A transformer predictive maintenance system

Abstract

The present invention belongs to the technical field of transformer maintenance, and discloses a transformer predictive maintenance system, which includes: collecting the operation data of the transformer, preprocessing the operation data to obtain operation characteristic data; clustering the operation characteristic data to obtain n operation state clusters; constructing a state directed graph with the n operation state clusters as nodes; solving the shortest path probability from each node to all other nodes in the state directed graph; constructing constraint conditions and an optimization function based on the state directed graph and the solved shortest path probability, and solving the optimization function by using the imperialist competitive algorithm based on the constraint conditions to obtain the optimal maintenance strategy of the transformer, which can not only predict potential fault risks, but also optimize the allocation of maintenance resources, avoiding unnecessary over-maintenance or under-maintenance; improving the scientificity and economy of transformer management.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer maintenance. More specifically, the present invention relates to a transformer predictive maintenance system. Background Art

[0002] The patent with the application publication number CN106643880A discloses a transformer early warning method for an operation and maintenance system, which solves the technical problem of identifying abnormal working conditions of a transformer. According to the rated parameters of the transformer and the transformer working condition data measured by the operation and maintenance system and the transformer working condition data measured by the manual inspection method, the temperature alarm coefficient and the current difference ratio of the transformer are calculated, and then the transformer alarm value is calculated, and the working state of the transformer is judged according to the transformer alarm value, so as to timely identify the abnormal working conditions of the transformer and the abnormal data acquisition quality, and can help maintenance personnel to timely discover equipment defects.

[0003] However, traditional maintenance strategies often rely on fixed time intervals or simple thresholds to trigger, lacking a comprehensive consideration of the actual operating state of the transformer, resulting in the maintenance work may be carried out too early or too late, causing waste of resources or equipment damage; secondly, in terms of data, the processing of data affects the accuracy of feature extraction, and the existing processing methods make the evaluation of the transformer state deviate; furthermore, traditional methods cannot accurately reflect the complex operating characteristics of the transformer under different working conditions; in addition, it is difficult to accurately describe the relationship between different operating states of the transformer, resulting in inaccurate prediction of future states; finally, there is a lack of an optimization method that comprehensively considers the operating state of the transformer, maintenance costs and long-term benefits, making it difficult for maintenance decisions to achieve the best balance between costs and benefits; these problems may lead to unnecessary power outages, shortened equipment life, increased maintenance costs, and even serious power accidents in actual operation.

[0004] In view of this, the present invention proposes a transformer predictive maintenance system to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A transformer predictive maintenance system, comprising: a data acquisition and processing module, configured to acquire the operating data of the transformer, preprocess the operating data to obtain operating feature data;

[0006] a clustering module, configured to cluster the operating feature data to obtain n operating state clusters;

[0007] a preliminary fitting module, configured to construct a state directed graph with the n operating state clusters as nodes; and solve the shortest path probability from each node to all other nodes in the state directed graph;

[0008] The strategy maintenance module constructs constraints and optimization functions based on the state directed graph and the shortest path probability obtained by solving it, and solves the optimization function based on the constraints using the Empire competition algorithm to obtain the optimal maintenance strategy for the transformer; each module is connected by wire and / or wireless means.

[0009] Further, the operation data includes a temperature data sequence, a load data sequence and an insulating oil quality data sequence;

[0010] The method of preprocessing the operating data includes:

[0011] Standardizing or normalizing the operating data to obtain standard operating data, and removing trend items from the standard operating data to obtain preliminary processed data;

[0012] Decompose the preliminary processed data to obtain m intrinsic modal components and residual sequences ;

[0013] For The integrated energy of the intrinsic modal components is calculated ,

[0014] ;in, is the first intrinsic modal component The residual value, For the The weight of the residual value, is a real or complex number; is a real number parameter; is a positive real number parameter;

[0015] Calculate the The proportion of the combined energy of an intrinsic modal component to the sum of the combined energies of all intrinsic modal components ;

[0016] Preset importance threshold, if If it is greater than or equal to the importance threshold, The intrinsic modal components are retained if If the value is less than the importance threshold, The intrinsic modal components are removed, and all the retained intrinsic modal components constitute the effective component set;

[0017] The intrinsic modal components in the effective component set are added and reconstructed to obtain a new sequence ; From the new sequence Extract running characteristic data.

[0018] Furthermore, the method of performing the trend item removal operation includes:

[0019] Define the window size of the moving window , and for the window size , set an adaptive window size adjustment strategy; the setting method of the adaptive window size adjustment strategy includes:

[0020] Define the evaluation index function ;

[0021] ; where represents the high-frequency wavelet coefficient of the -th level wavelet decomposition, represents the low-frequency approximation component coefficient of the -th level wavelet decomposition, is the probability density of the -th data point , is the mean value of the data points within the current moving window, is the total number of data points within the current moving window; is the standard deviation of the data points within the current moving window; , and are the weight parameters of the corresponding terms;

[0022] Calculate the value of the evaluation index function of the data points within the current moving window, denoted as the evaluation value ; preset the adjustment high threshold and the window size reduction amount specified by the adjustment high threshold ; preset the adjustment low threshold and the window size increase amount specified by the adjustment low threshold ; ;;

[0023] If is greater than , then reduce the window size by ; if is less than , then increase the window size by ; otherwise, keep the window size unchanged;

[0024] For the time point , calculate the mean value of the data points in the sequence centered on , from to , as a small trend item sequence , repeat the calculation of the small trend item sequence for the entire sequence, and splice to obtain the moving average sequence ;

[0025] Based on the original sequence and the moving average sequence , the detrended residual sequence is calculated ; where is the weight of the original sequence , is the weight of the moving average sequence ; all the detrended residual sequences are the preliminarily processed data.

