equipment state visualization platform according to an embodiment of the present invention. As shown in FIG. 1 , the big data-based power distribution equipment state visualization platform 100 includes a data processing module 110 , a data analysis module 120 , and an evaluation module 130 .

[0040] The data processing module 110 is configured to acquire multi-platform data, process the multi-platform data, and display the processed data. The multi-platform data includes, for example, at least production management system (PMS) data, online monitoring system data, spatial geographic information system (GIS) data, weather system data, and video monitoring platform data.

[0041] In other words, the data processing module 110 can implement multi-platform data acquisition. For example, the power distribution equipment status visualization platform needs to integrate multiple systems horizontally and access its data through the data processing module 110. The current stage includes, for example, production management system (PMS) data, online monitoring system data, and spatial geographic information system (GIS). Data, weather system data and video surveillance platform data. The distribution equipment status visualization platform and the horizontal data integration of these systems follow a unified interface specification, and the data interface method takes the Web Service method first, and combines the actual situation of the project.

[0042] In a specific example, such as shown in Table 1, an example of a partial data source system and its access objects is shown.

Table I

[0046] Further, the implementation manner of the interface includes the following:

[0047] Web Service service call interface: for the power distribution equipment state visualization platform, state monitoring information that needs further processing, and state detection has provided a service interface (for the power distribution device state visualization platform, online monitoring of data not provided with the service interface is required The data in the state monitoring is obtained through the service call, and the principle is that the data is stored in the distribution device state visualization platform database.

[0048] The page embedding integrated interface: for the power distribution device state visualization platform, no further processing status monitoring information is needed, and the state detection has provided the corresponding module page, and the corresponding function page is invoked through the url.

[0049] Structured data acquisition interface: for conventional relational database data, using JDBC/0DBC and other programming interfaces to directly obtain database data, for data with high security and high privacy, the interface provided by the business system is obtained/converted by data. The device calls the acquisition or is actively pushed by the business system, and sends the relevant data to the enterprise message bus, and the data acquisition/conversion device monitors the message bus to obtain the data.

[0050] Unstructured data acquisition interface: for unstructured data such as documents, audio, surveillance video, and pictures obtained by inspection, the data acquisition/conversion device directly reads and calls related files through a common file transfer protocol, and performs subsequent steps. Related cleaning, conversion and other processing work.

[0051] Grid spatial data acquisition interface: grid space data is more complex, including structured data such as coordinate axes, warp and latitude, and unstructured data such as images and texts. The data acquisition/conversion device acquires data from the system using structured data interfaces and unstructured data interfaces, respectively, according to different data types. For the data pulled from the business system by the data acquisition/conversion device calling the programming interface or the system interface, it is necessary to configure relevant policies in the device, and define related interfaces, cycles, calling frequencies, calling objects and other related parameters, data acquisition/ The conversion device automatically performs related tasks to pull data from the business system. Data acquisition is mainly divided into information intranet data acquisition and information extranet data acquisition. The data acquisition/conversion device is deployed on the information intranet, and the data acquisition of the service system in the information extranet needs to be obtained through a secure isolation device based on a secure transmission channel. The overall idea of ​​data acquisition is based on the cross-platform programming interface enterprise service bus, using data interface, data center sharing, and secure file transmission under network isolation to solve cross-platform database access, high-speed concurrent reading of cross-platform big data files, and cross-platform Key technologies such as secure transmission and synchronization of platform data.

[0052] On the other hand, the power distribution equipment state visualization platform needs to interact with a plurality of information systems, and needs to be connected by loose affinity. For example, Service-Oriented Architecture (SOA) can be used. S0A is a component model that links different functional units (called services) of an application through well-defined interfaces and contracts between these services. The interface is defined in a neutral manner and should be independent of the hardware platform, operating system, and programming language that implements the service. This enables services built into a variety of such systems to interact in a unified and versatile manner.

[0053] In an embodiment of the present invention, the data processing module 110 is configured to perform pre-processing and cleaning on the acquired multi-platform data. This is because the acquired data information has many sources, different structures and many attributes. Therefore, it is necessary to pre-process and clean the data in the data quality management. Specifically, the preprocessing is based on the service system, type, structure, and size of the data, and is marked with a uniform specification for identifying the source and type of the data. At the same time, combined with the preset data rule base, according to the marking of the data, the corresponding rules and data are encapsulated, and the encapsulated data can be identified, controlled and accompanied by corresponding cleaning rules, and can be sent to the data cleaning stage for cleaning work. For example, as shown in Figure 2.

