Transformer substation risk identification and alarm prediction method based on causal inference

Abstract

The invention discloses a transformer substation risk identification and alarm prediction method based on causal inference, and relates to the technical field of power grid intelligent alarm, and the method comprises the steps: preprocessing first data to obtain a first database, and processing the first data through a first algorithm to obtain a second data set; on the basis of the second data operation parameters, correlation analysis is conducted on the second data operation parameters through a first analysis method, and a first relation graph is initialized; analyzing the first relation graph through a second algorithm to generate a first relation model, and correcting and verifying the first relation graph to obtain a second relation graph. According to the method, the causal relationship between the alarm data is established through causal inference, so that the internal relationship between the data can be reflected more accurately, and the prediction accuracy is improved; the causal relationship is analyzed by using the equipment operation parameters, the operation parameters of the alarm equipment are analyzed, and the causal graph is established, so that the method has higher universality, is small in limitation and is more persuasive to data.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent grid warning, and specifically to a substation risk identification and warning prediction method based on causal inference. Background Art

[0002] In recent years, with the rapid development of smart grids, substations, as the core components of the power system, the stable operation of their equipment is crucial for ensuring the safety and efficiency of the overall power grid. During the long-term operation of substation equipment, due to the influence of various internal and external factors, it is inevitable that the performance will decline or even fail. In order to detect equipment anomalies in a timely manner, substations usually equip a large number of sensors to monitor the operation status of the equipment and record relevant data. When an equipment anomaly is detected, the sensors will send out a series of warning messages, which are recorded in the warning database, providing important equipment operation status information for maintenance personnel.

[0003] Currently, existing substation warning prediction methods mainly rely on statistical inference, such as density clustering algorithms, association rule mining, etc. Although these methods can discover the correlation between warning data, they cannot distinguish the causal relationship behind the correlation. For example, the density clustering algorithm can only divide warning data into clusters with similar characteristics, but cannot determine the causal relationship between warnings within the cluster. And association rule mining can only discover the frequent item sets between warning data, but cannot determine the causal relationship between frequent item sets. In addition, existing methods also have problems such as insufficient universality, incomplete coverage of causal relationships, and relatively high model complexity. To solve the above problems, the present invention proposes a substation risk identification and warning prediction method based on causal inference. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: existing substation warning prediction methods have problems such as insufficient universality, incomplete coverage of causal relationships, relatively high model complexity, and how to improve the accuracy of substation warning prediction.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A substation risk identification and warning prediction method based on causal inference, including obtaining historical first data of the substation, preprocessing the first data to obtain a first database, processing the first data through a first algorithm to obtain a second data set; obtaining real-time second data operation parameters of the substation, performing a correlation analysis on the second data operation parameters through a first analysis method, and initializing a first relationship graph; analyzing the first relationship graph through a second algorithm to generate a first relationship model, and based on the first relationship model, correcting and validating the first relationship graph to obtain a second relationship graph.

[0007] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: obtaining the first database includes classifying the acquired historical first data, sorting them, forming a first sequence, and storing the first sequence in a database to obtain the first database.

[0008] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: obtaining the second data set includes performing algorithm processing on each sequence in the first database through a first algorithm, and forming the processed data into a second data set.

[0009] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: performing correlation analysis on the operating parameters of the second data through the first analysis method includes based on the acquired operating parameters of the second data, and analyzing the correlation between data through the first analysis method.

[0010] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: initializing the first relationship graph includes analyzing and initializing the first relationship graph based on the first analysis method, and configuring the weight relationship of the relationship graph based on the strength of the association.

[0011] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: generating the first relationship model includes generating the first relationship model based on a second algorithm and identifying the relationship between the second data.

[0012] As a preferred embodiment of the substation risk identification and alarm prediction method based on causal inference according to the present invention, wherein: obtaining the second relationship graph includes pruning the data identified by the first relationship model that is irrelevant to the relationship graph to obtain the second relationship graph.