[0026] Furthermore, the method for decomposing the preliminarily processed data includes:

[0027] Initially, assign the original sequence to the residual sequence ; perform iterative processing on the residual sequence to extract an intrinsic mode component ;

[0028] The way of performing iterative processing includes: find all local maximum points and local minimum points, and fit the upper envelope through the local maximum points; fit the lower envelope

[0029] through the local minimum points; Calculate the mean of the upper envelope and the lower envelope

[0030] Remove the mean from the residual sequence ; repeat until meets the intrinsic mode constraint conditions; denote the that meets the intrinsic mode constraint conditions as the intrinsic mode component;

[0031] The intrinsic mode constraint conditions include: on , the number of local maximum points and local minimum points is equal or differs by no more than 1, and the mean of the upper envelope and the lower envelope of

[0032] tends to 0; Remove the extracted intrinsic mode component from the original residual sequence , that is, complete the update of the residual sequence , and repeat to obtain the intrinsic mode component for the updated residual sequence until the energy of

[0033] The calculation formula for energy is as follows: ; where represents the th residual value in the residual sequence.

[0034] Furthermore, the method for extracting the operation feature data from the new sequence includes:

[0035] Transform the new sequence to obtain the preliminary coefficients. The formula for the transformation is:

[0036] ; where is the preliminary coefficient, is the scale factor, is the translation factor, is the time factor exponent, is the wavelet basis function; is the time parameter;

[0037] Calculate the preliminary energy based on the preliminary coefficients, and calculate the preliminary entropy based on the preliminary energy ;

[0038] Extract along the maximum value line of the scale factor as the ridge line, and extract the slope of the wavelet ridge line in the logarithmic coordinate system as the slope; the operation feature data includes the preliminary energy, the preliminary entropy, the ridge line, and the slope.

[0039] Furthermore, the method for clustering the operation feature data includes:

[0040] Construct the extracted operation feature data into a multi-dimensional feature space, and divide the feature space into M1 rectangular cells; the division method of the rectangular cells includes:

[0041] Determine the boundary of the feature space, that is, multiply the intervals of the maximum and minimum values of each dimension of the operation feature data; regard the entire feature space as the root node, which is represented by a -dimensional hyper-rectangle. Based on the root node, divide the nodes. For each node, calculate its variance ; where is the variance factor of the th dimension; and are the maximum and minimum values of the th dimension respectively;

[0042] Preset the division threshold TH. If the variance of a node is greater than the division threshold TH, then recursively divide the node. The direction of the recursive division is to select the dimension with the largest variance​ dimension, and divide the node into two nodes at the midpoint of the interval along the th dimension; repeat until the variance of all nodes is less than the division threshold TH, that is, M1 rectangular cells are constructed.

[0043] For each rectangular cell, calculate the statistical information of the clustering points inside it. The statistical information includes the number of data points, the mean, and the variance;

[0044] Traverse each rectangular cell, and calculate the uniformity parameter according to its statistical information. The uniformity parameter is the weighted sum of the number of data points, the mean, and the variance;

[0045] Define the uniformity threshold. For the rectangular cells with the uniformity parameter greater than the uniformity threshold, regard them as the cells to be divided. When all rectangular cells no longer need to be divided after further dividing the cells to be divided, regard each corresponding rectangular cell as a running state cluster; obtain n running state clusters;

[0046] The method of the further division includes:

[0047] For the th clustering point in the cell to be divided , calculate its density. The calculation formula of the density is: , where is the total number of clustering points in the cell to be divided, is the kernel function, is the bandwidth parameter, is the th clustering point, and do not index the same clustering point at the same time, is the th clustering point 's density;

[0048] Define the initial neighborhood radius. Based on the neighborhood radius, construct a neighborhood for the th clustering point, and sum the densities of the clustering points in its neighborhood to obtain the total neighborhood density; preset the density threshold, and continuously increase the neighborhood radius. When the calculated total neighborhood density is greater than the density threshold, the neighborhood constructed by the neighborhood radius at this time is denoted as the high-density region. For each high-density region, divide it as a rectangular cell, that is, complete the further division.

[0049] Furthermore, the method of constructing the state directed graph includes:

[0050] Regard each running state cluster as a corresponding node in the state directed graph, and there are n nodes in total; for any two running state clusters And , calculate the similarity between them ;

[0051] ; Among them, is the attenuation parameter; is the mean vector of the operating state cluster , is the mean vector of the operating state cluster , is the dispersion term weight coefficient, is the dispersion of the operating state cluster , is the dispersion of the operating state cluster ; is the and Euclidean distance between;

[0052] ; Among them, is the weight adjustment parameter; is the total number of clustering points within the operating state cluster , is the density, is the clustering point within the operating state cluster , is the dispersion weight adjustment parameter, is the variance of the internal clustering points of the operating state cluster ; The calculation method of is the same as;

[0053] Preset similarity threshold , if , then add a directed edge in the state directed graph, and set the weight of the directed edge to be , that is, complete the construction of the state directed graph.

[0054] Furthermore, the solution method of the shortest path probability includes:

[0055] Create an empty minimum spanning tree MST, add any one node to the minimum spanning tree MST, for the remaining nodes, calculate their shortest distances to the nodes in the minimum spanning tree MST, and record them;

[0056] Select the node closest to the minimum spanning tree MST from the remaining nodes, add it to the minimum spanning tree MST, and update the shortest distances of the remaining nodes to the minimum spanning tree MST, repeat until all nodes are added to the minimum spanning tree MST;

[0057] For each pair of nodes , find their nearest common ancestor node in the minimum spanning tree MST , then the shortest path length from node to node ; where ; represents the shortest path length from node to node in the minimum spanning tree MST, represents the shortest path length from node to node in the minimum spanning tree MST;

[0058] Based on the shortest path length calculate the shortest path probability from node to node , where ; is the adjustment factor; is the path attenuation parameter, is the smoothing parameter.