[0054] The device state evaluation data source system generally maintains the original data characteristics, and the data conversion technology is used to improve the data quality, thereby improving the accuracy and performance of data mining or data stream mining. Data conversion mainly improves data quality and improves the accuracy and performance of data mining or data stream mining through data generalization, data normalization, and data attribute construction. Data generalization uses concept stratification, replacing low-level "raw" data with high-level concepts, and generalizing raw data in the database into aggregated, statistically significant metadata at the conceptual level of interest to the user.

[0055] The data analysis module 120 is configured to perform big data integration, storage, retrieval, and data mining analysis.

[0056] In order to facilitate understanding, first describe the current status of big data technology: with the development of smart grid, the establishment of various information platforms, the construction of intelligent substations and the gradual application of intelligent power distribution equipment, a large number of information networked from different sources Integration and sharing is an inevitable trend in the development of equipment state evaluation, and promotes the assessment and prediction of power distribution equipment to the direction of information integration and comprehensive analysis based on panoramic state. However, there are many factors affecting the operating state of power distribution equipment. Explosive growth of state monitoring data (such as PD, vibration, image, video, etc.) plus the amount of information related to grid operation and meteorological environment closely related to equipment status is huge. With rapid growth, it is difficult to establish a complete and accurate equipment state assessment mechanism model and causality model to analyze these data, which poses new challenges to improve the operation and maintenance management level of power grid equipment, while the field of power distribution equipment status assessment has multiple sources. The lag of the data mining technology is the bottleneck to control the state of the power distribution equipment and ensure the safety of the power grid. In this context, it is necessary to make full, reasonable and effective use of the data in the various information systems that have been built, and to integrate a large number of scattered distribution equipment status, operation and environmental meteorological multi-source information into and use advanced big data. The processing technology realizes the comprehensive analysis of differentiation, diversification and complexity. It finds valuable rules for equipment state evaluation from a large amount of data, timely captures the precursor information of early faults of equipment, and predicts the probability of fault occurrence, which is the fine state of equipment. The evaluation and prediction provide new solutions and technical means, and ultimately improve the accuracy of power distribution equipment evaluation, which is conducive to timely detection, rapid diagnosis and elimination of hidden troubles, improve equipment utilization, and ensure safe and reliable operation of equipment and power grid.

[0057] Based on this, in one embodiment of the present invention, the data analysis module 120 includes, for example, a sensing layer, a network layer, and an application layer. The sensing layer is used for data collection; the network layer is used for data transmission; the application layer further includes a service layer, a service layer, a presentation layer, and a tool set, and the service layer is used to provide data mining and analysis capabilities, and the service layer is used to implement The business requirements of the specific product, the presentation layer is used to provide an interactive interface, such as a mobile APP application, a web browser application, and a tool set for providing installation deployment tools, data mining tools, business modeling tools, and code generation tools.

[0058] Specifically, the three layers of the sensing layer, the network layer, and the application layer all have the functions of collecting, storing, analyzing, and transmitting, but the focus of each layer is different, and the collection for itself is used to optimize the system, and each time Each layer has the ability to access third-party systems. The storage ensures the ability of any node to fail data without loss. It analyzes the point-to-point analysis from different perspectives of the sensing layer, network layer and application layer.

[0059] In an embodiment of the invention, the interaction layer, the network layer and the application layer interact, the interaction includes a message flow and a data flow, and the processing of the data flow is controlled by the message flow.

[0060] Further, the application layer further includes, for example, a storage layer. The storage layer is used for data storage. For example, Red is responsible for real-time data storage. Event-driven data can be persisted to historical databases HBASE, Oracle/MySql, SqlLi te, HBASE for large-scale data, and Oracle/MySql for medium-sized. Scale data, SqlLite for small-scale data, you can use a unified interface to access Redis, HBASE, Oracle, MySql, SqlLite.

[0061] In a specific example, the service layer YARN is a resource scheduling manager of Had〇〇p2, on which Spark and MapReduce are provided, and MapReduce provides offline parallel computing capability of big data, and Spark Streaming provides a large Online streaming computing capabilities of data, Spark's Shark provides Sql-style interactive computing capabilities, Spark's GraphX ​​provides graph computing capabilities, Mahout and MLlib provide data mining and machine learning capabilities, and data mining in toolsets through embodiments of the present invention The tool (similar to PRiSM) is modeled, and then the model is injected into the IoT platform, and the data engine analysis can be performed in real time through the model engine driver.

