Transformer time-varying reliability evaluation method combining degradation process and real-time state

Abstract

The invention relates to the technical field of thermal power plant equipment state evaluation, and discloses a transformer time-varying reliability evaluation method combining a degradation process and a real-time state, and the method comprises the following steps: 1, collecting a basic data set; step 2, when missing data exists, filling the missing data in combination with a universal data reduction method; 3, constructing a degradation equation of a historical degradation track of the power transformer; calculating a time-varying aging index HAi based on the degradation equation; calculating the internal aging failure probability of the power transformer at the moment ti; 4, constructing and improving a logistic regression model; determining parameter values of the logistic regression model; establishing an external random fault probability evaluation model of a meteorological area where the power transformer is located; and 5, calculating to obtain the time-varying reliability of the power transformer in combination with the degradation process and the real-time state. According to the method, the time-varying reliability of the power transformer can be effectively evaluated, the precision is relatively high, and the practicability is relatively high.

Description

Technical Field

[0001] The invention relates to the technical field of equipment status assessment in thermal power plants, and in particular to a transformer time-varying reliability assessment method combining degradation process and real-time status. Background Art

[0002] With the rapid development of the economy and the continuous progress of society, electric energy has become an indispensable basic material support for all walks of life. As the world's most important power generation method, thermal power generation occupies an important position in power production. It is particularly important to ensure the reliable and continuous transmission of electric energy from thermal power plants. As the key hub equipment for realizing the transmission of electric energy from thermal power plants, the operating reliability of power transformers not only directly affects the economic benefits and service quality of power plants, but also has an important impact on the power supply reliability of the entire power system. However, with the continuous extension of the service life, the power transformers of the thermal power plants in operation are directly exposed to the outside world for a long time. After being subjected to the internal electrical, thermal, and mechanical stress shocks and the external high temperature, rainfall and other complex weather coupling effects for a long time, they will gradually enter the aging stage. This will lead to a significant increase in the frequency of generator set tripping and shutdown accidents caused by power transformer failures, which seriously threatens the reliable and continuous transmission capacity of electric energy and even causes certain safety risks. In this context, it is particularly important to evaluate the time-varying reliability level of power transformers.

[0003] However, most of the existing time-varying reliability level assessment schemes for power transformers perform poorly (low accuracy and lack of practicality). The specific reasons are as follows: (1) In terms of multi-source data processing of power transformers: Due to the limitations of objective conditions such as geographical environment and economic factors, most of the current power transformer monitoring devices use low-cost distribution carriers and industrial wiring transmission methods. When the above communication channels are subjected to strong magnetic interference such as high voltage and high current shocks, data loss often occurs during the collection, transmission and storage process, limiting the quality of multi-source data and its subsequent application value. Existing methods usually focus on directly using raw data to carry out reliability modeling and assessment, but fail to fully consider the objective problem of data loss in the field, making it difficult to effectively improve the accuracy and practicality of modeling. (2) In terms of internal fault probability modeling of power transformers: Existing methods usually only determine the current operating state of the power transformer based on the current monitoring data, and then conduct internal aging fault probability assessment, without making full use of the historical monitoring data of the power transformer in the past. (3) In terms of external random fault probability modeling of power transformers: Current methods mainly give the random fault probability of power transformers under different weather conditions (such as light rain, moderate rain, etc.) based on weather state classification (such as three states). However, these methods usually only consider the impact of a single weather condition and ignore the coupling effects of multiple weather factors such as precipitation, ambient temperature, air humidity, wind speed, etc. (4) In addition, the current methods have not yet provided a complete module architecture design, making it difficult to effectively realize the practical application of remote monitoring of the time-varying reliability of power transformers in thermal power plants. Summary of the invention

[0004] The present invention aims to provide a transformer time-varying reliability assessment method combining degradation process and real-time status, which can effectively assess the time-varying reliability of power transformers with high accuracy and strong practicability.

