JPH0540115A - Estimating method of cause of degradation of oil - Google Patents

Abstract

PURPOSE:To apply prescribed references and techniques to execution of an estimating method of a cause of degradation of oil wherein the cause of degradation of a specimen oil is estimated from a plurality of causes of degradation on the basis of the distribution of composition of a gas component contained in the specimen oil. CONSTITUTION:The degree of assurance of the genuineness of some cause of degradation is set beforehand as a function for the composition ratio of each gas component, in regard to each of a plurality of causes of degradation. The composition ratio of the gas component of the specimen oil is multiplied by the degree of assurance corresponding thereto in regard to each of the causes of degradation in a plurality, and the degree of assurance of that cause of degradation is judged to be higher as the product of the multiplication is larger.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present application relates to a method for estimating the cause of deterioration of oil such as insulating oil of an oil-filled transformer, and more specifically, a plurality of methods depending on the composition distribution of gas components contained in a sample oil. The present invention relates to an oil deterioration cause estimation method for estimating the deterioration cause of a sample oil from the deterioration cause.

[0002]

2. Description of the Related Art Hereinafter, as an example of oil whose cause of deterioration is to be estimated, an insulating oil sealed in an oil-filled transformer will be described. The insulating oil of such an oil-filled transformer deteriorates due to various causes such as aging and arc discharge. Therefore, the manager of the oil-filled transformer must periodically sample this insulating oil to judge the life, failure, etc. of the transformer and perform appropriate processing. At this time, the cause of deterioration of the oil is investigated, but conventionally, the method described below has been used to estimate the cause of deterioration.

That is, the collected sample oil is analyzed for gas components contained in the oil by a gas chromatograph or the like, and the cause of deterioration is determined from the distribution state of the component analysis (in the oil-filled transformer, the cause of failure). ) Was estimated based on the feeling of the experienced person and the judgment from time to time.

[0004]

However, this estimation work is performed by an experienced person using the gas component composition graph of the sample oil and the gas component composition graph of the oil of which the cause of deterioration is known (this is referred to as a reference graph). Is a work to judge by comparing,
The cause of deterioration and the large number of gas components are very complicated work for the worker and require skill, and this work has to be done only by relying on an experienced specialist. In addition, due to such circumstances, there is no uniform judgment standard among workers, and the standard varies depending on the management worker and estimation time.Therefore, management of oil-filled transformers based on rational judgment results It was not possible to create a plan, etc. (In the end, the safest side has been the management in the past).

Therefore, an object of the present application is to obtain an oil deterioration cause estimating method capable of performing the oil deterioration cause estimating work according to a certain standard and method.

[0006]

Therefore, the characteristic means of the method for estimating the cause of oil deterioration according to the present invention is, for each of a plurality of causes of deterioration, the degree of certainty that the cause of deterioration is the cause of deterioration. It is preset as a function to the ratio, and within a single cause of deterioration, the certainty factors corresponding to the composition ratios of the gas components of the sample oil are synergized, and the larger this product value, the greater the certainty factor of the deterioration factor. It is judged to be high, and its actions and effects are as follows.

[0007]

According to the estimation method of the present application, for each specific cause of deterioration with respect to a specific gas component composition ratio, the certainty factor that each deterioration cause is the deterioration cause of the sample oil (in other words, the deterioration cause is The appearance probability of this composition ratio in the case of being the cause of deterioration of is set in advance as a function of the composition ratio. This function is determined, for example, by making full use of expert's knowledge, or is determined from data obtained so far. Then, the composition ratio of each gas component appearing in the sample oil is converted into a certainty factor by the above-mentioned function, and this is integrated for all gas components within a specific cause of deterioration. Here, it is determined that the greater the certainty factor regarding the specific deterioration cause, the higher the certainty factor as the deterioration cause of the sample oil.

[0008]

Therefore, according to the method of the present invention, the estimation work of the cause of oil deterioration is performed based on the function set between the composition ratio of the gas component and the certainty factor, according to a certain standard and method. It can be done. Therefore, not only an expert with a long experience, but also a general worker can make inferences according to a certain standard and method, and the reproducibility of this inference work can be secured.