[0062] The presentation layer uniformly implements the mobile and web interaction interface through WebSocket, Node.js, JQuery, and HTML5 on the basis of the system platform.

[0063] The interaction between the storage layer, the service layer, and the presentation layer is event-driven, which ensures that the collected data can be instantly displayed to the user interface.

[0064] In a specific example, the description of the mining algorithm and the coupling analysis method adopted by the data analysis module is as follows: The heterogeneous big data storage of the system adopts the distributed file storage of HAD00P2.0, the distributed NoSQL column database, and the scalable data. Technologies such as warehouses provide support for big data applications and can meet the relevant technical requirements for big data management platforms. Under the Hadoop framework, the big data management platform software implements a distributed, column-oriented, multi-dimensional data storage system. Mainly composed of the following parts: distributed collaborative work system, distributed file system, distributed database, distributed data warehouse, unstructured data preprocessing and multi-level comprehensive index. Distributed File System: The system uses a distributed file system based on HDFS and HBASE. For massive unstructured small files, as well as complex and variable structured data, use HBASE's key-value storage. For larger single files, they can be stored directly in the HDFS file system. Distributed Database: A distributed NoSQL database with multiple reads, multiple writes, and multiple reads is built into the system. Through the master-slave replication technology based on the acceleration component, the data consistency between multiple databases is ensured, and the disaster recovery function is realized, and the pressure when reading data is shared.

[0065] Comprehensive analysis systems require complex data queries, but HADOOP's key-value-based simple indexing techniques are difficult to support multidimensional data queries. Therefore, it is necessary to establish a multi-level comprehensive index to improve the evaluation data retrieval performance of similar features. The specific primary index uses a multi-dimensional R-tree structure to achieve key data feature dimensions. The relative density of the clustered objects in the same cluster is close to each other, and the relative density of the objects in different clusters is low, so as to achieve the purpose of clustering objects. This can overcome the locality of a simple global distance criterion for most clustering or neighboring algorithms as a basis for detection: it is related to the distribution of neighbors within a certain range. By overcoming certain limitations caused by the global distance threshold, an adjacent scalable algorithm will form an adjustable and scalable effective clustering method, which can better support the super-grid data characteristics of power distribution equipment such as transmission lines. .

[0066] The evaluation module 130 is configured to generate a big data-based power distribution equipment evaluation model, and evaluate the power distribution equipment according to the power distribution equipment evaluation model, and generate a corresponding processing strategy according to the evaluation result.

[0067] In an embodiment of the present invention, the power distribution equipment evaluation model includes at least: a transformer-type equipment fault prediction model, a development trend of a switch and a combined electrical equipment state, and a fault probability dynamic prediction model, and a transmission based on a complex correlation relationship. Line fault prediction model. Considering the difference between different equipment structures and fault types, the transformer, GIS/circuit breaker and transmission line fault prediction methods are elaborated.

[0068] Transformer-based equipment failure prediction model: using traditional online monitoring, operation, test maintenance records, historical conditions, defect records combined with laboratory-related aging, defect pattern recognition, and new family differences in big data information systems And related analog test, etc. to construct the characteristic parameter data platform for fault diagnosis and dynamic prediction of transformer equipment; statistically analyze the typical defects of transformer equipment, and use the method of fitting and analogy in big data information to study the missing of some characteristic data missing Manual completion of data. The techniques of deep learning (classification) and cluster analysis are used to analyze a large number of sample data. The data association algorithm is used to mine the variation law of the typical defects of the power distribution equipment and the associated state information of the fault mode and the weight combination, and analyze the type, location, and The relationship between severity and related state, combined with poor working conditions, grid operation status, and family defects on equipment state changes, builds a multi-dimensional equipment fault diagnosis prediction model based on big data samples, as shown in Figure 3.