[0005] The basic scheme provided by the present invention is: a transformer time-varying reliability assessment method combining degradation process and real-time status, comprising the following steps:

[0006] Step 1, collecting historical oil chromatography monitoring data of power transformers, sensor data and weather monitoring data recorded by the meteorological platform as basic data sets;

[0007] Step 2: When there are missing data in the basic data set, a generalized data restoration method combining empirical classification and improved K-nearest neighbor algorithm is used to fill in the missing data in the oil chromatography monitoring data or weather monitoring data;

[0008] Step 3: Based on the oil chromatographic monitoring data, combined with the Wiener degradation process and recursive filtering algorithm, the degradation equation of the historical degradation trajectory of the power transformer is constructed, namely, s i =s i-1 +μ i-1 (ti -t i-1 )+σε i ; Among them, s i It indicates that the power transformer is estimated to be at t according to the oil chromatography monitoring data. i The estimated value of the degenerate observation at time; μ is the drift coefficient; σ is the diffusion coefficient and σ>0, ε i The mean is 0 and the variance is t. i -t i-1 Normal distribution of

[0009] Based on the degradation equation, calculate the power transformer at t i The probability density distribution function of the estimated value of the remaining life at time t is obtained from it. i Time-varying aging index HA i ; Based on the time-varying aging index HA i , calculate the power transformer at t i Internal aging failure probability at time;

[0010] Step 4, constructing a logistic regression model; based on weather monitoring data and historical fault record data of power transformers, using principal component analysis to improve the logistic regression model, analyzing the coupling effect of weather factors on the logistic regression model, and determining the parameter values ​​of the logistic regression model; based on the parameter values ​​of the logistic regression model, establishing an external random fault probability assessment model for the meteorological area where the power transformer is located; the external random fault probability assessment model is used to calculate the external random fault probability of the power transformer;

[0011] Step 5, combining the internal aging failure probability output from step 3 and the external random failure probability output from step 4, calculates the time-varying reliability of the power transformer combining the degradation process and the real-time status.

[0012] The working principle and advantages of the present invention are:

[0013] First, we collect multi-source data such as oil chromatogram information and weather data of power transformers, and clean the data through a general data restoration method that combines empirical classification with an improved K-nearest neighbor algorithm; secondly, based on oil chromatogram monitoring data, we use the Wiener degradation process, recursive filtering algorithm, etc. to construct an evaluation method for the probability of internal aging failure of power transformers; at the same time, based on a variety of weather data, we use an improved logistic regression model to construct an evaluation method for the probability of external random failure of power transformers. Finally, based on this method, through system module integration, we can realize the online application of this method. Based on real-time oil chromatogram monitoring data and weather data, we can evaluate the time-varying reliability level of power transformers online and support remote monitoring.

[0014] The transformer time-varying reliability assessment method combining the degradation process and the real-time status of the present invention can effectively assess the time-varying reliability of the power transformer with high accuracy and strong practicality. The key points are:

[0015] First, the evaluation is reasonable and the evaluation results have a high reference value. The power transformer data used in this scheme can all be derived from the existing database, which can fundamentally ensure the rationality of the evaluation results and significantly improve its feasibility in practical engineering applications. In addition, with the continuous increase of monitoring data, this method can continuously update the relevant models, thereby improving the accuracy of the time-varying reliability evaluation results of power transformers.

[0016] Second, the evaluation accuracy is high. This scheme combines multi-source data, comprehensively considers the impact of internal performance degradation of power transformers and external random weather factors, and performs targeted probability calculations. Combining the two probabilities, the time-varying reliability of power transformers combined with the degradation process and real-time status is calculated, which can reasonably and effectively evaluate the time-varying reliability level of power transformers in thermal power plants. It is of great significance to ensure the reliable operation of thermal power plants, and can also provide a more scientific decision-making basis for the on-site formulation of inspection cycles and maintenance strategies for power transformers.

[0017] In particular, this solution fully considers the objective problem of missing data in the field, and designs a generalized data restoration method that combines empirical classification and improved K-nearest neighbor algorithm to fill in the missing data, which can accurately restore the data and help improve the accuracy of subsequent evaluations. In addition, this solution selects the overall monitoring data sequence based on the power transformer for mining and analysis, which can fully analyze the entire performance degradation process involved in the power transformer, and then fully consider the inconsistency of the performance degradation process of individual power transformers, and obtain a more accurate evaluation result. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a method flow diagram of an embodiment of a transformer time-varying reliability assessment method combining degradation process and real-time status according to the present invention;

[0019] Figure 2 A schematic diagram of the input-output relationship between the modules constructed in the method of the embodiment of the transformer time-varying reliability assessment method combining degradation process and real-time status of the present invention;