[0009]

Embodiments of the present application will be described with reference to the drawings. FIG. 1 is a flowchart showing the process of estimating the cause of deterioration that employs the method for estimating the cause of deterioration of oil according to the present application. The following outlines this procedure, taking the case of applying to the insulating oil of an oil-filled transformer as an example. (1) Take a sample of insulating oil from the oil-filled transformer (#
1). (2) The composition distribution of gas components contained in this sample oil is detected and a composition ratio graph is created (# 2). Here, the concentration (absolute value) of each gas component is also detected separately. (3) The deviation between the above-mentioned composition ratio graph and the reference graph, which is a graph of the gas component composition of deteriorated oil that causes specific deterioration, is checked to determine the first certainty factor E1 of the deterioration cause (# 3). (4) Based on each composition ratio in the above composition ratio graph,
The certainty factor of this deterioration factor for each composition ratio under each specific deterioration factor is obtained by a membership function, and integrated for all gas components to obtain the second certainty factor E2 of the deterioration factor (#
4). (5) The final estimated cause is discriminated from the first and second certainty factors (E1, E2) determined by # 3 and # 4 (# 5). (6) For each cause, the cause certainty factor (probability that each cause is the cause of deterioration) is output (# 6). (7) Based on the cause certainty factor in # 6, a judgment (comprehensive judgment as to what causes this oil-filled transformer is deteriorated) is output (# 7). (8) Based on the above results, the future oil-immersed transformer processing method is output (# 8).

The outline of the procedure is as described above, but the configuration in each step will be described in more detail below. # 1 Step This step is a sample oil collection step and is usually # 2.
A sample amount (about 50 cc) necessary for the above is collected. In this case, it is necessary to be careful not to collect the oil remaining in the lower part of the oil-filled transformer where normal heat convection does not occur. Step # 2 This step is a step of creating a composition ratio graph of the sample oil, and a detailed diagnostic analysis table as shown in Table 1 and a composition ratio graph as shown in FIG. 2 are created. First, Table 1 will be described. This table describes various attributes (capacity, voltage, year of manufacture, serial number, manufacturer, etc.) of the oil-filled transformer to be diagnosed, as well as the date of sampling the sample oil. Now, the analysis of the gas components is performed by a gas chromatograph, and the target components are carbon monoxide, hydrogen,
Methane, acetylene, ethylene, ethane, propylene,
It is propane. Table 1 shows actual analysis values (ppm)
And composition ratio (composition ratio when total gas amount is 100%)
Is displayed.

[0011]

[Table 1]

Next, the composition ratio graph shown in FIG. 2 will be described. The abscissa of this graph shows the cause of deterioration described in the above-mentioned precise diagnosis analysis table, and the ordinate shows the composition ratio. Here, the analysis value of oil is shown by the solid line, and the reference graph by aging is shown by the broken line for reference. Steps # 3 and # 4 These steps are steps of estimating the cause of oil deterioration and include two inference processes. First, the background common to these processes will be described. The causes of the gas components contained in oil as described above (causes of oil deterioration) are arc discharge (including partial discharge), local heating (K type) of metal structure (other than coil), dehydrogenation reaction, aged It can be roughly divided into deterioration, overload, and coil heating (M type). The component composition of the gas generated by these causes shows a unique pattern according to the cause. These patterns are shown in FIG. 3 along with the cause of their deterioration. Further, here, for each pattern, data is distributed for each of the causes of deterioration and with a certain width (standard deviation) for each gas component. This distribution state is understood about the composition ratio of oil deteriorated due to a specific cause of deterioration. This distribution state is shown in the reference graph due to arc discharge shown in FIG. 3 when the gas component is hydrogen (here, the average value of the distribution is the composition ratio value position of each gas component in the reference graph. , Standard deviation is an amount that represents the range of values that the actual composition ratio can take.) Here, the stronger the correlation between the gas component and the specific cause of deterioration, the smaller the ratio (S _ij / m _ij ) of the standard deviation (S _ij ) of the composition ratio to the average (m _ij ) of the composition ratio. Table 2 shows the relationship between the average (m _ij ) of the composition ratios and the standard deviation (S _ij ) of the composition ratios when the cause of deterioration is arc discharge.