[0013] Another object of the present invention is to provide a substation risk identification and alarm prediction system based on causal inference, which can analyze alarm data by using a causal inference algorithm, construct a causal relationship graph model, and perform alarm prediction based on the causal relationship graph, solving the technical problems of the current substation alarm prediction technology based on statistical inference, including the inability to distinguish the causal relationship behind the association, insufficient universality, incomplete coverage of the causal relationship, and relatively high model complexity.

[0014] As a preferred solution of the substation risk identification and warning prediction system based on causal inference according to the present invention, it includes a data processing module, a relationship analysis module, and an identification and warning module; the data processing module is used to obtain the historical first data of the substation, preprocess the first data to obtain the first database, and process the first data through the first algorithm to obtain the second data set; the relationship analysis module is used to obtain the second data operation parameters of the substation, perform correlation analysis on the second data operation parameters through the first analysis method, and initialize the first relationship graph; the identification and warning module is used to analyze the first relationship graph through the second algorithm to generate the first relationship model, and based on the first relationship model, correct and verify the first relationship graph to obtain the second relationship graph.

[0015] A computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the substation risk identification and warning prediction method based on causal inference.

[0016] A computer-readable storage medium stores a computer program thereon, and the computer program, when executed by a processor, implements the steps of the substation risk identification and warning prediction method based on causal inference.

[0017] The beneficial effects of the present invention: The substation risk identification and warning prediction method based on causal inference provided by the present invention establishes the causal relationship between warning data through causal inference, which can not only more accurately reflect the internal relationship between data, but also improve the prediction accuracy; by using the equipment operation parameters to analyze the causal relationship and establishing a causal graph by analyzing the operation parameters of the warning equipment, it has stronger universality, smaller limitations, and more persuasive data; through the causal model, an intuitive and clear causal path graph is provided, which not only facilitates the maintenance personnel to understand the influence relationship between variables in the substation, but also helps the maintenance personnel to accurately judge the intervention result when implementing intervention measures, improving the operability of the fault prediction model; through the risk identification and warning prediction means provided by the present invention for the substation, it not only reduces the unstable risk caused by substation failures, but also effectively improves the operation efficiency and safety of the substation, ensuring the stable operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1The overall flowchart of a substation risk identification and warning prediction method based on causal inference provided for the first embodiment of the present invention.

[0020] Figure 2 The process flowchart of the DBSCAN density clustering algorithm for a substation risk identification and warning prediction method based on causal inference provided for the first embodiment of the present invention.

[0021] Figure 3 The flowchart for constructing a causal graph of a substation risk identification and warning prediction method based on causal inference provided for the first embodiment of the present invention.

[0022] Figure 4 The initial causal relationship graph of a substation risk identification and warning prediction method based on causal inference provided for the second embodiment of the present invention.

[0023] Figure 5 The corrected and verified causal relationship graph of a substation risk identification and warning prediction method based on causal inference provided for the second embodiment of the present invention.

[0024] Figure 6 The warning prediction accuracy graph of a substation risk identification and warning prediction method based on causal inference provided for the second embodiment of the present invention.

[0025] Figure 7 The overall flowchart of a substation risk identification and warning prediction system based on causal inference provided for the third embodiment of the present invention. Detailed implementation manners

[0026] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific implementation manners of the present invention will be given in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Embodiment 1, referring to Figures 1 - 3 , which is an embodiment of the present invention, provides a substation risk identification and warning prediction method based on causal inference, including:

[0028] S1: Obtain the historical first data of the substation, preprocess the first data to obtain the first database, and process the first data through the first algorithm to obtain the second data set.

[0029] Further, obtaining the first database includes classifying the acquired historical first data, sorting it, forming a first sequence, and storing the first sequence in the database to obtain the first database.

[0030] The first data includes, but is not limited to, substation alarm data.

[0031] It should be noted that in the embodiment of the present application, historical first data of the substation is acquired, and the first data is preprocessed to obtain the first database. The specific steps are as follows: Collect historical alarm data from multiple substations. The historical alarm data includes a timestamp and operating parameters.

[0032] The timestamp includes the time point when the alarm is recorded, usually represented in the form of a date and time.

[0033] The operating parameters include the operating parameters of the device where the alarm occurs at the alarm moment.