[0020] Figure 3 A schematic diagram of missing data restoration values ​​and error analysis obtained by different methods in an example of an embodiment of a transformer time-varying reliability assessment method combining degradation process and real-time status of the present invention;

[0021] Figure 4It is a schematic diagram of the comprehensive failure probability values ​​of two power transformers under the same operating time in an example of an embodiment of the transformer time-varying reliability assessment method combining degradation process and real-time status of the present invention. DETAILED DESCRIPTION

[0022] The following is a further detailed description through specific implementation methods:

[0023] The embodiment is basically as shown in the attached Figure 1 and Figure 2 As shown: The transformer time-varying reliability assessment method combining degradation process and real-time status includes the following steps:

[0024] Step 1: Collect multi-source heterogeneous data such as historical oil chromatography monitoring data of power transformers, sensor data, and weather monitoring data recorded by the meteorological platform as the basic data set.

[0025] Specifically, the sensor data includes daily ambient temperature collected by temperature sensors and air relative humidity data collected by humidity sensors. The oil chromatogram monitoring data includes historical oil chromatogram monitoring data of power transformers collected by oil dissolved gas analysis devices derived from equipment records. The weather monitoring data includes micro-meteorological data of the area where the thermal power plant is located derived from a meteorological platform or meteorological station.

[0026] In this embodiment, two specific data storage forms store oil chromatogram monitoring data and weather monitoring data respectively. n and M n As shown:

[0027]

[0028] Where: G n M is the oil chromatogram historical monitoring data matrix corresponding to power transformer n; n is the historical weather data matrix corresponding to power transformer n. T is the total length of the historical sampling period, in days.

[0029] They represent the daily average values ​​of the dissolved gas contents in seven types of oil of the power transformer obtained at the t-th time series collection point in history, corresponding to methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), hydrogen (H2), carbon monoxide (CO) and carbon dioxide (CO2), respectively.

[0030] It represents the six weather factor data of the power transformer obtained at the tth time series collection point in history; among them, the first five are numerical data, corresponding to ambient temperature, relative air humidity, precipitation, wind speed, and sunshine duration; the last one is text attribute label data, which is used to describe the overall meteorological conditions of the day. In addition, if there is missing data, special symbols are marked to distinguish it. For example, if the relative air humidity of the third collection point is missing, it is marked with * to facilitate the subsequent restoration of missing data.

[0031] This step also includes: constructing an overall historical database in the data station of the thermal power plant; the overall historical database is used to classify and store basic data sets. Specifically, the two data matrices of N power transformers are classified and stored respectively (each N matrix), and the construction of the overall historical database can be completed.

[0032] Step 2: When there are missing data in the basic data set, a generalized data restoration method combining empirical classification and improved K-nearest neighbor algorithm is used to fill in the missing data in the oil chromatography monitoring data or weather monitoring data.

[0033] Here, the daily ambient temperature and relative humidity data in the sensor data are incorporated into the weather monitoring data, and then processed together based on the weather data matrix. This setting ensures that the missing data in the basic data set are properly processed.

[0034] In this step, the generalized data restoration method includes:

[0035] Step 2.1, based on the empirical classification criteria, classify the oil chromatography monitoring data or weather monitoring data into different historical sample sets.

[0036] As shown in Table 1, the empirical classification standards corresponding to the oil chromatographic monitoring data include classification according to the alarm information displayed by the oil chromatographic data of the day, specifically including: Category 1, normal; Category 2, partial discharge; Category 3, internal overheating; Category 4, winding circulating current; Category 5, electric flashover; Category 6, continuous spark.

[0037] Table 1 Example of empirical classification standards for oil chromatography monitoring data

[0038]

[0039] As shown in Table 2, the empirical classification standards corresponding to weather monitoring data include: Category 1, thunderstorm; Category 2, strong wind; Category 3, heavy rain; Category 4, ice and snow; Category 5, high temperature; Category 6, cloudy.

[0040] Table 2 Example of empirical classification standards corresponding to weather monitoring data

[0041]

[0042]

[0043] Specifically, taking weather monitoring data as an example, each piece of weather data information is classified according to the overall meteorological conditions of the day (i.e., the characteristic data of the sixth column of the weather data matrix), and each type of historical sample set is recorded as S mi (i=1,…,6).

