Detection of loss or malfunctions in electrical distribution networks

The present invention relates to the field of power measurement.

Detection of non-technical losses of electrical energy is a problem for all electrical utilities. If electrical energy is lost due to fraudulent activity, for example by using a connection to a network that bypasses each electric meter, the utility company will take the location of the theft and the time / duration and value of the theft. Is very difficult to analyze.

Although gaining recognition, there are markets where more than 10% of the energy supplied is consumed by theft or without permission, which poses a major threat to income.

It was not possible to accurately detect the location, time, and amount of electrical energy theft, even if the exact values measured by smart meters, for example, were available.

Moreover, it is difficult to provide legal evidence of electrical energy theft after it has been detected.

In addition, the loss may indicate anomalies in different grids, such as chip level, equipment level, isolated networks, such as in-vehicle, plant or household consumer networks, or any other suitable distribution network. is there.

There are solutions based on data mining and neural networks. However, none of the known methods are accurate enough to provide an accurate indication of the location, time, and amount of electrical energy theft. The majority of these methods are based on historical data and load profiles or comparisons with others. These methods generally can only provide an indication of fraudulent activity, not an indication of the time and amount of theft.

In addition, there are solutions within smart meters intended to detect direct operation of the meter itself. However, one such smart meter cannot detect smart meter detours.

In addition, measurements by a group of meters placed within a point of transmission, ie a feed, can be compared to the combined readings from the supplied consumer's meters. However, the exact location and amount of electrical energy theft cannot be determined because the load in the grid and parts of it changes.

Even when combining different solutions, the resulting solution cannot provide an accurate indication of the location, time, and amount of theft.

<p> Therefore, an object of an embodiment of the present invention is to provide a method that makes it possible to provide accurate detection of losses in a power grid.</p>

<p> This problem is solved by a method of detecting losses or anomalies in the grid. The first step receives the average measured voltage and average measured current values associated with one or more supply points in the grid. Another step is to receive the average measured voltage and average measured current values associated with multiple points of consumption in the grid. The conductivity parameter is calculated based on the received average measured voltage value and average measured current value. Then, in a further step, further average measured voltage values and average measured current values are received from the point of supply in the grid. In another step, additional average measured voltage and average measured current values related to the point of consumption in the grid are received. Then, the further received average using the calculated conductivity parameter to detect any deviation of the measured current value from the calculated current value for a given point of consumption or point of supply. Analyze the measured voltage and average measured current.</p><p> In an embodiment of the invention, the received average measured voltage value and average measured current value are averaged over a predetermined time period.</p><p> In a further embodiment of the invention, it is determined from point analysis that the location of the loss is determined.</p><p> In a further embodiment of the invention, the amount of loss is determined from point analysis.</p><p> According to another embodiment of the invention, the calculation of conductive parameters is based on optimization methods.</p><p> In a further embodiment, the optimization method is selected from the group including quasi-Newton, conjugate gradient, simplex method, fuzzy logic, evolution method.</p><p> In a further embodiment of the present invention, a second conductive parameter is calculated based on the received further average measured voltage value and further average measured current value, and the first conductive parameter is referred to as the second conductive parameter. Analyzed by comparison.</p><p> According to another embodiment, the results of the analysis can be visualized.</p><p> This problem is also solved by systems configured to detect losses or anomalies within the grid. The system consists of a first logical receiver configured to receive average measured voltage and average measured current values associated with one or more supply points in the grid, and multiple consumptions in the grid. Includes a second logical receiver that receives the average measured voltage and average measured current values associated with the point. Further, the system includes a calculation unit configured to calculate conductivity parameters based on the mean values associated with the received mean measured voltage and current values. The first logical receiver is further configured to receive additional average measured voltage and average measured current values associated with the point of supply in the grid, and the second logical receiver is in the grid. It is further configured to receive additional measured average voltage and average current values related to the point of consumption. The calculation unit uses the calculated conductivity parameter to detect any deviation of the measured current value from the calculated current value for a given point of consumption or supply. It is further configured to analyze the average measured voltage and average measured current.</p><p> According to another embodiment of the invention, the calculator is further configured to average the received average measured voltage and average measured current values over a predetermined time period.</p><p> In a further embodiment of the invention, the calculation unit is further configured to determine the location of the loss from point analysis.</p><p> In another embodiment, the calculation unit is further configured to determine the amount of loss from point analysis.</p><p> According to a further embodiment of the present invention, the calculation unit is further configured to calculate conductivity parameters based on optimization methods.</p><p> In a further embodiment of the invention, the optimization method is selected from the group including quasi-Newton, conjugate gradient, simplex, fuzzy logic, evolutionary methods.</p><p> In a further embodiment of the invention, the computational unit is further configured to visualize the results of the analysis.</p><p> In a further embodiment of the invention, the calculation unit will calculate the second conductivity parameter based on the received further average measured voltage value and further average measured current value for calculating the second conductivity parameter. Further configured, the calculation unit is further configured to analyze by comparing the first conductive parameter with the second conductive parameter.</p><p> Some embodiments of the apparatus and / or method according to embodiments of the present invention will be described below with reference to the accompanying drawings only as examples.</p>