[0034] The collected historical alarm data is classified according to the affiliated substation, and sorted in ascending order of the alarm occurrence time in each alarm data classified based on the substation to form an alarm sequence. The alarm sequences of all substations are stored in a centralized alarm database, that is, the first database.

[0035] It should also be noted that obtaining the second data set includes performing algorithm processing on each sequence in the first database through a first algorithm, and forming the processed data into a second data set.

[0036] The second data set includes, but is not limited to, an alarm cluster set.

[0037] It should also be noted that the first algorithm can be a DBSCAN density clustering algorithm, a K-Means algorithm, or other algorithms suitable for clustering division.

[0038] It should also be noted that in the embodiment of the present application, the DBSCAN density clustering algorithm is used as the first algorithm to perform clustering division on each day's alarm sequence in the alarm database to obtain the alarm cluster of each alarm sequence.

[0039] There is an association relationship between the alarms within the divided clusters, and there is no association relationship between the alarms between the clusters. All the alarm clusters are formed into an alarm cluster set.

[0040] The steps of the DBSCAN density clustering algorithm for alarm cluster division are as follows:

[0041] Input: Alarm sequence X = {x1, x2, x3,..., xn}, distance function d(xi, xj) = |ti - tj|, neighborhood radius eps, minimum number of core points MinPts.

[0042] Output: The clustering division C = {c 1 , c 2 , …, c k}.

[0043] The process is as shown in Figure 3.

[0044] It should also be noted that in an alternative embodiment, the first algorithm uses the K-Means algorithm to process each sequence in the alarm database. The specific steps are as follows: Based on the first database, plot the curve of the number of clusters K and the sum of squared errors within the clusters, take the "elbow" of the curve as the value of K, regard each sequence in the first database as a sample point, calculate the silhouette coefficient of each sample point, select the value of K that maximizes the silhouette coefficient, randomly select K sample points as the initial centroids, calculate the distance between each sample point and each centroid, and assign the sample point to the cluster to which the nearest centroid belongs, update the centroid of each cluster. Updating the centroid of each cluster includes the mean value of all sample points within the cluster. Repeat determining the number of clusters K and the K-Means algorithm clustering until the change in the centroid is less than the threshold or the maximum number of iterations is reached, calculate the silhouette coefficient of each cluster, and evaluate the quality of the cluster. The closer the silhouette coefficient value is to 1, the better the quality of the cluster. Combine all clusters to form the alarm cluster set.

[0045] S2: Obtain the second data operation parameters of the substation, perform correlation analysis on the second data operation parameters through the first analysis method, and initialize the first relationship graph.

[0046] Furthermore, performing correlation analysis on the second data operation parameters through the first analysis method includes based on the obtained operation parameters of the second data, and analyzing the correlation between data through the first analysis method.

[0047] It should be noted that the first analysis method can be the correlation coefficient method, or it can be to draw a scatter plot for regression analysis, or other methods suitable for mechanical energy correlation analysis such as similarity-based analysis.

[0048] It should also be noted that initializing the first relationship graph includes initializing the first relationship graph based on the analysis of the first analysis method, and configuring the weight relationship of the relationship graph based on the strength of the correlation.

[0049] It should also be noted that the first relationship graph includes but is not limited to causal relationship graphs, Bayesian networks, directed graphs, etc.

[0050] It should also be noted that in the embodiments of the present application, the first relationship graph is a causal relationship graph.

[0051] Initialize the causal relationship graph. Specifically, initialize a causal relationship graph based on correlation analysis. In the causal relationship graph, nodes represent various warnings, edges represent the degree of correlation between nodes, and the weights of the edges represent the strength of the causal relationship.

[0052] If the correlation between two alarms is strong, it is considered that there is a high possibility of a causal relationship between the alarms, and the greater the weight of the edge.

[0053] It should also be noted that in an alternative embodiment, the first relationship graph is a Bayesian network.