[0044] Step 2.2, when the oil chromatography monitoring data or weather monitoring data of day a is missing, the neighbor sample with the highest similarity to the data of day a is selected from the historical sample set through the improved K nearest neighbor algorithm, and the missing data is restored according to the weighted value of the neighbor sample.

[0045] Specifically, taking weather monitoring data as an example, considering that under the same meteorological conditions, weather data such as ambient temperature, relative humidity, precipitation, wind speed, and sunshine duration usually show a certain correlation, this correlation reflects the comprehensive impact of different weather conditions on the overall meteorological conditions. Therefore, if the meteorological condition recorded by the historical weather data on day a is r, the historical sample set S is used mr The complete weather records in the data are used to infer the missing data on day a and restore the missing weather data on day a.

[0046] This setting not only avoids the deviations that may be introduced by blind interpolation or time series prediction, but also makes full use of the potential correlation in historical data and improves the rationality of the missing data restoration process.

[0047] The specific operation is as follows: First, the improved K nearest neighbor algorithm is used to select the historical sample set S mr Select num k The complete weather data samples with high similarity to the weather data of day a are selected, and then the last column (attribute label data) of these weather data is deleted. The specific expression is:

[0048]

[0049] Where: m n,a The weather data sample for day a. For example, if the relative humidity value of the second data is *, it means that the collected value is missing and needs to be restored. k Neighbors of m n,a Indicates selection and weather data m n,a The num with the closest similarity k A sample of neighbors.

[0050] Represents data sample m n,a With m base,i(i=1,……,num k ) can be calculated by the following formula.

[0051]

[0052] In the formula, j represents the dimension corresponding to the missing data in the weather data of day a. For example, if the relative humidity value of the second data is missing, the value of j is 2.

[0053] ξ h is the weight of each dimension data, which can be calculated by the following formula:

[0054]

[0055] In the formula, num(Sm r ) is the historical sample set S mr The total number of samples with no missing data.

[0056] p hb is the proportion of the h-th meteorological index in the j-th data sample, which can be calculated by the following formula:

[0057]

[0058] In the formula, It is the normalized value of the h-th meteorological indicator in the j-th data sample.

[0059] According to the above steps, the num with the highest similarity to the weather data sample of day a is selected k neighbor samples, according to num k The weighted value of the j-th dimension data of the neighbor samples is used to restore the missing j-th dimension data. The specific restoration formula is as follows:

[0060]

[0061] In the formula, is the missing value of the j-th dimension data missing in the weather data sample of day a.

[0062] According to the above steps, the missing data in the historical weather data of any power transformer n is restored, thereby improving the historical weather database of the power transformer. The same method is also applicable to repairing the missing values ​​in the historical oil chromatogram data.

[0063] This step also includes: building a data cleaning module in the data station of the thermal power plant; the data cleaning module is used to execute step 2.

[0064] Step 3: Based on the oil chromatographic monitoring data, the degradation equation of the historical degradation trajectory of the power transformer is constructed by combining the Wiener degradation process and the recursive filtering algorithm.

[0065] Specifically, the degradation equation based on the Wiener process in physics (Wiener degradation process) is used to represent the performance aging evolution process of the power transformer during its entire life cycle. The specific performance aging process can be expressed as:

[0066] S(t)=s i +μ(tt i )+σ(B(t)-B(t i ))=s i +μ(tt i )+σB(tt i ),t>t i ;

[0067] Where S(t) represents the performance degradation process function of the power transformer; s i It indicates that the power transformer is estimated at t according to the oil chromatography monitoring data. i The estimated value of the degraded observation at time; μ is the drift coefficient; σ>0 is the diffusion coefficient, B(t) represents the standard Brownian motion, which is a normally distributed random variable, indicating the randomness of the degraded observation.

[0068] The reason for using the Wiener process to describe the degradation process of the power transformer is that the small changes in the degradation of the power transformer in a small time interval are similar to the random fluctuations of small particles in the fluid and air. This scheme is set up in this way, which can effectively characterize and quantify the degradation state of the power transformer.

[0069] Furthermore, the power transformer in the above formula is i The estimated value of the degenerate observation at time s i The power transformer restored in step 2 can be i The oil chromatography data value at the time is calculated:

[0070]

[0071] Where, β(t) is the ratio vector of dissolved gas in oil obtained based on the three-ratio method in the IEC standard, which are C2H2 / C2H4 CH4 / H2 and C2H4 / C2H6 δ is a weight matrix, the sum of the matrix elements is 1. In practical applications, it can be determined by the experience and knowledge of on-site operators.

