A method, system, and equipment for maintaining power supply and transformation equipment based on parallel learning

Abstract

This application discloses a method, system, and equipment for maintaining power supply and transformation equipment based on parallel learning. The method includes: collecting real-time operating data of the power supply and transformation equipment and preprocessing the data; processing the preprocessed operating data using a fuzzy C-means clustering algorithm to divide the data into normal operating states and abnormal fault states; constructing a parallel learning model, training the parallel learning model based on a Markov decision process, inputting the operating data corresponding to the abnormal fault states into the parallel learning model, obtaining prediction results of the power supply and transformation equipment failure probability and remaining service life, and formulating maintenance decisions for the power supply and transformation equipment. This application can quickly obtain the failure probability and remaining service life of power supply and transformation equipment and specify maintenance decisions, reducing the pressure of maintenance and repair work.

Description

Technical Field

[0001] This application relates to the field of power equipment health management technology, and more specifically, to a method, system, and equipment for the maintenance of power supply and transformation equipment based on parallel learning. Background Technology

[0002] With the increasing level of industrialization and automation in China, and the continuous growth of industrial scale, the maintenance and repair work of key power supply and substation equipment has been under tremendous pressure in terms of both reliability and workload. In order to transform the current "passive" maintenance of power supply and substation equipment into "proactive" maintenance, it is necessary to establish a set of "proactive" maintenance methods for key power supply and substation equipment, thereby avoiding the reliability risks associated with scheduled maintenance and condition-based maintenance strategies.

[0003] Traditional condition-based maintenance strategies suffer from complex degradation models and poor versatility. However, with the rapid development of big data and artificial intelligence technologies, and the increasing adoption of Industrial Internet of Things (IIoT) technologies, predictive maintenance methods based on big data and artificial intelligence offer an effective solution to these problems. Summary of the Invention

[0004] To address at least one deficiency or improvement need in the prior art, this invention provides a method, system, and device for maintaining power supply and transformation equipment based on parallel learning. This method can quickly obtain the failure probability and remaining service life of power supply and transformation equipment and make maintenance decisions, thereby reducing the pressure of maintenance and repair work.

[0005] To achieve the above objectives, according to a first aspect of the present invention, a method for maintaining power supply and transformation equipment based on parallel learning is provided, the method comprising:

[0006] Collect real-time operating data of power supply and transformation equipment, and perform data preprocessing on the operating data;

[0007] The preprocessed running data is processed based on the fuzzy C-means clustering algorithm, and the running data is divided into normal operation state and abnormal fault state.

[0008] A parallel learning model is constructed and trained based on a Markov decision process. The operating data corresponding to the abnormal fault state is input into the parallel learning model to obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and to formulate maintenance decisions for the power supply and substation equipment.

[0009] Furthermore, in the above-mentioned parallel learning-based power supply and substation equipment maintenance method, the data preprocessing of the operating data further includes:

[0010] The operational data is treated as a normal distribution, and processed based on the 3σ principle of normal distribution to remove data with large errors that fall outside the (μ-3σ, μ+3σ) interval; where μ is the mean of the operational data and σ is the standard deviation of the normal distribution of the operational data.

[0011] Furthermore, in the above-mentioned parallel learning-based power supply and substation equipment maintenance method, the data preprocessing of the operating data further includes:

[0012] The running data is standardized by using a normalization method to process the running data and convert the data into values ​​between 0 and 1.

[0013] Furthermore, in the above-mentioned parallel learning-based power supply and substation equipment maintenance method, the data preprocessing of the operating data further includes:

[0014] Principal component analysis was used to select several representative composite data points from the operational data.

[0015] Furthermore, in the above-mentioned power supply and substation equipment maintenance method based on parallel learning, the processing of the preprocessed operating data based on the fuzzy C-means clustering algorithm specifically includes:

[0016] Using the set of preprocessed running data as a sample set, determine the number of categories in the sample set and set initial cluster centers;

[0017] Calculate the Euclidean distance between each sample point and the initial cluster center;

[0018] The cluster center points are updated based on the current Euclidean distance between each sample point and the cluster center, and the Euclidean distance is recalculated.

[0019] The operational data is classified into either the normal operation state category or the abnormal fault state category based on the final Euclidean distance.

[0020] Furthermore, the above-mentioned power supply and substation equipment maintenance method based on parallel learning, wherein the construction of the parallel learning model, training the parallel learning model based on a Markov decision process, inputting the operating data corresponding to the abnormal fault state into the parallel learning model, obtaining the predicted results of the power supply and substation equipment failure probability and remaining service life, and formulating maintenance decisions for the power supply and substation equipment, specifically includes:

[0021] Construct parallel learning models;

[0022] After performing calculation experiments and predictive learning on the operational data corresponding to the abnormal fault states, several artificial data points are generated.