[0069] The development trend of the state of the switch and the combined electrical equipment and the dynamic prediction model of the failure probability: firstly statistically analyze the major defect or fault history data, and the laboratory simulation defect data. Using correlation rule mining, multivariate correspondence analysis, principal component analysis and other related relationship recognition technologies to distinguish fault types and find effective data combinations that play a leading role. For GIS and circuit breakers, the known direct impact on the evaluation results is effective. The data includes: closing resistance, such as SF6 humidity, SF6 gas pressure, partial discharge, vibration, etc., combined with the equipment to carry out relevant tests, and collect new data, establish a valid data multi-logic model and correlation matrix of GIS typical failure mode. The association rules mining of big data information that integrates grid information, equipment status information and natural environment information, mining the variation law of GIS typical defects and failure mode effective data, and analyzing the relationship between defect type, location, severity and effective data Using time series model, grey model, support vector machine, regression model and other methods to calculate the relationship between data that is closely related to effective data (such as poor working conditions, grid operating state, family defects, device state changes, etc.) and effective data. index. It is used to dynamically adjust the weight of effective data (closing resistor, partial discharge, etc.), and build a multi-dimensional device dynamic fault diagnosis model based on big data samples. Some time series are time-dependent cluster time variables. Although the single sequence values ​​constituting the time series have uncertainties, the changes of the whole sequence have certain regularity, which can be approximated by the corresponding mathematical model. Combining the multi-logic model and state evolution history data of effective data of the equipment, based on the techniques of multivariate time series, the correlation evolution law and state distribution change of GIS switch equipment fault feature information are studied. Combined with the status confirmation and diagnostic analysis results, the ARMA (Auto- Regressive and Moving Average Model, regression model of the model, GIS switch device state development trend and fault probability dynamic prediction method.

[0070] The fault prediction model of transmission line based on complex correlation relationship: realizes the segmentation mapping of state data according to the time and position information of the fault of the transmission line; further calculates the support degree of all attributes after division, and analyzes the cause of the failure of the transmission line Correlation with other state parameter changes, such as the relationship between transmission line and ice coating, wind bias, thunderstorm, pollution flash, etc., analyze the objective law of fault development, get frequent itemsets, and extract faults from frequent itemsets. The association rules of development; using state evolution historical data, combined with equipment failure mode multi-dimensional association rule analysis, based on multivariate time series method to study the correlation evolution law and state distribution change of transmission line fault feature information, and finally establish a dynamic prediction model of transmission line failure probability. For example, as shown in Figure 4.

[0071] As a specific example, the following specific technical routes are respectively described by taking transformer type equipment, switches, GIS type equipment, and distribution line (cable) equipment as examples. According to the cascading relationship and connotation mechanism of equipment status information, the relationship between the global total data of the equipment and the equipment status is analyzed, the status evaluation index system of the equipment is determined, and the relevant characteristic parameters are reflected to reflect the equipment status and the equipment status comprehensive evaluation model. Using the online self-learning method of big data samples to analyze the individualization rules of equipment state changes from different manufacturers, different equipment types, different voltage levels, different operating years, different operating environments, different operating seasons, etc., based on which the evaluation model parameters are given. The individualized adjustment method for judging the threshold value, the individual attribute information of the statistical analysis device is obtained by the state evaluation correction index, and the differentiated evaluation model for the device state evaluation is established. The following is a description of the specific technical route of transformer equipment, switches and GIS equipment, and transmission line (cable) equipment. Through the combination of DS evidence reasoning theory and rules, the basic probability assignment function, reliability function and likelihood function based on the combination of multiple evidence bodies are obtained. As shown in Fig. 5, a multi-factor state evaluation analysis algorithm is formed.

[0072] In other words, in an embodiment of the present invention, the evaluation module 130 is configured to adopt a multi-factor state evaluation analysis algorithm, for example, as shown in FIG. 5, which specifically includes:

[0073] 1) analyzing the decision problem, constructing a set of system propositions, such as a GIS ontology evaluation module, that is, a recognition framework of the system Ω = {A1, A2, ..., Ak};

[0074] 2) for the target information system, constructing the evidence body Ei (i = I, 2, ..., m) based on the recognition framework, specific detection means, such as partial discharge, SF6 humidity, etc.;

[0075] 3) determining the basic credibility distribution of each evidence body mi (Aj), j = l according to the collected data of each evidence body, the global total data, and the characteristics of each proposition set in the recognition framework. 2, ..., K, that is, the ability of different state information to respond to the state of the device;

[0076] 4) assigning mi (Aj) according to the basic credibility, respectively calculating the reliability interval [Beli, Pli] of each proposition in the recognition framework under the action of the single evidence body;

[0077] 5) using the DS synthesis rule to calculate the basic credibility distribution m (Aj) and the reliability interval [Bel, Pl] under the joint action of all evidence bodies;

[0078] 6) constructing corresponding decision rules according to specific problems;

[0079] 7) Deriving a decision conclusion according to the decision rule.