[0072] Among them, g(·) is a normalization function, which is used to correct the deviation caused by the different dimensions of the characteristic gas ratio (for example, C2H2 / C2H4 is usually less than 0.1, while CH4 / H2 is usually between 0.5 and 1). The details are as follows:

[0073]

[0074] in, and They are the characteristic gas ratio β i (t) lower and upper bounds; U(β i The higher the (t) value, the more obvious the internal aging of the power transformer.

[0075] In the above steps, in order to consider the historical degradation trajectory of the power transformer in the evaluation of the power transformer degradation observation, a random walk strategy is also used to dynamically correct the drift coefficient μ corresponding to each time interval.

[0076] Specifically expressed as: i =μ i-1 +η;

[0077] Where η is the noise term, which obeys a normal distribution with a mean of 0 and a variance of Q. The noise term can be used to simulate the change of the drift coefficient μ in a short monitoring interval, so the expected μ i The μ should be adjusted by the noise term η. i-1 The present invention defines the initial drift coefficient μ0 according to the conditions of the random walk strategy to obey the normal distribution with mean A0 and variance P0. i is a random variable that changes with time, and its value is μ i-1 As a condition.

[0078] Furthermore, the degradation equation of the historical degradation trajectory of the power transformer is constructed: i =s i-1 +μ i-1 (t i -t i-1 )+σε i ;

[0079] Among them, s i It indicates that the power transformer is estimated to be at t according to the oil chromatography monitoring data. i The estimated value of the degenerate observation at time; μ is the drift coefficient; σ is the diffusion coefficient and σ>0, ε i The mean is 0 and the variance is t. i -t i-1 The normal distribution of .

[0080] The initial time when the power transformer is put into operation is t0 = 0; g0 = 0. i The mean is 0 and the variance is t. i -t i-1 The normal distribution of i -t i-1 As ε iThe variance of is required by the properties of the Brownian motion B(t).

[0081] Furthermore, based on the degradation equation, the power transformer is calculated at t i The probability density distribution function of the estimated value of the remaining life at time t is obtained from it. i Time-varying aging index HA i .

[0082] Specifically, the drift coefficient μ i The time series change of reflects the state evolution process of the power transformer, which can be seen from t0 to t i The observed so far is denoted as S 0:i ={s0,s1,s2,…,s i} power transformer oil chromatographic degradation data estimation. This step uses a recursive method to estimate μ i The probability density function p(μ i ∣S 0:i ) to estimate:

[0083]

[0084] In the formula, p(μ i ∣μ i-1 ) is the drift coefficient μ at the given previous moment i-1 In the case of i The conditional probability of p(g i ∣μ i-1 ,S 0:i-1 ) is the drift coefficient μ at a given previous moment i-1 The situation and all previous observations S 0:i-1 In the case of i The conditional probability of i-1 ∣S 0:i-1 ) is the given observation data S 0:i-1 In the case of drift coefficient μ i-1 The probability distribution of p(g i ∣S 0:i-1 ) is a normalization constant. Through this recursive process, the estimate of the drift coefficient can be updated at each time step. Since the model depends on the estimate at the previous moment and the current observation data, it allows the model to gradually improve its prediction as new data arrives.

[0085] In the recursive evaluation of the drift coefficient μ i In the process of , a strong tracking filter is introduced to deal with the occasional jumps or mutations that may exist in the drift coefficient and the degraded observation quantity, and the state prediction error is adaptively adjusted by using multiple suboptimal fading factors:

[0086]

[0087] In the formula, Λ i is the multi-vanishing factor matrix of the i-th recursive evaluation; F k is the Jacobian matrix of the degradation observations of the power transformer.

[0088] Based on the above steps, the probability density distribution of the drift coefficient μi can be obtained, which can be specifically expressed as:

[0089]

[0090] The above formula represents the drift coefficient μ i Satisfy the mean Variance is P i∣i =var(μ i ∣G 0:i )’s Gaussian distribution; μ i With G 0:i The dependencies between them are contained in μ i The mean of the probability density distribution of and variance P i∣i .