[0059] Furthermore, the formula of the optimization function is:

[0060] ; where is a corresponding recommended maintenance strategy determined by node , is a corresponding recommended maintenance strategy determined by node , is the cost required to execute the recommended maintenance strategy , is the cost required to execute the recommended maintenance strategy ; is the time to actually run and transfer from node to node ; is the probability of transferring from node to node ; is the running loss;

[0061] The constraint conditions include the first constraint condition and the second constraint condition; the first constraint condition is that the out-degree probability of each node is 1, and the in-degree probability of each node is 1; the second constraint condition is that the probability of transferring from any node to other nodes is non-negative, and the shortest path probability is used as the upper limit of the probability of transferring from any node to other nodes.

[0062] ​​Furthermore, the method for solving the optimization function includes:

[0063] Initialize a population, where each individual in the population represents a solution vector, that is, the probabilities of transfer between nodes; randomly generate an initial population that satisfies the constraint conditions;

[0064] For each individual, calculate the value of its optimization function, denoted as the allocation degree; define an empire based on the allocation degree of the individual, select the individual with the highest allocation degree as the emperor of the first empire, and calculate the difference between the allocation degrees of other individuals and this emperor. Arrange the other individuals in ascending order of the difference in allocation degrees and assign them to this empire in sequence until the power of this empire reaches the preset maximum power value; repeat until all individuals are assigned to different empires;

[0065] Define a weak threshold. Inside each empire, regard the individuals with an allocation degree less than the weak threshold as weak countries, and perform an assimilation operation on the weak countries to make them gradually approach the emperor of the empire. The formula for the assimilation operation is:

[0066] ; where is the new position of the weak country, is the current position of the weak country, is the position of the emperor of the empire, is a random number within the range of [0, 1], is a uniform random number within the range of [0.1, 0.9], is 's allocation degree, is 's allocation degree, is the difference adjustment parameter, is a positive number; is a random number that follows a standard normal distribution;

[0067] Define a boundary distance threshold, calculate the distances between all individuals inside the empire and the emperor, and mark the individuals with distances greater than the boundary distance threshold as being at the empire boundary;

[0068] For the individuals marked as being at the empire boundary, generate a random number within the range of [0, 1] according to the predefined revolution rate; if this random number is less than the revolution rate, re-randomly initialize the position of this individual;

[0069] After performing the assimilation operation and revolution, calculate the sum of the distribution degrees of all individuals within the empire as the power of the empire. Then, in descending order of the power of the empire, merge the smallest empire into other empires; repeat until the preset maximum number of iterations is reached, and obtain the individual with the highest distribution degree among all empires at this time as the optimal solution. Based on the optimal solution, obtain the transition probability matrix between each node;

[0070] Based on the obtained transition probability matrix and combined with the maintenance cost corresponding to each node, calculate the long-term total maintenance cost; enumerate different combinations of proposed maintenance strategies and find the combination of proposed maintenance strategies that can minimize the long-term total maintenance cost, which is the optimal maintenance strategy.

[0071] The technical effects and advantages of a predictive maintenance system for transformers according to the present invention:

[0072] The present invention can effectively remove the noise and trend terms of the original data, improve the accuracy of feature extraction, and use a method combining recursive partitioning and density clustering to achieve a fine division of the operating state of the transformer. By constructing a state directed graph and calculating the shortest path probability, the transition relationship between different operating states is accurately characterized; thus, comprehensively analyzing the actual operating state of the transformer, realizing the intelligence and refinement of the maintenance strategy, greatly improving the pertinence and efficiency of maintenance; effectively reducing the long-term maintenance cost, extending the service life of the transformer, reducing the occurrence of unexpected failures, thereby improving the overall reliability and stability of the power system; in addition, its adaptive characteristics enable it to automatically adjust the maintenance strategy according to changes in the operating environment of the transformer to ensure that the optimal maintenance effect is always maintained; it can not only predict potential failure risks, but also optimize the allocation of maintenance resources to avoid unnecessary over-maintenance or under-maintenance; improving the scientificity and economy of transformer management. Brief Description of the Drawings

[0073] Figure 1 It is a schematic diagram of a predictive maintenance system for transformers according to the present invention;

[0074] Figure 2 It is a schematic diagram of a predictive maintenance method for transformers according to the present invention. Detailed Embodiments

[0075] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0076] Embodiment 1:

[0077] Please refer to Figure 1 As shown, a predictive maintenance system for a transformer in this embodiment includes:

[0078] A data acquisition and processing module, which is used to acquire the operation data of the transformer, preprocess the operation data, and obtain operation characteristic data;

[0079] A clustering module, which is used to cluster the operation characteristic data to obtain n operation state clusters;

[0080] A preliminary fitting module, which is used to construct a state directed graph with n operation state clusters as nodes; solve the shortest path probability from each node to all other nodes in the state directed graph;

[0081] A strategy maintenance module, which constructs constraint conditions and an optimization function based on the state directed graph and the obtained shortest path probability, and uses the imperialist competitive algorithm to solve the optimization function based on the constraint conditions to obtain the optimal maintenance strategy of the transformer; each module is connected by wired and / or wireless means to realize data transmission between modules.

[0082] Furthermore, the operation data includes a temperature data sequence, a load data sequence, and an insulating oil quality data sequence; temperature sensors are usually installed inside the transformer, and the temperature sensors sample and record the temperature values at preset time intervals to form a temperature data sequence. Current transformers and voltage transformers are used to detect current and voltage, calculate the real-time load of the transformer according to the current and voltage, and sample and record the real-time load at preset time intervals to form a load data sequence.