[0080] Further, in an embodiment of the present invention, combined with FIG. 6, the process of evaluating the differentiated state of the transformer-type device is described as follows: First, the relevant parameters of the induction transformer state evaluation are analyzed, and the relevant parameter data types are determined, and the pair is determined. Data statistical analysis, classification and identification methods, proposed feature extraction methods for unstructured data such as images, videos, texts, etc., and then using principal component analysis, correlation analysis and other big data core mining analysis methods to determine the state of the transformer. The characteristic parameters and their coupling relationship with the device state complete the existing feature parameter set. Finally, for the specified parameters or parameter sets, multi-dimensional statistical analysis, multi-dimensional correlation analysis and other methods are used to determine the evaluation criteria and evaluation model of transformer state evaluation, and establish a complete evaluation index system; through regular or irregular data analysis, Dynamic maintenance of the indicator system; statistical analysis of equipment state, special working conditions, different structures, etc., the error of transformer state judgment, determine the correction index under the corresponding conditions, use artificial neural network, fuzzy clustering and other methods to establish the individualized state of the transformer Evaluate the model to achieve a differentiated assessment of the state of the transformer equipment.

[0081] The GIS/circuit breaker device differentiation state evaluation process is described as follows: the relevant parameters of the power distribution equipment operating conditions are extracted from the big data comprehensive analysis platform, and the meteorological environment information, the operating condition information, the online monitoring information, and the pre-test are integrated. The information of the inspection information, manual inspection and individualized data of the equipment is analyzed by systematic hierarchical clustering method to analyze the dependence between the parameters and the parameters and the state of the GIS/circuit breaker, and establish the partial discharge and breaking of the GIS/circuit breaker. Evaluation model for key performance such as short-circuit current. Using sequence rule mining method to analyze the individual differences of equipment (operating period, manufacturer model, operating conditions) and the impact of GIS / circuit breaker key performance degradation, expand the influencing factors of the evaluation model, form a GIS / circuit breaker based on data state dependence Personalized and differentiated evaluation methods for key performance.

[0082] The process of evaluating the differentiated state of the distribution line is described as follows: extracting relevant parameters of the transmission line and cable operating conditions from the big data platform, including meteorological environment, operating conditions, online monitoring, manual inspection, pre-testing The data is systematically clustered to analyze the dependence between the above parameters and the parameters and the state of the distribution line, and establish key performances such as ice distribution, sag, insulator contamination and lightning protection level of distribution lines (cables). Evaluation model. The sequence rule mining method is used to analyze the influence of individual differences (operating years, manufacturer models, operating conditions) and the performance degradation of distribution lines (cables), and the influencing factors of the evaluation model are extended to form a distribution based on data state dependencies. Personalized and differentiated evaluation methods for key performance of the line.

[0083] A quick evaluation method of device status will be described below with reference to FIG. 7. Specifically, under the support of big data hardware platform, using the methods of predictive model, orphan analysis, clustering and partitioning, the method of rapid detection and early warning of abnormal state based on real-time data stream mining technology of state information is proposed to realize the rapid state of abnormal state. Detection and early warning to improve the timeliness of assessment. State Information A data stream is an ordered sequence of time consisting of a large number of continuously arriving, potentially infinitely long, constantly changing, multi-source state information data. With the improvement and improvement of the distribution equipment condition monitoring system and production management system and the real-time integration of grid information and environmental meteorological information, the data of the status information of the distribution equipment presents the characteristics of large and continuous data flow, rapid excavation and inspection. Outliers in the data stream provide early warning, status evaluation, and decision support for power distribution equipment. On the one hand, the rapid excavation and early warning research on the abnormal value of real-time data flow of the state information requires real-time mining of a large number of structured data flow abnormal states of the device, on the other hand, it needs image, video, vibration (waveform, fingerprint), partial discharge (waveform, Quick extraction of unstructured data eigenvalues ​​such as maps and test reports.