[0013]

[Table 2]

On the other hand, when inferring the cause by looking at the composition ratio graph of the sample oil, the "probability (confidence)" that the deterioration is "probably due to this cause" is compared with the above-mentioned reference graph. It is also possible to express functionally in relation to the composition ratio of each causative gas component. Such "confidence"
The membership function is expressed by the probability of 0 to 1. FIG. 4 shows the distribution of the certainty factor that the deterioration cause is arc discharge. Explaining the certainty factor when the gas component is hydrogen, a1 and a5 parts are regions with a certainty factor of 0, a2 and a4 parts are regions located at intermediate positions of the certainty factors 0-1 and a3 part is This is a part with a certainty factor of 1. Also in FIG.
In addition, the membership function in which the cause of deterioration is arc discharge is shown for each gas component. Here, in this membership function, the stronger the correlation between the gas component and the specific cause of deterioration, the greater the certainty factor and the wider the region of the greater certainty factor with respect to the composition ratio of the gas component.
In addition, regarding each cause of deterioration, correspondence with gas component elements with high certainty for a specific cause of deterioration and wide range of area is acetylene for arc discharge, ethylene for local heating other than coil, dehydrogenation. Hydrogen for the reaction, carbon monoxide and hydrogen for aging, hydrogen, ethylene, ethane, propylene for overload operation, and carbon monoxide, ethylene, propylene for local heating of the coil. As described above, the cause of deterioration, the average value on the reference graph for each of the gas components, the standard deviation and the cause of deterioration,
The membership function for each gas component is appropriate for the expert. Where mean, standard deviation,
There are 48 membership functions (the number of causes of deterioration 6 × the number of gas components 8) = 48, and the membership function is defined as a trapezoidal function, and experience is often represented.

Under the above assumptions, the inference process and the fuzzy inference process based on the root mean square error will be described below. (A) Inference step based on root mean square error (# 4, FIG. 6) This step is step # 4 shown in FIG. 1, and the details of this step are shown in FIG. As shown in the figure, this step is a step of obtaining the deviation value for each deterioration cause with respect to the reference graph of the composition ratio graph, and in # 31, the average value and the standard deviation of the composition ratio of each gas component for each deterioration cause are read from the table. While calling, the deviation value for each deterioration cause shown in # 32 is calculated. The details of the formula for calculating the deviation value for each cause of deterioration are shown below.

[0016]

[Equation 1] i: the type of degradation causes (1-6; 1: aging ......) j: type of gas (1~8; 1: CO2: H 2, ......) e i: deterioration caused by deviation value x _j: Composition Composition ratio (%) of each gas j in the ratio graph m _ij : Average value of composition ratio for each gas component j in the specific cause of deterioration i (%) S _ij : For each gas component j in the specific cause of deterioration i Standard deviation of composition ratio (%) (This is an estimate of the standard deviation of _mij as seen by experts.)

The deviation value e _i for each deterioration cause is obtained from the above equation, and the position of the obtained deviation value e _i for each deterioration cause in the normal distribution is determined to determine the first certainty factor E1 (# 33). ). For example, if the deviation value e _i for each deterioration cause is 0, the first certainty factor is 100%. This step (# 3)
The purpose of is to find out how much the composition ratio graph deviates from the reference graph, and the average value (m _ij ) and standard deviation (S _ij ) of the composition ratio described above The deviation in the composition ratio distribution of each gas component of oil is obtained, and the deviation is averaged for all gas components and converted using a normal distribution table to obtain the first certainty factor E1.
By adopting this method, it is possible to determine the order of which reference graph is closest to any input data.

(B) Inference step by fuzzy inference (# 5, FIG. 7) This step is step # 5 shown in FIG. 1, and the details of this step are shown in FIG. As shown in the figure, this step is a step of directly obtaining the second confidence factor E2 in this inference method based on the membership function, and in # 41, the membership function of the composition ratio of each gas component in each cause of deterioration is While calling from the table, the second certainty factor E2 shown in # 42 is calculated.
The calculation formula applied to the second certainty factor E2 will be described in detail below. In this embodiment, "and" is calculated for the membership function μ _ij (x _j ) for each gas. In fuzzy arithmetic, a limiting product or an algebraic product is used, but this time the algebraic product is adopted. Therefore, the second certainty factor by the fuzzy inference of the deterioration cause i with respect to the composition ratio (x _j ) of each gas component in the composition ratio graph is

[0019]

[Equation 2] Becomes

The purpose of this step is to proceed with the inference according to the criteria of the expert, and the membership function of each gas component described above is used. And by adopting this method, even for any input data,
You can make judgments according to the judgment criteria of experts. Further, in the present embodiment, more accurate determination is possible by combining with the above-mentioned root mean square error.