[0083] Online monitoring sensors are installed in the insulating oil storage tank or oil circulation system of the transformer. These sensors detect the dielectric strength, acid value, and particle content of the insulating oil in real time, perform weighted calculation on the dielectric strength, acid value, and particle content to obtain insulating oil quality data, and sample and record the insulating oil quality data at preset time intervals to form an insulating oil quality data sequence.

[0084] Furthermore, the method for preprocessing the operation data includes:

[0085] Normalize or standardize the operation data to make it have a zero mean to obtain standard operation data, and perform a trend term removal operation on the standard operation data to obtain preliminary processed data.

[0086] Specifically, define the window size of the moving window (initially set to 5%-25% of the sequence length), and for the window size , set an adaptive window size adjustment strategy; the setting method of the adaptive window size adjustment strategy includes:

[0087] Define the evaluation index function ;

[0088] ; where represents the high-frequency wavelet coefficient of the -th level wavelet decomposition (the data within the current moving window is wavelet decomposed to obtain wavelet coefficients of different scales), represents the low-frequency approximation component coefficient of the -th level wavelet decomposition, is the probability density of the -th data point , is the mean value of the data points within the current moving window, is the total number of data points within the current moving window; is the standard deviation of the data points within the current moving window; , and are the weight parameters of the corresponding items, and the weight parameters can be adaptively adjusted by methods such as Bayesian models and online learning.

[0089] It should be noted that the data value at each time point in the temperature data sequence, load data sequence, and insulating oil quality data sequence is used as a data point.

[0090] Calculate the value of the evaluation index function for the data points within the current moving window, denoted as the evaluation value ; Preset the adjustment high threshold and the window size reduction specified by the adjustment high threshold ; Preset the adjustment low threshold and the window size increase specified by the adjustment low threshold .

[0091] If is greater than , then reduce the window size by ; If is less than , then increase the window size by ; Otherwise, keep the window size unchanged.

[0092] For the time point , calculate the mean value of the data points within the sequence from to with as the center, and use it as a small trend item sequence , repeatedly calculate the trend item small sequence for the entire sequence (temperature data sequence, load data sequence and insulating oil quality data sequence), and splice them to obtain the moving average sequence (trend item) ; It should be explained that the calculations are all for the corresponding sequences, for example, the temperature data sequence is calculated for itself.

[0093] Based on the original sequence and the moving average series , calculate the residual sequence after detrending ;in, For the original sequence The weight of is a moving average series The weight of .

[0094] All the residual series after detrending are the preliminary processed data;

[0095] The residual series after detrending (preliminary processing data) Decompose to obtain m intrinsic modal components and residual sequences .

[0096] Specifically, initially, the original sequence Assign to the residual sequence ; For the residual sequence Perform iterative processing to extract an intrinsic modal component .

[0097] The iterative processing method includes: finding All local maximum points and local minimum points are used to fit the upper envelope through the local maximum points. ; Fit the lower envelope through the local minimum point Specifically, a cubic spline curve is used for interpolation fitting to obtain an upper envelope and a lower envelope.

[0098] Calculate the upper envelope and lower envelope The mean ;

[0099] From the residual sequence Remove the mean ; Repeat until Satisfy the intrinsic modal constraints; are denoted as intrinsic modal components.

[0100] The intrinsic modal constraints include: The number of local maximum points and local minimum points is equal or the difference does not exceed 1, and The mean of the upper and lower envelopes of is close to 0;

[0101] The extracted intrinsic modal components are extracted from the original residual sequence Remove it, that is, complete the residual sequence The updated residual sequence Repeatedly obtain the intrinsic mode components (also part of the residual sequence) until Energy Less than the preset energy threshold.

[0102] The energy calculation formula is: ;in, is the first Residual values.

[0103] Through the above iterative process, the original sequence is decomposed into m intrinsic modal components and the remaining residual sequence.

[0104] For The integrated energy of the intrinsic modal components is calculated ,

[0105] ; in, is the first intrinsic modal component The residual value, For the The weight of the residual value, is a real number or a complex number, thereby introducing a phase factor, making the energy calculation result have vector characteristics, which can reflect the oscillation and fluctuation characteristics of the residual; is a real number parameter used to control the curvature of the logarithmic function; is a positive real number parameter that controls the decay rate.

[0106] Calculate the The proportion of the combined energy of an intrinsic modal component to the sum of the combined energies of all intrinsic modal components .

[0107] Preset importance threshold, if If it is greater than or equal to the importance threshold, Intrinsic modal components are retained if If the value is less than the importance threshold, The intrinsic modal components are removed, and all the retained intrinsic modal components constitute the effective component set.

[0108] The intrinsic modal components in the effective component set are added and reconstructed to obtain a new sequence ; New sequence retains the main feature components of the original sequence while removing some noise and irrelevant components.

[0109] Extract running feature data from the new sequence Specifically, transform the new sequence to obtain preliminary coefficients. The formula for the transformation is as follows:

[0110] ; where is the preliminary coefficient, is the scale factor, is the translation factor, is the time factor exponent used to change the influence of time translation, is the wavelet basis function (such as Haar wavelet, Coiflet wavelet, etc.); is a time parameter that controls the broadening degree of the function on the time axis.

[0111] Calculate the preliminary energy based on the preliminary coefficients , and calculate the preliminary entropy based on the preliminary energy .

[0112] Extract the maximum value line along the scale factor as the ridge line, and extract the slope of the wavelet ridge line in the logarithmic coordinate as the slope; the running feature data includes the preliminary energy, the preliminary entropy, the ridge line, and the slope.