[0091] Based on the predefined transformer aging failure threshold w (defined as 1 in this embodiment), the current t i Remaining life of power transformer at time: R i = inf{r i :S(n+t i )≥w|S 0:i};

[0092] Then, based on the degradation equation, the power transformer is calculated at t i The probability density distribution function of the estimated value of the remaining life at time

[0093]

[0094] Where Φ(·) is the cumulative probability density function of the standard normal distribution; μ of the variable i The historical observation data and variance are updated recursively. The four unknown parameters in the above formula are: the mean and variance a0, P0 corresponding to the initial drift coefficient μ0. The mean and variance Q, σ of the normal distribution corresponding to the η parameter. 2 The estimation can be dynamically updated using the maximum expectation algorithm.

[0095] Furthermore, the time-varying aging index HA i Refer to the following formula for calculation:

[0096]

[0097] Where, Lmean is the average aging failure life of the same type of transformer; E(R i ) is the power transformer at t i The estimated remaining life at time R i Expectations; HA i The larger it is, the shorter the remaining life of the power transformer is, that is, the more serious the aging condition is.

[0098] Furthermore, based on the time-varying aging index HA i , based on the two-parameter Weibull distribution model, the power transformer is calculated at t i Internal aging failure probability at time:

[0099]

[0100] Where P aging (t i ) is the power transformer at t i The internal aging failure probability at the moment; α mean and κ mean are the shape parameter and scale parameter of the Weibull distribution respectively. In practical applications, they can be obtained by fitting and evaluating the aging failure statistics of the same type of power transformers.

[0101] This step also includes: building an internal modeling and evaluation module in the data station of the thermal power plant; the internal modeling and evaluation module is used to execute step 3.

[0102] Step 4, constructing a logistic regression model; based on weather monitoring data and historical fault record data of power transformers, using principal component analysis to improve the logistic regression model, analyzing the coupling effect of weather factors on the logistic regression model, and determining the parameter values ​​of the logistic regression model by maximizing the likelihood function; based on the parameter values ​​of the logistic regression model, establishing an external random fault probability assessment model for the meteorological area where the power transformer is located; the external random fault probability assessment model is used to calculate the external random fault probability of the power transformer.

[0103] Specifically, the method includes the following sub-steps:

[0104] Step 4.1, based on weather monitoring data and historical fault record data of power transformers, construct the model fitting sample data required for this step. The data format and construction rules of each group of samples are as follows:

[0105]

[0106] Assignment rules:

[0107] In the formula, is a sample constructed based on the weather data and operation records of power transformer n in the tth time period; DY n,t It is the output corresponding to the weather sample, which is determined by the historical fault record data and is a binary variable.

[0108] Step 4.2, after obtaining the fitting sample data, the input data in the fitting sample data is standardized to eliminate the influence of the magnitude difference of each weather index:

[0109]

[0110] Step 4.3, use principal component analysis to determine the weight of each factor, filter out variables with low weight or redundant, and retain the num cri The principal component analysis method is relatively mature, so this step will not be introduced in detail.

[0111] Step 4.4, after the normalization, cri The input data of weather factors are the dependent variables, DY is the binary output variable, and the sigmoid function is used to build a logistic regression model:

[0112]

[0113] In the formula, l i (j=0,1,…,num cri ) is the coefficient to be estimated in the logistic regression model; V(l) is the output of the logistic regression model. It should be noted that the advantage of logistic regression is that in the model fitting stage, the output variable of the model is a binary variable, DY = 1 indicates that an external random fault has occurred in the event, and DY = 0 indicates that no external random fault has occurred in the power transformer; in the fitting completion application stage, the output external random fault probability can take a continuous value between 0 and 1.

[0114] Step 4.5, based on the extracted N all The maximum likelihood estimation method is used to estimate the logistic regression model parameters. The num is determined by maximizing the likelihood function estimation function. cri The specific values ​​of the model parameters are:

[0115]

[0116] Where L(l) is the likelihood function,

[0117] Step 4.6, determine the specific values ​​of the logistic regression model parameters by maximizing the logarithmic value of the likelihood function, as follows:

[0118]

[0119] Step 4.7, after completing the above steps, establish the power transformer at t based on the logistic regression model parameter value. i External random failure probability assessment model at time:

[0120]

[0121] In the formula, is the weather monitoring data of the power transformer at time ti. It should be noted that with the increase of external weather data and fault log data of the power transformer, the parameters of the logistic regression model can be dynamically estimated adaptively to improve the accuracy of the evaluation.