<figref num="1">It is a figure which shows typically the exemplary power grid.</figref><figref num="2">It is a figure which shows schematic the equivalent circuit of the example distribution network of FIG.</figref><figref num="3">It is a figure which shows schematicly the equivalent circuit of the example distribution network of FIG. 1 used in detection.</figref><figref num="4">It is a figure which shows the table of the numerical result derived by the method by the embodiment of this invention.</figref><figref num="5">It is a figure which shows the graph representation of the numerical result derived by the method by the embodiment of this invention.</figref><figref num="6">It is an exemplary flow chart according to a different embodiment of the present invention.</figref><figref num="7">It is a figure which shows typically the exemplary system by this invention.</figref><figref num="8">It is a figure which shows the table of the numerical result derived by the method by the embodiment of this invention.</figref><figref num="9">It is a figure which shows the graph representation of the numerical result derived by the method by the embodiment of this invention.</figref>

The next embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows an exemplary power grid 1 of an embodiment.

Within the network 1, the power supplier SP is shown on the right side of the figure. The power supplier SP has a potential U facing the network.₁With current I₁To deliver. Further, the network 1 includes a plurality of consumers C.₁, C₂... and C_nIncluding, where n represents a predetermined integer.

Each consumer consumes a current that is a voltage (or a potential difference). The current can be understood as the measured current at the measured voltage. For example, consumer C₁Is the measured current I₂And the measured voltage U₂Consume power that can be evaluated by consumer C₂Is the measured current I₃And the measured voltage U₃Consume power that can be evaluated by consumer C_nIs the measured current I_nAnd the measured voltage U_nConsume power that can be evaluated by.

The net 1 can be meshed within any given arrangement, as shown in the figure, i.e. the net 1 may even include connections of different consumers through loops or different paths. Can be done. That is, the consumer C1 is interconnected to the consumer C2 or Cn via the first path P1 (dotted line) and the second path P2 (broken line).

Meshed network topologies are complex and do not allow the adoption of known solutions for hierarchical networks.

The grid may not be fully aware of its topology and conductivity for this procedure.

All loads in the network, whether consumed or supplied, lead to negative or positive voltage drops that have an effect on the voltage across the network.

The network of FIG. 1 and associated loads can be described as a conductance network, as shown in FIG.

In Figure 2, multiple consumers C₁, C₂, And C_nAnd the power supplier SP are connected to each other by a conductance.

The consumer represents the point of consumption, and the power supplier represents the point of supply.

The conductance network is the dashed cloud G_cRepresented by.

Within the network, each supplier and each consumer measures voltage and current. It should be understood that these measured voltages and currents can also be their own mean, that is, the mean over a given period of time.