[0084] Specifically, in conjunction with FIG. 7, the device state information data stream is first summarized, that is, a fixed length window is set, and the window contains all the data in the limited acquisition period. There are three kinds of outlier detection methods for the summarized data, which are the threshold, trend analysis method and time series transfer function model in the comparative state evaluation guide. The three methods can detect three types of outliers, an outlier that exceeds the threshold of the state quantity, an outlier that is generated by external disturbances, and an abnormal value that changes the trend of the potential fault, including the following steps:

[0085] A) according to the relevant requirements in the power distribution equipment status evaluation guide, each state quantity threshold in the corresponding guide scans the data one by one, and when any one of the data exceeds the threshold defined in the guide, the data is marked as abnormal. Value, separated from the original data.

[0086] B) transforming the data into a multivariate time series, calculating a cross-covariance function and a cross-correlation function of each one-dimensional time series, thereby obtaining the order and delay parameters of the transfer function numerator and the denominator polynomial, and then fitting the transfer function The model finally determines the time of the disturbance and the abnormal data generated based on the ACF test of the model residual sequence. Specifically, the power distribution equipment may be affected by external disturbances at a certain time during operation, which affects the distribution of state quantity data (for example, when the transformer is subjected to a short-circuit shock, the oil temperature may rise for a short time), in this case The data will have a certain initial migration effect when the time T interference occurs, and then a permanent horizontal migration or temporary horizontal migration will occur due to the interference cause and the state quantity attribute difference. Such anomalous values ​​can be quickly detected by a time series transfer function model, that is, the data is first transformed into a multivariate time series, and the cross-covariance function and the cross-correlation function of each one-dimensional time series are calculated, thereby obtaining a transfer function molecule, The order of the denominator polynomial and the delay parameters are then fitted to the transfer function model. Finally, the interference time and the generated abnormal data are determined according to the ACF test of the model residual sequence.

[0087] C) Based on incremental recursive least squares regression parameter estimation and generalized likelihood ratio change point detection, an incremental mechanism is used to determine data sequence regression model parameters and segmentation points, and data trend characteristics are extracted in real time, and the trend is changed. Data is marked as anomalous data. This is due to the possibility of insulation aging, mechanical defects, etc. during the operation of the power distribution equipment. The state quantity data may have a trend change (such as the deterioration of the oil-paper insulation degradation of the transformer will lead to oil dielectric loss, oil CO and CO 2 gas rise). The trend is strengthened), so the separation of outliers of such trend changes is of great significance for detecting potential failures of power distribution equipment. In an embodiment of the present invention, the detection method of such outliers is based on incremental recursive least squares regression parameter estimation and generalized likelihood ratio change point detection, and the algorithm uses an incremental mechanism to determine data sequence regression model parameters and Split points, extract data trend characteristics in real time, and mark trend-changing data as abnormal data.

[0088] Further, in order to establish a more accurate power plant outage probability model, based on the historical accident record of the power enterprise and the real-time state monitoring information of the equipment, the internal relationship between the internal state of the device, the evolution of the external environment and the forced disconnection of the device is revealed. Establish a "temporary learning-event-driven" type of spatiotemporal state model reinforcement learning system. Uncertainty theory, such as credibility theory, cloud model, etc., is used to give a switchable time-varying equipment outage model in the absence of collected data. Establish a parameter learning library, so that the equipment outage model has adaptive feedback correction and security check function. The forced outage rate of equipment is mainly affected by both time and space factors. The time factor is mainly reflected in the aging of equipment. The spatial factors are mainly reflected in the different locations of the equipment in the power system and the different meteorological environments around it. Modeling the transformer failure rate based on spatiotemporal state analysis, the model should have strong generalization ability and versatility. The power distribution equipment failure rate model considering big data can not only characterize the general aging of the equipment, but also consider the impact of big data and multiple factors on the equipment. It can quantify the impact of internal covariates and external covariates on the failure rate, such as certain equipment. Detection information, external environment in which the equipment is operating, weather conditions, and system conditions. The model also takes into account the stochastic process of state transitions. Considering the stochastic process of a particular device will make the model more precise and special, and closer to reality. In addition to the general model, different devices have many specific models under different conditions. The main cause of aging failure of transmission lines is the loss of tensile strength of the wire, which is a gradual accumulation and irreversible process. Theoretical analysis and experimental results show that the annealing of high temperature conductor is the main reason for the loss of tensile strength of the wire. The overhead transmission line temperature is mainly determined by the conductor current, ambient temperature, wind speed, wind direction, and solar heat.

[0089] In one embodiment of the present invention, the failure rate of the distribution line under different weather conditions is the number of times the time is converted into a unit when the failure occurs, and the average of the failure rate is in units of one calendar year.