[0054] Initialize the Bayesian network. Specifically, define each alarm as a node. Based on the results of correlation coefficient analysis, determine the dependency relationship between alarms. If the correlation coefficient between two alarms is large, it is considered that there is a dependency relationship between them. A directed edge can be used to represent the dependency relationship, and the arrow points to the affected alarm. Determine the parent nodes of the node. The parent node refers to the node that has a direct impact on the current node. The parent nodes can be determined based on correlation coefficient analysis and domain knowledge. Each node can have multiple parent nodes or no parent nodes. For each node, construct a conditional probability table to represent the conditional probability of the node under different states of the parent nodes. The conditional probability can be estimated by using historical alarm data.

[0055] S3: Analyze the first relationship graph through a second algorithm to generate a first relationship model. Based on the first relationship model, correct and verify the first relationship graph to obtain a second relationship graph.

[0056] Furthermore, generating the first relationship model includes generating the first relationship model based on the second algorithm and identifying the relationship between the second data.

[0057] It should be noted that in the embodiment of the present application, the second algorithm used is a causal inference algorithm, and the first relationship model is a causal relationship model.

[0058] Generating the first relationship model includes analyzing based on the causal inference algorithm to generate a causal relationship graph model and identifying the causal relationship between alarms.

[0059] The causal inference theory is based on the structural causal model. The core content of the structural causal model is the formalized operator of the "intervention" concept - do(·). Define the causal effect according to the do operator. For example, the average causal effect of Z on Y is defined as:

[0060]

[0061] Where and represent the expected values of the result variable when the intervention variable is 1 and 0 respectively. The greater the difference between the two, the stronger the causal effect.

[0062] Use the identified causal effect as the weight of the edge.

[0063] It should be noted that obtaining the second relationship graph includes pruning the data identified by the first relationship model that is irrelevant to the relationship graph to obtain the second relationship graph.

[0064] It should also be noted that in the embodiment of the present application, the second relationship graph is a causal relationship graph after being corrected and verified.

[0065] Prune the alarms without causal relationships in the causal relationship graph according to the identified causal relationships, and retain the alarms and paths with strong causal relationships, and finally obtain a complete causal relationship graph.

[0066] Embodiment 2, refer to Figures 4 - 6 , which is an embodiment of the present invention, provides a method for substation risk identification and alarm prediction based on causal inference. In order to verify the beneficial effects of the present invention, scientific demonstrations are carried out through economic benefit calculation and simulation experiments.

[0067] In order to verify the effectiveness and superiority of the method for substation risk identification and alarm prediction based on causal inference, this embodiment is provided. The embodiment shows some parameters. The alarms recorded in the alarm database of a substation in this embodiment include transformer overload, too high transformer ambient temperature, too high transformer oil temperature, too low transformer oil level, too high transformer winding temperature, transformer winding fault, abnormal bus current, circuit breaker trip, abnormal grid frequency, too high transformer voltage, transformer fault, and transformer oil level indicator trip.

[0068] The following clusters are obtained by clustering through the DBSCAN clustering algorithm:

[0069] Cluster 1: {Transformer overload, too high transformer temperature, too high transformer oil temperature, too low transformer oil level, too high transformer winding temperature, transformer winding fault, abnormal bus current, too high transformer voltage, transformer fault, transformer oil level indicator trip, circuit breaker trip}.

[0070] Cluster 2: {Abnormal grid frequency}.

[0071] Perform a correlation analysis on the data in Cluster 1, calculate the correlation coefficient, and obtain the correlation coefficient analysis data shown in Table 2.

[0072] Table 2 Correlation coefficient analysis of Cluster 1

[0073]

[0074]

[0075] Initialize a causal relationship graph based on relevance, as shown in Figure 4 shown.

[0076] Calculate the average causal effect, expressed as:

[0077] ACE = E{Y / do(X)=1} - E{Y / do(X)=0}

[0078] Obtain the results shown in Table 3 below according to the calculation.

[0079] Table 3 Average causal effect data

[0080]

[0081] According to the calculated average causal effect ACE, if the calculated value is lower than 0.01, it is determined that the causal relationship between the two variables is not significant, and the corresponding edge should be removed from the causal relationship graph. The causal relationship graph after causal analysis and correction is shown in Figure 5 shown.