For convenience, without loss of generality, it is now assumed that the measured potentials and currents are measured at the same time. If the measured potentials and currents are for their own mean, they are for the same time period, eg, a length of time, eg 1 minute, 5 minutes, 15 minutes, 30 minutes, 60 minutes. Suppose it is also the mean for, ...

Here we state that there is a voltage drop across each element, assuming there is conductance between each element in grid 1 (supplier (s) and consumer (s)). be able to. The voltage drop can be calculated as follows, based on the measured voltage across each element of the grid: δU₁₁= U₁-U₁... δU_1n= U₁-U_n Based on the derived voltage drop, the simulated current through each element can be determined as follows: Is₁= g₁₁* δU₁₁+ g₁₂* δU₁₂+ ... g_1n* δU_1n... Is_n= g_n1* δU_n2+ g_n2* δU_n2+ ... g_nn* δU_nnHere, Is represents a simulated current through a network element.

In this example, Is₁Represents the simulated current from the power supplier SP, Is₂Is Consumer C₁Represents a simulated current to Is₃Is Consumer C₂Represents a simulated current to Is_nIs Consumer C_nRepresents a simulated current to.

These simulated currents are based on a network of conductances represented as a matrix as follows:

<maths num="1"><img file="JP2013545425A_D0001.tif" /></maths> Conductive matrix G_cIs determined based on the following understanding: Err (t) = (I₁-Is₁)²+ ... + (I_n-Is_n)²Err (t₁... t_k) = Err (t₁) + Err (t₂) + ... Err (t_k) Where t₁... t_kIndicates the time period.

It can be shown that the number of criteria given by the above two equations needs to be large enough to calculate the number of unknown matrix elements.

The number of unknown matrix elements depends on the size of the net, which affects the coupling of the matrix elements. That is, it is expected that the larger the net, the less the influence from the elements in the net elements that are separated from each other. Using this finding, it can be assumed that a certain number of matrix elements are kept at 0, i.e. it is not necessary to determine these matrix elements.

Assuming a smaller net and therefore a stronger bond, the number p of unknown matrix elements can be determined as follows: p = 0.5 * n * (n + 1) Here, n is the number of judgment criteria.

Assuming a larger net and therefore a looser coupling, the number p of unknown matrix elements can be defined as: p &lt;0.5 * n * 50 Here, the calculation is based on the estimated band matrix. As can be seen from above, if the network involved contains more than 50 nodes, we assume that loose coupling can be assumed. This estimate is a good assumption and reduces the overall required calculations. Still, if calculation time is not an issue, it is not necessary to assume loose coupling.

Equations Err (t) and Err (t)₁... t_kIt can be shown that the number n of the judgment criteria given by) is equal to or greater than the number p of unknown matrix elements.

Time period t₁... t_kTherefore, the number of judgment criteria is n. By considering the above inequality p &lt;0.5 * n * 50 for loose coupling, G_cIt should take a 25 hour period to calculate.

Unknown matrix element g_mnIs calculated using any suitable optimization method such as quasi-Newton, conjugate gradient, simplex method, fuzzy logic, evolution method.

The optimization method can be selected so that additional parameters take into account time needs, and / or accuracy, and / or efficiency.

The intended optimization target is the matrix element g until it reaches the intended given target._mnBy changing Err (t₁... t_k) Should be minimized as close to 0 as possible.

After determining the matrix, loss detection can be performed.

Time and position analysis is based on the following equation: Err₁(t₁) = (I₁-Is₁)²... Err_n(t_n) = (I_n-Is_n)² The analysis of the amount of stolen electrical energy is based on the following equation: If_n(t₁) = (I₁-Is₁) ... If_n(t_n) = (I_n-Is_n) Supplier SP and 3 Consumers C shown in Figure 1.₁, C₂, And C_nMove to the network containing the conductance matrix G_cAssuming that the elements in are already determined, the values shown in the table shown in FIG. 4 are received and further processed.