Step # 5 This step is a step of comprehensively judging the results of the inference process using the above-mentioned mean square error and the fuzzy inference process. By averaging the certainty factors obtained by both, the final result is obtained. Confidence is being determined. # 6, # 7, # 8 Steps These steps are steps for outputting the results obtained in the above steps. The output format is shown in Table 3.

[0022]

[Table 3]

Here, in the determination shown in # 7, # 5
According to the final certainty factor obtained in, for example, aged deterioration 6
If 0%, overload 15%, and other 0%, it is determined to be deterioration over time. Furthermore, depending on the final confidence level,
Three types of judgments are made: "Extremely 0 tendency", "Extremely 0 tendency" and "Slightly 0 tendency". (If none of the final confidence factors is small, "it does not apply to any cause.") Next, in the output of the processing method shown in # 8, the absolute value (analytical value ppm) of the input data is finally inferred. Depending on the cause, the following 6 types are output as to what to do. Output format is "Stop immediately and inspect""Need to consider countermeasures""1 / M monitor""1 / 3M monitor""1 / 6M monitor""1 / 3Y monitor" Be divided.

[Reference Data] In the above example, the deterioration cause is an aging change. For reference, the results when the deterioration cause is arc discharge are shown in Table 4 and FIG. An example is shown in Table 5 and FIG.

[0025]

[Table 4]

[0026]

[Table 5]

[Other Examples] In the above examples, the standard deviation of the composition ratio and the membership function were set empirically. The value may be set from the measured value of Further, the method of the present application can be applied not only to gas analysis of an oil-filled transformer but also to oil diagnosis of other equipment.

It should be noted that reference numerals are given in the claims for convenience of comparison with the drawings, but the present invention is not limited to the configurations of the accompanying drawings by the entry.

[Brief description of drawings]

FIG. 1 is a flowchart showing a deterioration cause estimating step.

FIG. 2 is a diagram showing a composition ratio graph of sample oil.

FIG. 3 is a diagram showing a gas component composition distribution of deteriorated oil due to each deterioration cause.

FIG. 4 is a diagram showing a relationship between a gas component and a membership function when arc discharge is a cause of deterioration.

FIG. 5 is a diagram showing a membership function when arc discharge is a cause of deterioration.

FIG. 6 is a diagram showing details of a cause inference step based on a root mean square error.

FIG. 7 is a diagram showing details of the cause inference step by fuzzy.

FIG. 8 is a diagram showing a composition ratio graph of sample oil in which deterioration is caused by arc discharge.

FIG. 9 is a diagram showing a composition ratio graph of sample oil in which deterioration is caused by coil heating.

Claims (4)

[Claims] 1. An oil deterioration cause estimation method for estimating a deterioration cause of a sample oil from a plurality of deterioration causes based on a composition distribution of gas components contained in the sample oil, wherein each of the plurality of deterioration causes is The certainty factor that this cause of deterioration is the cause of deterioration is preset as a function for the composition ratio of each of the gas components, and within the single cause of deterioration, the composition ratio of each of the gas components of the sample oil is changed. A method for estimating the cause of deterioration of oil, which synergizes the corresponding certainty factors and determines that the larger the product value, the higher the certainty factor of the deterioration cause. 2. The stronger the correlation between the gas component and the specific cause of deterioration, the larger the certainty factor, and the wider the width of the region having the greater certainty factor with respect to the composition ratio of the gas component. 1
The method for estimating the cause of oil deterioration as described. 3. The plurality of causes of deterioration are arc discharge, local heating other than a metal structure, dehydrogenation reaction, aging, overload operation, local heating, respectively, and the gas component is monoxide, respectively. The method for estimating the cause of deterioration of oil according to claim 1, which is carbon, hydrogen, methane, acetylene, ethylene, ethane, propylene, or propane. 4. When the specific cause of deterioration is arc discharge, local heating other than coil, dehydrogenation reaction, aging deterioration, overload operation, local heating of coil, the certainty factor for the specific cause of deterioration is The gas components are large and the width of the region is wide, respectively, acetylene for the arc discharge, ethylene for local heating other than the coil, hydrogen for the dehydrogenation reaction, carbon monoxide and hydrogen for the aging deterioration, The method for estimating the cause of deterioration of oil according to claim 2, wherein hydrogen, ethylene, ethane, and propylene for overload operation and carbon monoxide, ethylene, and propylene for local heating of the coil are used.