[0113] Furthermore, the methods for clustering the running feature data include:

[0114] Construct a multi-dimensional feature space from the extracted running feature data, divide the feature space into M1 rectangular cells. Specifically, determine the boundaries of the feature space, that is, multiply the intervals of the maximum and minimum values of each dimension (preliminary energy, preliminary entropy, ridge line, and slope) of the running feature data; regard the entire feature space as the root node, represented by a dimensional hyper-rectangle, divide the nodes based on the root node. For each node, calculate its variance ; where is the variance factor of the th dimension, generally taking a value of ; and are the maximum and minimum values of the th dimension respectively.

[0115] ​Preset a division threshold TH. If the variance of a node is greater than the division threshold TH, then recursively divide the node. The direction of recursive division is to select the dimension with the largest variance as the th dimension, and divide the node into two nodes at the midpoint of the interval along the th dimension; repeat until the variances of all nodes are less than the division threshold TH, that is, M1 rectangular cells are constructed. The root node corresponds to the entire feature space, and other nodes correspond to rectangular cells.

[0116] For each rectangular cell, calculate the statistical information of the clustering points inside it. The statistical information includes the number of data points, the mean, and the variance.

[0117] Each clustering point corresponds to a twelve-dimensional vector, which is a combination of the vector representations of the four operating characteristic data of the preliminary energy, preliminary entropy, ridge line, and slope corresponding to the temperature data sequence, load data sequence, and insulating oil quality data sequence respectively.

[0118] Traverse each rectangular cell, and calculate the uniformity parameter according to its statistical information. The uniformity parameter is the weighted sum of the number of data points, the mean, and the variance.

[0119] Define a uniformity threshold. Take the rectangular cells with a uniformity parameter greater than the uniformity threshold as the cells to be divided. When the cells to be divided are further divided until all rectangular cells no longer need to be divided, each rectangular cell corresponds to an operating state cluster; obtain n operating state clusters; each operating state cluster represents an operating state of the transformer.

[0120] The method of the further division includes:

[0121] For the th clustering point in the cell to be divided , calculate its density. The calculation formula of the density is: , where is the total number of clustering points in the cell to be divided, is the kernel function (such as Gaussian kernel), is the bandwidth parameter, is the th clustering point, and do not index the same clustering point at the same time, is the th clustering point 's density.

[0122] Define the initial neighborhood radius, and for the For each clustering point, construct a neighborhood based on the neighborhood radius, sum up the densities of the clustering points within its neighborhood to obtain the total neighborhood density; preset a density threshold, and continuously increase the neighborhood radius. When the calculated total neighborhood density is greater than the density threshold, the neighborhood constructed with the neighborhood radius at this time is denoted as a high-density area. For each high-density area, divide it as a rectangular cell, that is, the further division is completed.

[0123] Further, the method for constructing the state directed graph includes:

[0124] Take each operation state cluster as a corresponding node in the state directed graph, with a total of n nodes (each corresponding to an operation state of the transformer); for any two operation state clusters and , calculate the similarity between them .

[0125] ; where, is the attenuation parameter for controlling the similarity attenuation rate, taking a positive value; is the mean vector of the operation state cluster (the mean of all clustering points within it), is the mean vector of the operation state cluster , is the dispersion term weight coefficient, used to adjust the influence of data dispersion degree on the similarity, is the dispersion degree of the operation state cluster , is the dispersion degree of the operation state cluster ; is and the Euclidean distance between.

[0126] ; where, is the weight adjustment parameter, with a value in the range of [0, 1]; is the total number of clustering points within the operation state cluster , is 's density (the calculation method is the same as the density calculation method of ), is the clustering point within the operation state cluster , is the dispersion weight adjustment parameter, is the variance of the internal clustering points of the operation state cluster ; 's calculation method is the same as .

[0127] Preset similarity threshold If , then add a directed edge in the state directed graph , and set the weight of the directed edge to be .

[0128] Create an empty minimum spanning tree MST, add any node (such as node A) to the minimum spanning tree MST. For the remaining nodes, calculate their shortest distances to the nodes in the minimum spanning tree MST (obtained by adding the weights of the directed edges to the Euclidean distance calculation), and record them.

[0129] Select the node closest to the minimum spanning tree MST from the remaining nodes (such as node B), add it to the minimum spanning tree MST, and update the shortest distances of the remaining nodes to the minimum spanning tree MST. Repeat until all nodes are added to the minimum spanning tree MST.

[0130] For each pair of nodes , find their nearest common ancestor node in the minimum spanning tree MST , then the shortest path length from node to node is ; where represents the shortest path length from node to node in the minimum spanning tree MST, represents the shortest path length from node to node in the minimum spanning tree MST.

[0131] Based on the shortest path length calculate the shortest path probability from node to node , where is the adjustment factor, used to control the overall size of the probability; usually set to a positive value less than 1, such as 0.9; is the path attenuation parameter, used to control the influence degree of the path length on the probability, the larger it is, the greater the influence of the path length on the probability, and the faster the probability value decreases; is the smoothing parameter, used to smooth the change of the probability value.

[0132] Furthermore, the formula of the optimization function is:

[0133] ; where is node Determine a corresponding recommended maintenance strategy (including the maintenance operations to be performed in the corresponding state and their costs), for the node determine a corresponding recommended maintenance strategy, for the cost required to execute the recommended maintenance strategy ; for the cost required to execute the recommended maintenance strategy ; for the time to transfer from the actual operation of the node to the node ; for the probability of transferring from the node to the node ; for the operation loss (transferring to the node after operating for a period of time at the node may result in some operation losses, such as equipment aging, efficiency decline, etc.). The operation loss includes losses caused by temperature, losses caused by load, and losses caused by the quality of insulating oil.

[0134] The loss caused by temperature is estimated using the Arrhenius model, the loss caused by load is estimated using the exponential model, and the loss caused by the quality of insulating oil is estimated using the linear model; for example, the loss caused by load ; where and are empirical constants, is the average load within the operation time at the node , is the rated load of the transformer.