It can be expressed as:

[0090]

Where N is the expected duration of normal weather, S is the expected duration of bad weather, λ is the expected value of the component failure rate in normal weather, and λ' is the expected value of the component failure rate in bad weather.

[0092] Most of the transformers used in the power grid are oil-immersed transformers. The main cause of transformer aging failure is the loss of mechanical strength of the insulating paper, which is a gradual accumulation and irreversible process. The insulation failure of the transformer is related to the temperature at which it operates. It is generally believed that the transformer hot spot temperature is the highest temperature experienced by the transformer insulation system, and the hot spot is near the top of the transformer high voltage or low voltage winding. The transformer aging process is often described by the Weibull distribution, and its long-term failure Arrhenius-Weibull model. Therefore, the failure rate and cumulative probability distribution function of the transformer can be expressed as:

[0093]

[0094]

[0095] It should be noted that the model assumes that the parameters β and C remain unchanged at different temperatures. In the case of having enough samples, these parameters can be estimated by least squares or maximum likelihood. A two-state weather model is used to describe the accidental failure mode failure rate of the transformer, and a two-state weather model is used to describe the accidental failure mode failure rate of the transformer. The expression is:

[0096]

[0097] wherein

For the statistical average of accidental failure of the transformer, N is the duration of normal weather, S is the duration of bad weather, F is the proportion of faults occurring in bad weather, w is the current weather condition of the transformer, normal weather W = 0, bad weather w=l.

[0098] Further, the evaluation module 130 is further configured to perform device importance evaluation according to device status and system risk. The reliability of the operation of power distribution equipment is directly related to the safety and stability of the power system. With the continuous expansion of power scale and the introduction of market economy, power companies have put forward higher requirements for the safety and economic operation of power distribution equipment. The improvement of maintenance plan and the formulation of maintenance strategy directly determine the cost of power equipment use phase. And the service life; therefore, the risk analysis of power equipment failure and the importance assessment of system risk not only help to develop an appropriate maintenance plan, improve the reliability of the system operation, but also better avoid the maintenance shortage of the traditional preventive maintenance program. Problems such as over-maintenance, reducing maintenance costs and operating costs, and effectively improving the reliability and economy of power equipment operation.

[0099] In an embodiment of the present invention, the device importance evaluation is performed according to the device status and the system risk, and specifically includes the following steps:

[0100] a) calculating the real-time failure probability of the system component considering the big data by using the PHM model according to the big data state evaluation result, the operation information, and the micro-meteorological data;

[0101] b) using an enumeration method to select a system state, enumerating to a third-order fault, forming an expected fault event, and calculating a probability of occurrence of the fault event;

[0102] c) performing static security analysis on the selected system state, using the optimal power flow to calculate whether the system state satisfies the abundance, and if the load is to be cut, the system state is an emergency state, and proceeds to step d), if no load is required The status of the system is an alert state or a healthy state, and the system performs NI verification. If the security criterion is met, it is in a healthy state, and returns to step b), otherwise it is in an alert state, and proceeds to step d);

[0103] d) calculating an emergency index or a number of police rings in the state of the system, and using a risk tracking model to calculate a contribution value of each faulty component in the state;

[0104] e) returning to step b) until traversing all fault events of the expected fault set;

[0105] f) calculating the total emergency index and the total number of police rings, and calculating the component emergency importance index and the alert importance index, and sorting according to the importance index to determine the weak device of the system.

[0106] In summary, according to the big data-based power distribution equipment state visualization platform according to the embodiment of the present invention, a relatively mature development technology of software engineering is used in the development process of the system, and a demand analysis of the system function is performed to construct a network application system. At the same time, a safe operation mechanism is established to ensure the continuous and reliable operation of the system. According to relevant management regulations, a reasonable work flow is formulated, mainly involving power supply analysis, status maintenance, distribution network equipment life cycle management, and power information collection service module. . The platform uses information technology to establish a stable and efficient operation and maintenance and data verification system, promotes the practical and deep application of the system, guarantees the sustainable development of each system, and realizes the power supply company's power supply analysis and information warning for the distribution equipment within its jurisdiction. Claim. The big data-based power distribution equipment status visualization platform is mainly improved from the following aspects:

[0107] (1) More powerful device status monitoring function.