[0122] This step also includes: building an external modeling and evaluation module in the data station of the thermal power plant; the external modeling and evaluation module is used to execute step 4;

[0123] Step 5: Combine the internal aging failure probability output from step 3 and the external random failure probability output from step 4 to calculate the time-varying reliability of the power transformer combined with the degradation process and the real-time state. The calculation formula is as follows:

[0124]

[0125] Where P com (t i ) is the comprehensive time-varying fault probability of the power transformer. The third negative term in the formula is used to consider that the power transformer generally does not suffer from external random faults and internal aging faults at the same time.

[0126] This step also includes: building a result output module in the data station of the thermal power plant; the result output module is used to execute step 5. Combined with the modules constructed in steps 1 to 4, this solution can also construct a system module architecture suitable for remote monitoring of time-varying reliability of power transformers in thermal power plants, which can realize dynamic evaluation of power transformer reliability based on online monitoring data.

[0127] In addition, in order to intuitively demonstrate the superiority of this solution, an example is also used here to illustrate the application effect of this solution.

[0128] Two in-service transformers T1 and T2 in the power transformer database of a power company are selected for example analysis. Based on the historical oil chromatographic monitoring data of the two transformers and regional weather data, their time-varying comprehensive fault probabilities at different status monitoring points are evaluated.

[0129] First, we take the restoration of missing oil chromatographic data as an example to verify the effectiveness of this scheme in data restoration, and compare it with the traditional KNN method and ARIMA method. The specific analysis is shown in Table 3. Figure 3 and Figure 4 .

[0130] Table 3 Oil chromatogram sample construction with missing data

[0131]

[0132] Note: “*” indicates that the gas content monitoring data is missing and needs to be restored.

[0133] from Figure 3 It can be seen that when restoring the missing samples of the four oil chromatogram monitoring data, the relative error of this scheme is controlled within 15%, while the relative error of the traditional KNN method and ARIMA method will exceed 30% in some cases, resulting in data restoration failure. This is because the ARIMA method relies on time series prediction, but the monitoring data such as transformer oil chromatogram may have mutations due to internal defects (such as short-term overheating, internal discharge); and the traditional KNN method searches for similar oil chromatogram data samples in all unclassified data samples, which is also prone to cause large errors. The above results show that compared with the traditional KNN method and ARIMA method, the missing data restoration method set in this scheme is more suitable for dealing with the missing problem of transformer monitoring data and has higher accuracy.

[0134] Figure 4 It further shows the comprehensive failure probability values ​​calculated at different monitoring points (same service time) for two power transformers with the same service time. Figure 4 It can be seen that the comprehensive failure probability of both transformers gradually increases with the increase of service time. However, due to the differences in the historical performance degradation process and the meteorological conditions of the day, there are differences in the comprehensive failure probability evaluation values ​​of the two power transformers at the same service time. At the second and third monitoring points, the comprehensive failure probability of transformer T1 at each state monitoring point is significantly higher than that of T2 in the same period. The main reason is that the internal aging failure probability of transformer T1 is higher than that of T2. This phenomenon is specifically manifested in that the daily load rate of transformer T1 is significantly higher than that of T2 during on-site operation, which leads to its performance degradation rate being significantly faster than that of T2. The comprehensive failure probability of the first monitoring point T2 is higher than that of T1, because the internal aging failure probability of both transformers is very small at this time, and the random failure probability caused by external weather is dominant at this time. The above results verify the effectiveness of the evaluation of this scheme, and prove that this scheme can reasonably evaluate the comprehensive failure probability of transformers. It is a promising method applicable to engineering practice and has a high value of promotion and application.

[0135] The present embodiment provides a transformer time-varying reliability assessment method that combines the degradation process with the real-time status, which can effectively assess the time-varying reliability of the power transformer with high accuracy and strong practicality.

[0136] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field are aware of all the common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement the scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for the ordinary technicians in the relevant field to implement this application. It should be pointed out that for the technicians in this field, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent.