[0135] The said constraint conditions include the first constraint condition and the second constraint condition; the first constraint condition is that the out-degree probability of each node (the sum of the probabilities of transferring from the node to all other nodes) is 1, and the in-degree probability of each node (the sum of the probabilities of transferring from all other nodes to the node) is 1; the second constraint condition is that the probability of transferring from any node to other nodes is non-negative, and the shortest path probability is used as the upper limit of the probability of transferring from any node to other nodes.

[0136] The methods for solving the optimization function include:

[0137] Initialize a population, and each individual in the population represents a solution vector, that is, the probabilities of transfer between each node; randomly generate an initial population that satisfies the constraint conditions.

[0138] For each individual (solution vector), calculate the value of its optimization function, denoted as the allocation degree; define an empire based on the allocation degree of the individual, select the individual with the highest allocation degree as the emperor of the first empire, and calculate the difference between the allocation degrees of other individuals and this emperor. Arrange the other individuals in ascending order of the difference in allocation degrees and assign them to this empire one by one until the power of the empire (the sum of the allocation degrees of all individuals within the empire) reaches the preset maximum power value; repeat until all individuals are assigned to different empires.

[0139] Define a weak threshold. Within each empire, consider the individuals with an allocation degree less than the weak threshold as weak countries and perform an assimilation operation on them to make them gradually approach the center (emperor) of the empire. The formula for the assimilation operation is:

[0140] ; where, is the new position (solution vector) of the weak country, is the current position (solution vector) of the weak country, is the position of the emperor of the empire, is a random number within the interval [0, 1], is a uniform random number within the interval [0.1, 0.9] used to control the degree of approach, is 's allocation degree, is 's allocation degree, is a difference adjustment parameter used to control the influence of fitness difference on the degree of approach, is a small positive number used to control the intensity of random perturbation; is a random number following a standard normal distribution used to introduce random perturbation; in this way, the weak country will gradually approach the emperor but will not completely coincide.

[0141] Define a boundary distance threshold, calculate the distance (Euclidean distance) between all individuals within the empire and the emperor, and mark the individuals with a distance greater than the boundary distance threshold as being on the empire's boundary; for the individuals (countries) marked as being on the empire's boundary, give them the opportunity to revolutionize, that is, randomly re-initialize the position of this individual (country).

[0142] For each border country, generate a random number within the interval [0, 1] according to the predefined revolution rate (such as 0.1); if this random number is less than the revolution rate, then randomly re-initialize the position (solution vector) of this individual (country).

[0143] After performing the assimilation operation and the revolution, calculate the sum of the distribution degrees of all individuals (countries) within the empire as the power of the empire. Arrange the empires in descending order of their power and merge the smallest empire into other empires; repeat this process until the preset maximum number of iterations is reached, and obtain the individual with the highest distribution degree among all empires at this time as the optimal solution. Based on the optimal solution, obtain the transition probability matrix between each node (operating state cluster), and this transition probability matrix represents the probability of the transformer transitioning from the current operating state to other operating states.

[0144] For each node (operating state cluster), determine a recommended maintenance strategy based on the characteristics of its corresponding operating state and in combination with historical data. Substitute the recommended maintenance strategy corresponding to each node into the maintenance cost in the optimization function to calculate the cost required to execute the recommended maintenance strategy at this node.

[0145] Based on the obtained transition probability matrix and in combination with the maintenance cost corresponding to each node, calculate the long-term total maintenance cost; enumerate different combinations of recommended maintenance strategies to find the combination of recommended maintenance strategies that can minimize the long-term total maintenance cost, which is the optimal maintenance strategy.

[0146] The optimal maintenance strategy includes which specific maintenance operations should be performed and the corresponding maintenance schedule under each possible operating state.

[0147] This embodiment can effectively remove the noise and trend terms in the original data, improve the accuracy of feature extraction, use a method combining recursive partitioning and density clustering to achieve a fine division of the transformer operating state, and accurately describe the transition relationship between different operating states by constructing a state directed graph and calculating the shortest path probability; thus comprehensively analyzing the actual operating state of the transformer, realizing the intelligence and refinement of the maintenance strategy, greatly improving the pertinence and efficiency of maintenance; can effectively reduce the long-term maintenance cost, extend the service life of the transformer, reduce the occurrence of unexpected failures, thereby improving the overall reliability and stability of the power system; in addition, its adaptive characteristics enable it to automatically adjust the maintenance strategy according to changes in the transformer operating environment to ensure that the optimal maintenance effect is always maintained; it can not only predict potential failure risks, but also optimize the allocation of maintenance resources to avoid unnecessary over-maintenance or under-maintenance situations; improving the scientificity and economy of transformer management.

[0148] Embodiment 2:

[0149] Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. Provide a transformer predictive maintenance method, including:

[0150] S1. Collect the operation data of the transformer, preprocess the operation data, and obtain the operation characteristic data;

[0151] S2. Cluster the operation characteristic data to obtain n operation state clusters;

[0152] S3. Construct a state directed graph with the n operation state clusters as nodes; Solve the shortest path probability from each node to all other nodes in the state directed graph;

[0153] S4. Construct constraint conditions and an optimization function based on the state directed graph and the obtained shortest path probability, and use the imperialist competitive algorithm to solve the optimization function based on the constraint conditions to obtain the optimal maintenance strategy of the transformer.

[0154] Embodiment 3:

[0155] This embodiment publicly provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the above-provided method for predictive maintenance of a transformer.

[0156] Since the electronic device introduced in this embodiment is the electronic device used to implement the method for predictive maintenance of a transformer in the embodiments of the present application, based on the method for predictive maintenance of a transformer introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the method for predictive maintenance of a transformer in the embodiments of the present application, it falls within the scope of protection of the present application.

