Flame monitor

[Detailed Description of the Invention]

<Use public expenditure with industry> The present invention relates to the flame monitor which a flame monitor can be started, especially surveillance accuracy of a flame can be sharply made with Mukai. <Prior art and its problem> In various kinds of combustion equipments including large-sized Silly of a place of business, it is very important from the standpoint of economical efficiency and safety to carry out right mG This detection of the combustion state. The increase in the The-Na installation number accompanying large scale-ized * of a boiler when the large-sized boiler for plant is explained to an example, Much more improvement in the reliability of flame detection of a burner is desired by change etc. of the burning behaviour accompanying the burner ignition accompanying the formation of middle load employment of a large-sized boiler, the increase in the number of times of fire extinguishing, the implementation of the combustion method that reduces nitrogen oxide (NOx), and also diversification of fuel. Although it is what detects any of heat, light, and electric nature which a flame has as a flame detector here they are, and detects a flame, Among these, thing for which the method of detecting light and detecting a flame not only detects the existence of a flame, but it analyzes the wave number i of the light which a flame emits .. Since it is fundamentally possible to analyze the quality of a flame, the information on formation flame About me many can be acquired. The flame detection judgment value (threshold) beforehand set up about the light volume which detected the supervising system in the conventional flame monitor, When a size relation with the detected light volume is compared simply and the detected light volume has not reached a threshold, it is judged as the inside of flame-less Firefighting, When the detected light volume is higher than a threshold, it is judged as those with a flame (under combustion), and it constitutes so that this detection signal may be outputted to a control device, 5 news device, a monitoring device, etc. Although there are some which have set up the threshold respectively according to the kind of fuel with a complicated device somewhat, it is to each fuel. ' Even if it attached, and only one threshold was given but the use range was expanded, the accuracy itself did not carry out separate installation improvement. light volume with those high whom the burner which there is a problem in the judgment logic itself that the adult direction of light volume is burning in actual flame surveillance, and performs flame detection is extinguishing -- " -- it may detect. When it is going to detect the existence of ignition to ignition burner 31 in flame detector 32 in Drawings 5 and 6, While 1 ignition burner 31 extinguishes the fire, a lot of light of flame FL of the main burner 3o is detected, on the other hand, the light of this main burner will be interrupted by the flame of ignition burner F2 with less light volume during ignition burner 31 ignition, and detection light volume will fall on the contrary. For this reason, an incorrect judgment will be made and the signal mistaken in the control system will be outputted. This means that incorrect judgment may be produced also about the case where not only between an ignition burner and the main burners but two or more burners approach, and are only arranged, and there is a possibility that proper control of combustion equipment may become impossible. Object > of Invention The present invention is constituted in view of the problem mentioned above, and it aims at providing the flame monitor which can detect the existence of the flame of a predetermined burner correctly. Outline > of Invention In short, the present invention compares with a threshold based on these characteristic values light volume which set a threshold to that light volume of a state of each, and was detected with a characteristic value of light volume under ignition to a burner for which a flame should be detected, and fire extinguishing. Therefore, it is the flame monitor which carried out circuit composition so that existence of a flame might be detected. <Example> It explains using a drawing per example of the present invention below. By 1st [ The ] figure smell, numerals 4 are 7 REME detector A head which performs flame detection, for example, use for optical detection and its transfer the optical fiber which the artificer etc. have proposed separately. Numerals 16 are main parts of a monitoring instrument which judge the existence of a flame by this detected light, and these flame monitors are respectively installed to each burner. Frame detector head 4 is first installed in the position which was most suitable for detecting the flame of a burner. Head watch detects the light volume (or flicker of a flame) of a flame, passes through table 13 as a lightwave signal (in the case of an optical fiber, a detection sensor is an electric signal in Ah 6 case at photoelectric conversion members, such as 7 Otosell), is inputted into Teplo amplifier 15, and it is changed into an electric signal and it is amplified. Human power of the amplified electric signal (voltage, current value) is carried out to the analog / digital Funny Piss #(a /D conversion W is called below) 6 in the main part of a device on table 14. This digital signal is inputted into 7-Rie converter 7, it is changed into the size for every frequency, and this signal is inputted into smooth machine 8 and characteristic value operation part 9. It learns by calculating the characteristic of the burner flame which should be detected beforehand, before calculating the flame characteristic in predetermined light volume in characteristic value operation part 9, and inputting this operation result into storage part 10. That is, since the existence of a flame is not detectable yet at the time of study, the existence of a flame is directed in this circuit. A