[0082] Based on the corrected causal relationship graph, complete the alarm prediction, and obtain the result as shown in Figure 6 shown, with an accuracy rate of approximately 92%, while the accuracy rate of the method based on statistical inference is only 85.7%. Compared with our invention, the causal inference method can more accurately identify the risks of the substation and give early warnings.

[0083] The present invention applies the causal inference method to the field of substation risk identification and alarm prediction, overcomes the limitations of traditional statistical inference methods, and improves the prediction accuracy and interpretability.

[0084] Example 3, referring to Figure 7 , is an embodiment of the present invention, which provides a substation risk identification and alarm prediction system based on causal inference, including a data processing module, a relationship analysis module, and an identification and warning module.

[0085] Among them, the data processing module is used to obtain the historical first data of the substation, preprocess the first data to obtain the first database, and process the first data through the first algorithm to obtain the second data set; the relationship analysis module is used to obtain the second data operation parameters of the substation, perform correlation analysis on the second data operation parameters through the first analysis method, and initialize the first relationship graph; the identification and warning module is used to analyze the first relationship graph through the second algorithm to generate the first relationship model, and based on the first relationship model, correct and verify the first relationship graph to obtain the second relationship graph.

[0086] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0087] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a predefined sequence list of executable instructions for implementing a logical function, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0088] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or more wirings, a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber device, and portable compact disc read-only memory (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0089] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A substation risk identification and alarm prediction method based on causal inference, characterized in that: include: Acquire historical first data of the substation, pre-process the first data to obtain a first database, and process the first data using a first algorithm to obtain a second data set; Acquire a second data operating parameter of the substation, perform a correlation analysis on the second data operating parameter using a first analysis method, and initialize a first relationship graph; The first relationship graph is analyzed by a second algorithm to generate a first relationship model, and based on the first relationship model, the first relationship graph is corrected and verified to obtain a second relationship graph.

2. The substation risk identification and alarm prediction method based on causal inference according to claim 1 is characterized in that: The obtaining of the first database includes classifying and sorting the acquired historical first data to form a first sequence, storing the first sequence in a database, and obtaining the first database.

3. The substation risk identification and alarm prediction method based on causal inference according to claim 2 is characterized in that: The obtaining of the second data set includes performing algorithmic processing on each sequence in the first database by using the first algorithm, and forming the processed data into the second data set.

4. The substation risk identification and alarm prediction method based on causal inference according to claim 3 is characterized in that: The performing correlation analysis on the operating parameters of the second data by the first analysis method includes analyzing the correlation between the data by the first analysis method based on the acquired operating parameters of the second data.

5. The substation risk identification and alarm prediction method based on causal inference according to claim 4 is characterized in that: The initializing the first relationship graph includes analyzing and initializing the first relationship graph based on a first analysis method, and configuring a weight relationship of the relationship graph based on the strength of the correlation.

6. The substation risk identification and alarm prediction method based on causal inference according to claim 5 is characterized in that: The generating the first relationship model includes generating the first relationship model based on a second algorithm and identifying the relationship between the second data.

7. The substation risk identification and alarm prediction method based on causal inference according to claim 6 is characterized in that: The obtaining of the second relationship graph includes pruning data irrelevant to the relationship graph identified by the first relationship model to obtain the second relationship graph.

8. A system using the substation risk identification and alarm prediction method based on causal inference as claimed in any one of claims 1 to 7, characterized in that: Including data processing module, relationship analysis module, identification and early warning module; The data processing module is used to obtain historical first data of the substation, pre-process the first data to obtain a first database, and process the first data using a first algorithm to obtain a second data set; The relationship analysis module is used to obtain the second data operation parameter of the substation, perform correlation analysis on the second data operation parameter by using the first analysis method, and initialize the first relationship graph; The identification and early warning module is used to analyze the first relationship graph through a second algorithm to generate a first relationship model, and based on the first relationship model, correct and verify the first relationship graph to obtain a second relationship graph.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the substation risk identification and alarm prediction method based on causal inference described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the substation risk identification and alarm prediction method based on causal inference described in any one of claims 1 to 7 are implemented.

Images