First time period T₁Within, average current and average voltage from suppliers and consumers are received. From these received values, the difference voltage drop δU₁₁... δU_1nAnd simulated current Is for each supplier and consumer₁... Is_nIs determined.

Theoretically, the simulated current and the measured current should be the same. However, due to some numerical calculations and different voltages and different loads, a certain amount of deviation can be assumed to be normal. In the current case, it is assumed that deviations within 5% are within normal. This value can be subject to further fine-tuning depending on numerical accuracy, averaging time period, network size, and so on. The same thing is the additional time period T₂... T₇Is repeated about.

Here time period T₃Move to. Here, consumer C₂However, it can be seen that the deviation exceeds 5%. The measured value is 5,9831162 simulated current Is₂However, the measured current shows only 2,017319.

Therefore, the above method is not only to detect the location of the theft, that is, customer 2, but also the time period T.₃It also makes it possible to quantify the theft of almost 4 units of.

The results shown in this table can also be visualized in Figure 5. Time period T₃In addition, it can be easily detected that customer 2 has consumed more power than measured.

Therefore, distribution company employees can easily detect fraud, have service employees inspect their equipment, and / or initiate legal action. Nevertheless, this can be automated, and after fraud is detected, service employees can be inspected in an automated manner, while at the same time starting legal action with stylized letters.

Therefore, the method according to the invention can be summarized as follows: Average measured voltage value associated with one or more supply points (such as supply location) SP in grid 1 U₁And average measured current value I₁Is received in step 100.

In further step 200, multiple points of consumption within grid 1 (such as all consumer locations) C₁, C₂, C_nAverage measured voltage value associated with U₂, U₃, U_nAnd average measured current value I₂, I₃, I_nTo receive.

Steps 100 and 200 are repeated as long as necessary to provide sufficient criteria for calculating the matrix elements. Therefore, in step 250, it is determined whether there are sufficient criteria available. If the number is not sufficient, repeat steps 100 and 200 (No branch), and if the number is sufficient, the method continues to step 300 (Yes branch).

The received average measured voltage value U₁, U₂, U₃, U_nAnd average measured current value I₁, I₂, I₃, I_nBased on the matrix G_cThe first conductivity parameter of is calculated in step 300.

After the first conductivity parameter is determined, an additional average measured voltage value U associated with said supply point SP in grid 1.₁And average current value I₁Was received in step 400, and in step 500, the point of consumption C in grid 1.₁, C₂, C_nFurther average measured voltage value related to U₂, U₃, U_nAnd average measured current value I₂, I₃, I_nTo receive.

As a result, the further received average measured voltage and average measured current using the calculated first conductivity parameter were analyzed in step 600 to calculate for a given point of consumption or point of supply. Detects deviation of the measured current value from the current value.

It should be understood that this calculation may require occasional updates to reflect the changing grid 1. Updates can be triggered by the construction work of grid 1 and / or can be repeated periodically, for example weekly, monthly, and so on. This update can lead to a periodic (re) restart of this method.

In an embodiment of the invention, the received measured average voltage and average current values are averaged over a predetermined time period. The predetermined time can be any suitable time period, for example 1 minute, 5 minutes, 15 minutes, 30 minutes, 60 minutes.

In one embodiment, this method can be complemented by step 700 following the analysis, said step relating to determining the location of the loss from the point analysis. This determination can be based on a threshold, i.e., after exceeding a threshold, the deviation is determined to be theft. Such a threshold can be any suitable value, eg 5%. The above values can be subject to further fine-tuning.

In a further embodiment, step 600 can be followed by step 800 for analysis, in which the amount of loss is determined from the point analysis. This amount is based on the deviation of the calculated and measured current values for each network element.

In a further embodiment, the calculation of the conductivity parameters in step 300 is based on the optimization method. Unknown matrix element g_mnIs calculated via any suitable optimization method such as quasi-Newton, conjugate gradient, simplex method, fuzzy logic, evolution method, etc.