[0023] Artificial systems are constructed using the artificial data.

[0024] Furthermore, the above-mentioned power supply and substation equipment maintenance method based on parallel learning, wherein the construction of a parallel learning model, training the parallel learning model based on a Markov decision process, inputting the operating data corresponding to the abnormal fault state into the parallel learning model, obtaining the predicted results of the power supply and substation equipment failure probability and remaining service life, and formulating maintenance decisions for the power supply and substation equipment, further includes:

[0025] Based on the Markov decision process, the current state of the artificial system is randomly obtained, and the current learning action is recorded.

[0026] A current knowledge matrix is ​​established based on the current artificial system state and the current learning action, and an immediate reward is calculated based on the current artificial system state, the next moment's artificial system state, and the current learning action.

[0027] The knowledge matrix for the next time step is updated based on the current knowledge matrix and the immediate reward.

[0028] The knowledge matrix is ​​stored in the state transition function.

[0029] Furthermore, the above-mentioned power supply and substation equipment maintenance method based on parallel learning, wherein the construction of a parallel learning model, training the parallel learning model based on a Markov decision process, inputting the operating data corresponding to the abnormal fault state into the parallel learning model, obtaining the predicted results of the power supply and substation equipment failure probability and remaining service life, and formulating maintenance decisions for the power supply and substation equipment, further includes:

[0030] Find the action value that maximizes the knowledge matrix;

[0031] Repeat the step of randomly obtaining the state of the artificial system until all the artificial systems are traversed to obtain the failure probability and service life of the power supply and transformation equipment;

[0032] Maintenance decisions for power supply and transformation equipment are made based on the failure probability and service life of the equipment.

[0033] According to a second aspect of the present invention, a power supply and transformation equipment maintenance system based on parallel learning is also provided, the system comprising:

[0034] The data acquisition and processing module is used to collect real-time operating data of power supply and transformation equipment and to perform data preprocessing on the operating data;

[0035] The fuzzy C-means clustering algorithm module is used to process the preprocessed running data and divide the running data into normal operation state and abnormal fault state;

[0036] The prediction and decision-making module is used to construct a parallel learning model. The parallel learning model is trained based on a Markov decision process. The operating data corresponding to the abnormal fault state is input into the parallel learning model to obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and to formulate maintenance decisions for the power supply and substation equipment.

[0037] According to a third aspect of the present invention, a power supply and substation equipment maintenance device based on parallel learning is also provided, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program that, when executed by the processing unit, causes the processing unit to perform the steps of any of the methods described above.

[0038] According to a fourth aspect of the invention, a storage medium is also provided, which stores a computer program executable by a power supply and substation maintenance device based on parallel learning, wherein when the computer program is run on the power supply and substation maintenance device based on parallel learning, the power supply and substation maintenance device based on parallel learning performs the steps of any of the methods described above.

[0039] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0040] (1) The power supply and transformation equipment maintenance method based on parallel learning provided by the present invention processes the collected power supply and transformation equipment operation data through the fuzzy C-means clustering algorithm to directly obtain the current operating status of the equipment; through the parallel learning model, the complex engineering modeling process is avoided, and the failure probability and remaining service life of the power supply and transformation equipment can be quickly obtained, reducing the pressure of maintenance and repair work.

[0041] (2) The power supply and transformation equipment maintenance method based on parallel learning provided by the present invention creates a predictive maintenance method based on big data and artificial intelligence technology, avoids the use of degradation models, and has good versatility and generalization ability. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 A flowchart illustrating the power supply and transformation equipment maintenance method based on parallel learning provided in this application embodiment;

[0044] Figure 2A schematic diagram of the structure of a power supply and transformation equipment maintenance system based on parallel learning provided in an embodiment of this application. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0046] The terms "first," "second," "third," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0047] On the one hand, this application provides a method for maintaining power supply and transformation equipment based on parallel learning. Figure 1 Please refer to the flowchart of the power supply and transformation equipment maintenance method based on parallel learning provided in the embodiments of this application. Figure 1 The method includes the following steps:

[0048] (1) Collect real-time operating data of power supply and transformation equipment, and perform data preprocessing on the operating data;

[0049] The collected real-time operational data is cleaned in the following ways: data with missing fields is removed or completed; data with incorrect format and content is removed or modified; data with logical errors is removed or modified; unnecessary data is removed; the correct feature attributes are selected by verifying the correlation between data; all relevant data are summarized and integrated, and multi-source heterogeneous data are modeled into a unified data model to integrate the data.