[0157] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0158] The above is only the preferred embodiment of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A transformer predictive maintenance system, characterized in that: include: The data acquisition and processing module is used to collect the operation data of the transformer, pre-process the operation data, and obtain the operation characteristic data; A clustering module is used to cluster the operation characteristic data to obtain n operation status clusters; A preliminary fitting module is used to construct a state directed graph using n running state clusters as nodes; Solve the shortest path probability from each node to all other nodes in the state directed graph; The strategy maintenance module constructs constraint conditions and optimization functions based on the state directed graph and the shortest path probability obtained by solving the problem, and solves the optimization function based on the constraint conditions using the Empire competition algorithm to obtain the optimal maintenance strategy for the transformer; each module is connected by wire and / or wireless means; The operation data includes a temperature data sequence, a load data sequence and an insulating oil quality data sequence; The method of preprocessing the operating data includes: Standardizing or normalizing the operating data to obtain standard operating data, and removing trend items from the standard operating data to obtain preliminary processed data; Decompose the preliminary processed data to obtain m intrinsic modal components and residual sequences ; For The integrated energy of the intrinsic modal components is calculated , ;in, is the first intrinsic modal component The residual value, For the The weight of the residual value, is a real or complex number; is a real number parameter; is a positive real number parameter; Calculate the The proportion of the combined energy of an intrinsic modal component to the sum of the combined energies of all intrinsic modal components ; Preset importance threshold, if If it is greater than or equal to the importance threshold, Intrinsic modal components are retained if If the value is less than the importance threshold, The intrinsic modal components are removed, and all the retained intrinsic modal components constitute the effective component set; The intrinsic modal components in the effective component set are added and reconstructed to obtain a new sequence ; From the new sequence Extracting operation characteristic data; The method for solving the shortest path probability includes: Create an empty minimum spanning tree MST, add any node to the minimum spanning tree MST, and for the remaining nodes, calculate their shortest distances to the nodes in the minimum spanning tree MST and record them; From the remaining nodes, select the node closest to the minimum spanning tree MST, add it to the minimum spanning tree MST, and update the shortest distance from the remaining nodes to the minimum spanning tree MST, repeat until all nodes are added to the minimum spanning tree MST; For each pair of nodes , find their nearest common ancestor node in the minimum spanning tree MST , then the node To Node The shortest path length ;in, Indicates that in the minimum spanning tree MST, from node To Node The shortest path length, Indicates that in the minimum spanning tree MST, from node To Node The shortest path length; Based on the shortest path length Calculate the node To Node The shortest path probability ,in, is the regulating factor; is the path attenuation parameter, is the smoothing parameter; The method for solving the optimization function includes: Initialize a population, where each individual in the population represents a solution vector, i.e. the probability of transfer between nodes; randomly generate an initial population that satisfies the constraints; For each individual, calculate the value of its optimization function, recorded as the allocation degree; define the empire according to the individual's allocation degree, select the individual with the highest allocation degree as the emperor of the first empire, and calculate the difference between the allocation degrees of other individuals and the emperor. In the order of the difference in allocation degrees from small to large, allocate other individuals to the empire in turn until the power of the empire reaches the preset maximum power; repeat until all individuals are allocated to different empires; Define the weak threshold. Within each empire, treat individuals with a distribution degree less than the weak threshold as weak countries. Perform assimilation operations on weak countries to gradually make them closer to the emperor of the empire. The formula for assimilation operations is: ;in, A new position for small and weak countries, For the current position of weak countries, For the position of the emperor of the empire, is a random number in the interval [0, 1]. is a uniform random number in the interval [0.1, 0.9], for The distribution degree, for The distribution degree, is the difference adjustment parameter, is a positive number; is a random number that follows a standard normal distribution; Define a border distance threshold, calculate the distance between all individuals in the empire and the emperor, and mark individuals whose distance is greater than the border distance threshold as being at the border of the empire; For individuals marked as being at the border of the empire, a random number in the interval [0, 1] is generated according to the predefined revolution rate; if the random number is less than the revolution rate, the position of the individual is randomly initialized again; After the assimilation operation and the revolution, the sum of the distribution degrees of all individuals in the empire is calculated as the strength of the empire. In order of the strength of the empires from large to small, the smallest empire is merged into other empires; repeat until the preset maximum number of iterations is reached, and the individual with the highest distribution degree in all empires at this time is obtained as the optimal solution. Based on the optimal solution, the transition probability matrix between each node is obtained; Based on the obtained transition probability matrix, combined with the maintenance cost corresponding to each node, the long-term total maintenance cost is calculated; different combinations of recommended maintenance strategies are enumerated, and the recommended maintenance strategy combination that can minimize the long-term total maintenance cost is found, which is the optimal maintenance strategy; The method of removing the trend item includes: Defines the window size of the moving window , and for the window size , set the adaptive window size adjustment policy; the setting methods of the adaptive window size adjustment policy include: The data value at each time point in the temperature data series, the load data series, and the insulating oil quality data series is taken as a data point; Define the evaluation indicator function ; ;in, Indicates The high-frequency wavelet coefficients of the first-order wavelet decomposition, Indicates The low-frequency approximate component coefficients of the first-order wavelet decomposition, For the Data points The probability density of is the mean of the data points in the current moving window, is the total number of data points in the current moving window; is the standard deviation of the data points in the current moving window; , and is the weight parameter of the corresponding item; Calculate the value of the evaluation index function of the data point in the current moving window, recorded as the evaluation value ; Preset adjustment high threshold and adjust the high threshold Decrease the window size by the specified amount ; Preset adjustment low threshold and adjust the low threshold Increase the window size by the specified amount ; like Greater than , then the window size Reduce ;like Less than , then the window size Increase ; Otherwise, keep the window size constant; For time point , calculated with Centered on arrive , take the mean of the data points in the sequence as a small sequence of trend items , repeatedly calculate the trend item small sequence for the entire sequence, and splice it to get the moving average sequence ; Based on the original sequence and the moving average series , calculate the residual sequence after detrending ;in, For the original sequence The weight of is a moving average series weights; all residual sequences after detrending are the preliminary processed data.