In a further embodiment, the method also includes step 900, in which step 900 visualizes the results of the analysis provided in steps 600 and / or 700 and / or 800.

Here, moving to application examples in different fields, the present invention can also be used for anomaly detection.

Anomalies are the rapid degradation experienced by two types, namely wire cuts or short cuts, and the slow degradation experienced when the wire or device is exhibiting a change in resistance, for example due to an aging process. Can be done.

However, the slow and quick as used herein relate to methods and repeated use rather than fixed size.

Anomalies lead to changes in conductivity between the nodes involved. Assume that path P1 in the figure is damaged and has increased resistance or the wire is completely interrupted.

Anomalies that represent rapid degradation are perceived as changes in the error function of the process described above. In that sense, the speed can be within a small number of time periods as compared with the number of time periods that form the basis for the calculation of the conductive matrix. For example, if the number of time periods required to calculate the conductivity matrix is 25, then the quick degradation can be within a time period of 10 or less.

As can be seen from Figures 8 and 9, the deviations derived in the process described earlier have different time periods T.₁From T₆Is rather small. But time period T₇To customer C₂And C₃The two nodes represented by are experienced approximately the same deviation. Customer C because the deviation is about two nodes rather than a single customer₂And C₃It can be deduced if the interconnect with is damaged, i.e., experiencing rapid deterioration. This determination can be based on a threshold, i.e., after exceeding a threshold, the deviation is determined to be theft. Such a threshold can be any suitable value, eg 5%. The value can be subject to further fine-tuning.

Therefore, the analysis performed in step 600 can provide both an understanding when power is lost and when the wire has anomalies due to a quick degradation event.

However, if the anomaly represents slow degradation, such as when the link between the customer or customer and / or supplier is aging and leads to a slow change in resistance, then the conductive matrix G_cDue to such changes, anomalies that represent slow degradation may not be derived via error functions. The slowness in that sense can be within a certain number of time periods compared to the number of time periods that form the basis for the calculation of the conductive matrix. For example, if the number of time periods required to calculate the conductive matrix is 25, the slow degradation can be within 10 or more time periods. However, such slow degradation can be rather easily detected by comparing the conductive matrices over time.

First matrix G with first conductivity parameter_{c, 1}But at the time t₁Another matrix G calculated in and having a second conductivity parameter_{c, 2}But t₁At a different time t₂Assuming that it is calculated in, the conductive matrices should be almost identical if no changes in the grid are observed.

However, if the wire, customer, or supplier experiences slow deterioration, it concerns the wire between the abnormal customer, supplier, and / or customer (s) and / or supplier. The first and second conductivity parameters deviate over time.

Therefore, by comparing each parameter of the conductive matrix related to the second time period with each parameter of the previous one, that is, the conductive matrix related to the first time period, even anomalies that represent slow degradation. It is possible to detect. This determination can be based on a threshold, i.e., after exceeding a threshold, the deviation is determined to be theft. Such a threshold can be any suitable value, eg 5%. The value can be subject to further fine-tuning.

Obviously, slow degradation detection can be performed independently. This is because it is not necessary to calculate the current value based on the calculated conductivity parameters as well as the average measured voltage and average measured current received.

On the other hand, since the conductivity parameter calculation can be repeated, this can be easily integrated.

In another embodiment of the invention, the above method is incorporated into a system 10 configured to detect losses in the grid.

First logical receiver RX such system 10 configured to receive average measured voltage and average measured current values associated with one or more supply points in the grid._SPAnd the second logical receiver RX, which receives the average measured voltage and average measured current values associated with multiple points of consumption in grid 1._CIt can be implemented within a computer or controller that includes.

The logical receiver RX_spAnd RX_cThe supplier (s) SP and customer C₁, C₂, C_nIt can be carried out within one or more network cards connected to each meter located in. As shown, said logical receiver RX_spAnd RX_cCan be implemented within a common physical receiver NIC.