[0108] Fully adapt to the data structure requirements of the future PMS2.0 of the State Grid Corporation, and centrally display the status information of various types of equipment. Apply visualization technology to convert numbers and text into graphics, and display individual differences and trends in device status using friendly methods such as dynamic curves, charts, and lists. Develop the "Devices I Care" module to facilitate the monitoring of personnel at different levels and units to dynamically monitor the status of the equipment as needed to achieve visual monitoring.

[0109] (2) Based on big data analysis, the panoramic display device status function.

[0110] The platform will access PMS2.0 system data, distribution automation system data, and mine core state parameters closely related to equipment load capacity, and establish equipment load capacity under different service environments, different equipment aging states, and different equipment defects. Short-term, medium-term and long-term multi-scale dynamic assessment and prediction models, study the relationship between overload operation of distribution equipment and equipment health status and life, and propose a dynamic calibration method for equipment real-time dynamic capacity expansion control strategy and equipment load safety margin. . Multi-dimensional display of device status information to improve the fine management of distribution network equipment.

[0111] (3) More abundant device fault diagnosis function.

[0112] Highly integrate information from different application systems, equipped with advanced diagnostic modules such as status detection, maintenance work, auxiliary work, power quality analysis, risk warning, etc., deep mining and multi-angle analysis of massive data, combined with power distribution equipment status Monitoring information and equipment inspection test information in PMS, online diagnosis of power distribution equipment status, and use of typical faults, defect aids library and equipment standard library to improve the real-time and accuracy of state diagnosis, and support managers to make quick and accurate decisions.

[0113] (4) A more comprehensive operational risk warning function.

[0114] Performing continuous real-time scanning on weak points of the power grid according to the established rules, automatically searching for power distribution equipments with serious faults, state degradation, and the like, and displaying them through risk information aggregation, inspection defects, patrol hidden dangers, monitoring abnormalities, and the like, and Automatically issue warning information.

[0115] On the other hand, the platform adopts a power distribution equipment evaluation method based on power big data analysis, extracts relevant parameters of the operating conditions of the substation equipment from the big data comprehensive analysis platform, and analyzes individual differences and devices by using the sequence rule mining method. The impact of key performance degradation forms a critical performance assessment method for equipment based on data state dependencies.

On the other hand, the platform adopts the minimum covariance determinant M(3) robust multivariate detection method for oil chromatogram H2, C0 and total hydrocarbon class 3 characteristic gas outliers. Using the iterative and Mahalanobis distance idea to construct a robust covariance estimator to detect outliers, strengthen the statistical law of oil chromatographic data; through the tracking and evaluation of the outlier interval, it can more clearly reflect the change of transformer operating state.

[0117] Further, the platform adopts a transformer partial discharge pattern recognition method based on the improved dual-tree complex wavelet transform and BP neural network, and designs a transformer partial discharge detection system based on the ultra-high frequency method, and applies the double-tree complex wavelet transform. The collected partial discharge envelope signal is denoised, and the BP neural network is used to identify the type of transformer discharge, which effectively eliminates the spatial electromagnetic wave and hardware circuit noise interference in the original signal, and improves the correct rate of partial discharge type identification.

[0118] The big data-based power distribution equipment state visualization platform according to the embodiment of the present invention utilizes multi-dimensional visual display functions to maximize reuse of various distribution equipment basic data and maintenance management resources, and utilize data integration. And data mining technology to achieve comprehensive analysis of equipment, for the transmission, substation equipment to carry out state evaluation, fault diagnosis, risk assessment, life prediction and maintenance decision-making work, assist with technical supervision and intelligent report management, provide a panoramic, real-time for the operation and maintenance department Multi-dimensional, intelligent equipment management and control information platform, so that a large number of new, old and different, different state of the equipment is always under close monitoring, prevent equipment operation risks in advance, improve the response speed to emergencies, and meet intensive development Lean management business needs, improve the multi-source information exchange and fusion capabilities of equipment, realize panoramic real-time perception of power distribution equipment, multi-dimensional intelligent monitoring and control, and make the life cycle management of equipment transparent and efficient.

[0119] In the description of the present specification, the description with reference to the terms “one embodiment”, “some embodiments”, “example”, “specific example”, or “some examples”, etc. Specific features, structures, materials, or characteristics are included in at least one embodiment or example of the invention. In the present specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples.

[0120] While the embodiments of the present invention have been shown and described, it will be understood by those skilled in the art The scope of the invention is defined by the claims and their equivalents.