Claims

1. A transformer time-varying reliability assessment method combining degradation process and real-time status, characterized in that: The following steps are involved: Step 1, collecting historical oil chromatography monitoring data of power transformers, sensor data and weather monitoring data recorded by the meteorological platform as basic data sets; Step 2: When there are missing data in the basic data set, a generalized data restoration method combining empirical classification and improved K-nearest neighbor algorithm is used to fill in the missing data in the oil chromatography monitoring data or weather monitoring data; Step 3: Based on the oil chromatographic monitoring data, combined with the Wiener degradation process and recursive filtering algorithm, the degradation equation of the historical degradation trajectory of the power transformer is constructed, namely, s i =s i-1 +μ i-1 (t i -t i-1 )+σε i ; Among them, s i It indicates that the power transformer is estimated at t according to the oil chromatography monitoring data. i The estimated value of the degenerate observation at time; μ is the drift coefficient; σ is the diffusion coefficient and σ>0, ε i The mean is 0 and the variance is t. i -t i-1 Normal distribution of Based on the degradation equation, calculate the power transformer at t i The probability density distribution function of the estimated value of the remaining life at time t is obtained from it. i Time-varying aging index HA i ; Based on the time-varying aging index HA i , calculate the power transformer at t i Internal aging failure probability at time; Step 4, constructing a logistic regression model; based on weather monitoring data and historical fault record data of power transformers, using principal component analysis to improve the logistic regression model, analyzing the coupling effect of weather factors on the logistic regression model, and determining the parameter values ​​of the logistic regression model; based on the parameter values ​​of the logistic regression model, establishing an external random fault probability assessment model for the meteorological area where the power transformer is located; the external random fault probability assessment model is used to calculate the external random fault probability of the power transformer; Step 5, combining the internal aging failure probability output from step 3 and the external random failure probability output from step 4, calculates the time-varying reliability of the power transformer combining the degradation process and the real-time status.

2. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: The sensor data includes daily ambient temperature collected by a temperature sensor and air relative humidity data collected by a humidity sensor.

3. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: In step 2, the universal data restoration method includes: firstly classifying the oil chromatography monitoring data or weather monitoring data into different historical sample sets based on empirical classification criteria; when the oil chromatography monitoring data or weather monitoring data of the ath day is missing, selecting the neighbor sample with the highest similarity to the data of the ath day from the historical sample set by improving the K nearest neighbor algorithm, and restoring the missing data according to the weighted value of the neighbor sample.

4. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 3 is characterized in that: The empirical classification standards corresponding to oil chromatography monitoring data include: Category 1, normal; Category 2, partial discharge; Category 3, internal overheating; Category 4, winding circulating current; Category 5, electrical flashover; Category 6, continuous spark.

5. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 3 is characterized in that: The empirical classification standards corresponding to weather monitoring data include: Category 1, thunderstorm; Category 2, strong wind; Category 3, heavy rain; Category 4, ice and snow; Category 5, high temperature; Category 6, cloudy.

6. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: In step 3, a random walk strategy is also used to dynamically correct the drift coefficient μ corresponding to each time interval.

7. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: The time-varying aging index HA i Refer to the following formula for calculation: Where, L mean is the average aging failure life of the same type of transformer; E(R i ) is the power transformer at t i The estimated remaining life at time R i Expectations; HA i The larger it is, the shorter the remaining life of the power transformer is, that is, the more serious the aging condition is.

8. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: In step 3, the power transformer is calculated based on the two-parameter Weibull distribution model at t i Internal aging failure probability at time: Where P aging (t i ) is the power transformer at t i The internal aging failure probability at the moment; α mean and κ mean are the shape parameter and scale parameter of the Weibull distribution, respectively.

9. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: In step 4, the parameter values ​​of the logistic regression model are determined by maximizing the likelihood function.

10. The transformer time-varying reliability assessment method combining degradation process and real-time status according to claim 1 is characterized in that: Step 1 also includes: constructing an overall historical database in a data station of the thermal power plant; the overall historical database is used to classify and store basic data sets; Step 2 also includes: building a data cleaning module in a data station of the thermal power plant; the data cleaning module is used to execute step 2; Step 3 also includes: building an internal modeling and evaluation module in a data station of the thermal power plant; the internal modeling and evaluation module is used to execute step 3; Step 4 also includes: building an external modeling and evaluation module in a data station of the thermal power plant; the external modeling and evaluation module is used to execute step 4; Step 5 also includes: building a result output module in the data station of the thermal power plant; the result output module is used to execute step 5.

Images