The system 10 also includes a calculation unit ALU configured to calculate the first conductivity parameter based on the received average measured voltage value and average measured current value. Each compute unit ALU can be any type of microprocessor, microcontroller, FPGA, ASIC, or analog.

First logical receiver RX_SPIs further configured to receive further measured average voltage and average current values associated with said supply point SP in grid 1, said second logical receiver RX._CFurther, the point of consumption C in the power grid 1.₁, C₂, C_nFurther configured to receive further measured average voltage and average current values associated with.

The logical receiver RX_spAnd RX_cThe supplier (s) SP and customer C₁, C₂, C_nIt can be carried out within one or more network cards connected to each meter located in. As shown, the logical receiver RX_spAnd RX_cCan be implemented within a common physical receiver NIC.

The calculation unit ALU further uses the calculated conductivity parameters to detect all deviations of the measured current value from the calculated current value for a given point of consumption or supply. It is configured to analyze the further received average measured voltage and average measured current.

If the ALU does not provide sufficient memory for the calculation, a conventional memory MEM can be provided, which is used for the purpose of storing the result, the intermediate value of the calculation, and calculating and / or displaying it. It enables the retrieval of the data. Specifically, the memory can be arranged to store one or more previously calculated conductive parameter matrices to detect deviations associated with slow degradation of the connection.

In a further embodiment of the invention, the compute unit ALU is further configured to average the received measured average voltage and average current values over a predetermined time period.

In a further embodiment of the invention, the compute unit ALU is further configured to determine the location of the loss from point analysis.

In another embodiment of the invention, the computational unit ALU is further configured to determine the amount of loss from point analysis.

According to another embodiment of the invention, the calculation unit ALU is further configured to calculate conductivity parameters based on an optimization method. The optimization method is selected from the group including quasi-Newton, conjugate gradient, simplex method, fuzzy logic, and evolution method.

According to a further embodiment of the present invention, the calculation unit ALU further calculates the second conductivity parameter based on the received further average measured voltage value and further average measured current value for calculating the second conductivity parameter. The calculation unit ALU is further configured to analyze by comparing the first conductive parameter with the second conductive parameter.

However, it should be understood that any suitable method can be selected. The optimization method can be selected so that additional parameters such as time need and / or accuracy and / or efficiency are taken into account.

The goal of the optimization is to minimize the error, i.e. the difference between the measured current and the simulated current is minimized and eventually reach 0, i.e. no error and the model The matrix elements are modified when calculating the matrix elements so that they are true equivalent circuits of grid 1.

In another embodiment of the invention, the computational unit ALU is further configured to visualize the results of the analysis.

Such visualization can be performed by an internal WWW server that provides computationally based graphics data via the transmitter TX of the network interface card NIC, or as a video signal DIS directed to a monitor. be able to.

With the present invention, it is now possible to detect theft simply by receiving and calculating measured values without knowing the exact network setup of the grid. This not only makes this method inexpensive, but also achieves that it does not require specially trained personnel.

Accuracy is greatly enhanced by making it possible to detect not only the time of theft, but also the location and amount of theft.

Although the present invention has been illustrated and described in detail in the drawings and the aforementioned description, such illustration and description must be considered to be exemplary or exemplary and not limiting, and the present invention is: It is not limited to the disclosed embodiments.

Other modifications to the disclosed embodiments can be understood and implemented by those skilled in the art from the drawings, the present disclosure, and an examination of the appended claims in the practice of the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude more than one. The mere fact that certain means are listed in different dependent claims does not indicate that the combination of these means cannot be used in an advantageous manner. All codes within the scope of the claims shall not be construed as limiting the scope.

The present invention can be practiced with other specific devices and / or methods. The embodiments described are merely exemplary in all and must be considered non-limiting.

Those skilled in the art will appreciate that all block diagrams herein represent a conceptual diagram of an exemplary network that implements the principles of the invention.

Specifically, the scope of the present invention is shown not by the description and drawings of the present specification but by the appended claims. The meaning of the claims and all changes contained within the scope of equality must be included within that scope.