detection signal is calculated in this state, the flame characteristic at the time of a flame owner and the flame characteristic at the time of-less [ flame ] are computed respectively, and this characteristic is respectively inputted into storage part 10. In this case, it enables it to call memory corresponding to switching signal 17 of burner No. 9, and in the case of an opening signal, the characteristic of With flame is called, and, in the case of a closed signal, the characteristic of flame heat is called respectively. The flame characteristic computed in operation part 9 is the characteristic in a statistical meaning, and average value and distribution are shown. In the device (however, study is suitably performed also in flame detection work) which finished study, a digital signal is inputted into characteristic value operation part 9 and smooth sections 8 like the above-mentioned. The operation process is first described about characteristic value operation part 9. In the operation part, the average value of a signal (light volume is explained to an example below) and distribution about the detected light are calculated. Namely, when the output of smooth sections 8 mentioned below is set to X and the switching signal of a former valve is Open, it averages with a stone, it sets distribution to delta 1, in the case of a closed signal, average x2. distribution is set to delta, and it inputs it into storage part 10. Under the present circumstances, the characteristic value to last time required in order to calculate a characteristic value is read from storage part 10, and is used. Assuming here that the light volume in a combustion state is a normal distribution if the whole combustion system, such as a boiler's, is stable is that there is a statistics top meaning, Since the characteristic value of the light volume of a predetermined time should follow the collective probability distributions sampled from the population of the normal distribution, following 7 distribution (student distribution) is predicted. When probability degree variable X generally follows 7 distribution, the probability that X belongs to the group by average value and the deviation of X can be searched for. Probability P (probability which is extinguishing the fire) belonging to probability PK (probability which is burning) that present value (output from smooth sections 8) X belongs to the group at the time of combustion by this, and the group at the time of fire extinguishing can be searched for. Although the signal which removed the noise on the other hand and was smoothed is outputted from smooth sections 8 and a characteristic value signal is outputted also from another side characteristic value operation part 9, both are inputted into comparison operation section 11, and comparison operation section 11 calculates based on the probability theory mentioned above. For example, in the case of 4>4 (graph of the lower berth of Drawing 2), probability P8 that present value X is burning serves as a portion shown with a slash, and can calculate P2 similarly. If a threshold is more specifically set up about X8, an interference value is set up about 4 and distribution of delta and an interference value is set to delta 1 for distribution of a threshold, as shown in Drawing 4, they will be 99% or more of the probability of Tamotsu of a flame, and -3delta at -delta, for example. The existence of a flame can be judged as less than 99% of probability of flame heat by less than 99% of the probability of With flame, and +delta'' by 99% or more of the probability of flame heat, and +3delta. In X Engineering <x! (i.e., when the direction at the time of fire extinguishing detects large light volume), 2nd [ The ] figure upper row can show, and can search for probability Pl which turns into probability P2 that the slash part by present value X is not burning, and is burning similarly. Next, the existence of a flame is judged by the method which compares RR size A situation by judgment part 12 smell, and is shown in Drawing 4 by it. However, the case of P and -P, and the difference of P1P2 are very those (portion of the slash of Drawing 4) to which the reliability of judgment falls to a small Low case by the above method, and it judges by adding the following logic in this case. That is, like Drawing 3, the frequency characteristic of light volume change shows the peculiar characteristic respectively by the existence of a flame, and shows a clear difference between frequency f thru/or f, especially. That is, by With flame, on the frequency between f, it inclines, and Le 1 is smaller than inclination As in the section of flame heat, and inclination 4 of With flame has the characteristic of becoming larger than inclination A4 of flame heat, between 12 and f, on the contrary. Therefore, when judgment by the above-mentioned method is difficult, flame existence is judged using the frequency characteristic of this light volume. Effect> The existence of a flame can be correctly judged by carrying out the present invention, and according to this invention, the existence of a flame is certainly detectable, although especially the direction at the time of fire extinguishing produced incorrect judgment conventionally in many cases when detection light volume increased.

[Brief Description of the Drawings]

The diagram in which Drawing 1 shows the flowchart of the device of the present invention, and Drawing 2 shows the relation between light volume and probability distributions, the diagram in which Drawing 3 shows the relation between the frequency of light volume change and light volume, the key map in which Drawing 4 shows the judging method of flame existence, Drawing 5, and Drawing 6 are schematic diagrams showing the detection state of flame existence. 8- -... Smooth machine 9 [ ...... Judgment part ] ...... Characteristic value operation part 10 ...... Storage part 11 ...... Comparison operation section 12 Drawing 2 f t2 t3 Okasu collection 1000 few Imaginary