2. A transformer predictive maintenance system according to claim 1, characterized in that: The method of decomposing the preliminary processed data includes: Initially, the original sequence Assign to the residual sequence ; For the residual sequence Perform iterative processing to extract an intrinsic modal component ; The iterative processing method includes: finding All local maximum points and local minimum points are used to fit the upper envelope through the local maximum points. ; Fit the lower envelope through the local minimum point ; Calculate the upper envelope and lower envelope The mean ; From the residual sequence Remove the mean ; Repeat until Satisfy the intrinsic modal constraints; Denoted as intrinsic modal component; The intrinsic modal constraints include: The number of local maximum points and local minimum points is equal or the difference does not exceed 1, and The mean of the upper and lower envelopes of is close to 0; The extracted intrinsic modal components are extracted from the original residual sequence Remove it, that is, complete the residual sequence The updated residual sequence Repeat the acquisition of intrinsic modal components until Energy Less than a preset energy threshold; The energy calculation formula is: ;in, is the first Residual values.

3. A transformer predictive maintenance system according to claim 2, characterized in that: The new sequence The methods for extracting running feature data include: will new sequence Transform and obtain preliminary coefficients. The transformation formula is: ;in, is the preliminary coefficient, is the scale factor, is the translation factor, is the time factor index, is the wavelet basis function; is the time parameter; Calculate the preliminary energy based on the preliminary coefficients , based on the initial energy Calculate the initial entropy ; extract Along scale factor The maximum value line of the wavelet is taken as the ridge line, and the slope of the wavelet ridge line in the logarithmic coordinate is extracted as the slope; the running characteristic data include preliminary energy, preliminary entropy, ridge line and slope.

4. A transformer predictive maintenance system according to claim 3, characterized in that: The method of clustering the running characteristic data includes: The extracted operation feature data is constructed into a multi-dimensional feature space, and the feature space is divided into M1 rectangular cells; the rectangular cells are divided in the following ways: Determine the boundary of the feature space, that is, multiply the interval of the maximum and minimum values ​​of each dimension of the running feature data; take the entire feature space as the root node, and use a dimensional hyperrectangular representation, divide the nodes based on the root node, and for each node, calculate its variance ;in, For the The variance factor of each dimension; and Respectively The maximum and minimum values ​​of the dimensions; The partition threshold TH is preset. If the variance of a node is greater than the partition threshold TH, the node is recursively partitioned. The direction of the recursive partition is the node with the largest variance. dimension, move the node along the The midpoint of the interval of the dimension is divided into two nodes; repeat until the variance of all nodes is less than the division threshold TH, that is, M1 rectangular cells are constructed; For each rectangular cell, calculate the statistical information of the cluster points inside it, including the number, mean and variance of data points; Traverse each rectangular cell and calculate the uniform parameter based on its statistical information. The uniform parameter is the weighted sum of the number, mean and variance of data points. Define a uniform threshold, take the rectangular cells whose uniform parameters are greater than the uniform threshold as cells to be segmented, further divide the cells to be segmented until all rectangular cells do not need to be further divided, and take each rectangular cell as a running state cluster; obtain n running state clusters; The further division method includes: For the first Cluster points , calculate its density, the density calculation formula is: ,in, is the total number of cluster points in the cell to be segmented, is the kernel function, is the bandwidth parameter, For the Cluster points, and Do not index the same cluster point at the same time. For the Cluster points density; Define the initial neighborhood radius and A cluster point builds a neighborhood based on the neighborhood radius, and the density of the cluster points in its neighborhood is summed to obtain the total density of the neighborhood; a density threshold is preset, and the neighborhood radius is continuously increased. When the calculated total neighborhood density is greater than the density threshold, the neighborhood constructed by the neighborhood radius at this time is recorded as a high-density area. For each high-density area, it is segmented out as a rectangular cell to complete further division.

5. A transformer predictive maintenance system according to claim 4, characterized in that: The method of constructing the state directed graph includes: Each running state cluster is regarded as a corresponding node in the state directed graph, with a total of n nodes; for any two running state clusters and , calculate the similarity between them ; ;in, is the attenuation parameter; Running state cluster The mean vector of Running state cluster The mean vector of is the weight coefficient of the dispersion term, Running state cluster The dispersion of Running state cluster The dispersion of for and The Euclidean distance between ;in, is the weight adjustment parameter; Running state cluster The total number of points in the cluster, for The density of Running state cluster The cluster points within To disperse the weight adjustment parameters, Running state cluster The variance of internal cluster points; The calculation method and same; Preset similarity threshold ,like , then add a directed edge to the state directed graph , and set the directed edge The weight of , that is, the construction of the state directed graph is completed.

6. A transformer predictive maintenance system according to claim 5, characterized in that: The formula of the optimization function is: ;in, For Node Determine a corresponding recommended maintenance strategy, For Node Determine a corresponding recommended maintenance strategy, Recommend maintenance strategies for execution The cost required, Recommend maintenance strategies for execution the costs required; For slave nodes Actual operation is transferred to the node time; For slave nodes Transfer to Node The probability of For operating losses; The constraints include a first constraint and a second constraint; the first constraint is that the out-degree probability of each node is 1, and the in-degree probability of each node is 1; the second constraint is that the probability of transferring from any node to other nodes is non-negative, and the shortest path probability is used as the upper limit of the probability of any node transferring to other nodes.

Images