[0050] Furthermore, assuming the operational data follows a normal distribution, the 3σ principle is applied to process the data. In a normal distribution, the probability of a value falling within (μ-σ, μ+σ) is 0.6826, within (μ-2σ, μ+2σ) is 0.9544, and within (μ-3σ, μ+3σ) is 0.9974. This means that almost all values ​​in a set of operational or historical data are concentrated within the (μ-3σ, μ+3σ) interval, with the probability of values ​​exceeding this range being less than 0.3%. Here, μ is the mean of the operational data, and σ is the standard deviation of the normal distribution. Therefore, an error equal to ±3σ is typically considered the limiting error. If the absolute value of the residual error of a data point in a set of operational or historical data is greater than 3σ, then the measurement value is considered an outlier, and the corresponding sample is an anomaly that should be removed.

[0051] As a preferred embodiment, the running data is processed using a normalization method. Without changing the data distribution, the data is transformed into continuous variables between 0 and 1, accelerating the convergence of machine learning. This can be expressed as Equation (1), where the original running data value is X = (x1, x2, ..., x...). n ),

[0052]

[0053] Where max(X) and min(X) represent the maximum and minimum values ​​of the running data, respectively.

[0054] As a preferred embodiment, the approach employs a reduction strategy, transforming a large amount of operational data into a few comprehensive datasets, i.e., principal components. Each principal component reflects most of the information from the original operational data variables, and the information contained within them is non-overlapping. This method simplifies the problem by introducing multiple variables and reducing complex factors to a few principal components, while simultaneously yielding more scientifically sound and effective data.

[0055] (2) Based on the fuzzy C-means clustering algorithm, the preprocessed running data is processed and divided into normal operation state and abnormal fault state;

[0056] The objective function of the fuzzy C-means clustering algorithm is the sum of the products of the membership degree of each sample point and the Euclidean distance between that point and the cluster center. The fuzzy C-means clustering algorithm is to find the partition matrix U and the cluster center matrix C that minimize the clustering objective function, which can be expressed as equation (2).

[0057]

[0058] Where n is the number of sample datasets; c is the cluster center; m is the fuzzy weighting index; d ijThis represents the Euclidean distance between a sample point and the cluster center.

[0059] Using the preprocessed data set as the sample set, the number of categories in the sample set is determined, and an initial cluster center is set. The cluster center can be expressed as Equation (3).

[0060]

[0061] The Euclidean distance between each sample point and the initial cluster center can be expressed as Equation (4).

[0062] d ij (x j ,c i )=||x j -c i ||(4)

[0063] The cluster center points are updated based on the current Euclidean distance between each sample point and the cluster center, and the Euclidean distance is recalculated.

[0064] The membership degree of each sample point calculated based on Euclidean distance can be expressed as Equation (5).

[0065]

[0066] The operational data is categorized into normal operating status or abnormal fault status based on the final Euclidean distance.

[0067] (3) Construct a parallel learning model, train the parallel learning model based on the Markov decision process, input the operating data corresponding to the abnormal fault state into the parallel learning model, obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and formulate maintenance decisions for the power supply and substation equipment.

[0068] A parallel learning model is constructed. After performing calculation experiments and predictive learning on the operational data corresponding to abnormal fault states, several artificial data are generated. An artificial system is constructed through the artificial data.

[0069] Based on the Markov decision process, the current state of the artificial system is randomly obtained, and the current learning action is recorded. A current knowledge matrix is ​​established based on the current artificial system state and the current learning action. An immediate reward is calculated based on the current artificial system state, the next artificial system state, and the current learning action. The knowledge matrix for the next moment is updated based on the current knowledge matrix and the immediate reward. The knowledge matrix is ​​stored in the state transition function. The action value that maximizes the knowledge matrix is ​​solved. The step of randomly obtaining the state of the artificial system is repeated until all artificial systems are traversed, obtaining the failure probability and service life of the power supply equipment. The above process can be expressed as equations (6) and (7).

[0070]

[0071]

[0072] Where Q is the knowledge matrix, s is the artificial system state, a is the learning action, and l, p, and k represent the l-th real code used for transfer learning, the p-th search, and the k-th iteration, respectively; R ij LP is an immediate reward, generally derived from optimization goals; Q ij l and ΔQ ij Here, represents the knowledge matrix and its increment; q0 is a random value in the unified probability distribution; ε is the local greedy search parameter; a rand This represents a global random search action. α is the learning parameter, and γ is an adjustable parameter.

[0073] Maintenance decisions for power supply and transformation equipment are made based on the failure probability and service life of the equipment. These decisions include the level of repair and the optimal repair time. The level of repair includes minor repairs, major repairs, and equipment replacement.

[0074] On the other hand, this application also provides a power supply and transformation equipment maintenance system based on parallel learning. Figure 2 Please refer to the schematic diagram of the power supply and transformation equipment maintenance system based on parallel learning provided in the embodiments of this application. Figure 2 The system includes: a data acquisition and preprocessing module, a fuzzy C-means clustering algorithm module, and a prediction and decision-making module.

[0075] The data acquisition and processing module is used to collect real-time operating data of power supply and transformation equipment and to perform data preprocessing on the operating data;

[0076] The fuzzy C-means clustering algorithm module is used to process the preprocessed running data and divide the running data into normal operation state and abnormal fault state;

[0077] The prediction and decision-making module is used to construct a parallel learning model. The parallel learning model is trained based on a Markov decision process. The operating data corresponding to the abnormal fault state is input into the parallel learning model to obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and to formulate maintenance decisions for the power supply and substation equipment.

[0078] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, as well as magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0079] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0080] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0081] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0082] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0083] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0084] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0085] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0086] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of embodiments of this disclosure upon considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

[0087] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0088] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A power transformer maintenance method based on parallel learning, characterized by, include: Collect real-time operating data of power supply and transformation equipment, and perform data preprocessing on the operating data; The preprocessed running data is processed based on the fuzzy C-means clustering algorithm, and the running data is divided into normal operation state and abnormal fault state. A parallel learning model is constructed, and the parallel learning model is trained based on the Markov decision process. The operating data corresponding to the abnormal fault state is input into the parallel learning model to obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and to formulate maintenance decisions for the power supply and substation equipment. The construction of the parallel learning model involves training the model based on a Markov decision process, inputting the operational data corresponding to the abnormal fault states into the parallel learning model, obtaining prediction results of the failure probability and remaining service life of the power supply and substation equipment, and formulating maintenance decisions for the power supply and substation equipment. Specifically, this includes: Construct parallel learning models; After performing calculation experiments and predictive learning on the operational data corresponding to the abnormal fault states, several artificial data points are generated. Artificial systems are constructed using the artificial data; Based on the Markov decision process, the current state of the artificial system is randomly obtained, and the current learning action is recorded. A current knowledge matrix is ​​established based on the current artificial system state and the current learning action, and an immediate reward is calculated based on the current artificial system state, the next moment's artificial system state, and the current learning action. The knowledge matrix for the next time step is updated based on the current knowledge matrix and the immediate reward. The knowledge matrix is ​​stored in the state transition function; Find the action value that maximizes the knowledge matrix; Repeat the step of randomly obtaining the state of the artificial system until all the artificial systems are traversed to obtain the failure probability and service life of the power supply and transformation equipment; Maintenance decisions for power supply and transformation equipment are made based on the failure probability and service life of the equipment.

2. The method for power distribution equipment maintenance based on parallel learning according to claim 1, wherein, The data preprocessing of the operational data specifically includes: The operational data is considered to follow a normal distribution, based on the 3... The principle is to process the operational data and remove data distributed in (…). Data with large out-of-range errors; among them, The mean of the running data, The standard deviation of the normal distribution of the running data.

3. The power supply and transformation equipment maintenance method based on parallel learning as described in claim 2, wherein, The data preprocessing of the operational data also includes: The running data is standardized by using a normalization method to process the running data and convert the data into values ​​between 0 and 1.

4. The power supply and transformation equipment maintenance method based on parallel learning as described in claim 3, wherein, The data preprocessing of the operational data also includes: Principal component analysis was used to select several representative composite data points from the operational data.

5. The method for power distribution equipment maintenance based on parallel learning according to claim 1, wherein, The process of processing the preprocessed running data based on the fuzzy C-means clustering algorithm specifically includes: Using the set of preprocessed running data as a sample set, determine the number of categories in the sample set and set initial cluster centers; Calculate the Euclidean distance between each sample point and the initial cluster center; The cluster center points are updated based on the current Euclidean distance between each sample point and the cluster center, and the Euclidean distance is recalculated. The operational data is categorized into normal operating status or abnormal fault status based on the final Euclidean distance.

6. A power transformer maintenance system based on parallel learning, characterized by, include: The data acquisition and processing module is used to collect real-time operating data of power supply and transformation equipment and to perform data preprocessing on the operating data; The fuzzy C-means clustering algorithm module is used to process the preprocessed running data and divide the running data into normal operation state and abnormal fault state; The prediction and decision-making module is used to construct a parallel learning model. The parallel learning model is trained based on a Markov decision process. The operating data corresponding to the abnormal fault state is input into the parallel learning model to obtain the prediction results of the failure probability and remaining service life of the power supply and substation equipment, and to formulate maintenance decisions for the power supply and substation equipment.

7. A power transformer maintenance device based on parallel learning, characterized by, It includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program that, when executed by the processing unit, causes the processing unit to perform the steps of the method according to any one of claims 1 to 5.

Images