A power failure prediction system and method
By establishing a Gaussian mixture model and knowledge graph, the problem of accuracy in judging power equipment faults was solved, enabling power equipment fault prediction and fault cause localization, and providing decision support.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STATE GRID FUJIAN ELECTRIC POWER CO LTD
- Filing Date
- 2023-12-29
- Publication Date
- 2026-07-07
AI Technical Summary
Existing technologies cannot accurately determine power equipment faults and their causes, especially when there is a high degree of mutual influence between power equipment in complex power grids, making it difficult to trace the specific functional node of the fault.
By establishing a Gaussian mixture model and knowledge graph, and combining historical and topological data of power equipment, the probability of power equipment failure can be predicted and the cause of failure can be located.
It enables accurate prediction of power equipment failures and accurate location of failure causes, providing decision support.
Smart Images
Figure CN117909864B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power fault technology, and specifically to a power fault prediction system and method. Background Technology
[0002] Currently, with the development and evolution of power grids, electrical equipment is becoming increasingly complex, and the degree of mutual influence between these devices is growing. The number of functional nodes connecting these devices is also increasing, including transformers and control devices. Often, equipment failures are caused by faults in these functional nodes. This increasing interdependence makes it difficult for technicians to determine whether a particular device is faulty based solely on power information. Furthermore, the growing number of functional nodes makes it impossible to accurately pinpoint which node caused the fault.
[0003] Therefore, in order to solve this problem, the present invention provides a power fault prediction system and method. Summary of the Invention
[0004] The purpose of this invention is to address the shortcomings of existing technologies by establishing a Gaussian distribution for each fault event and further constructing a Gaussian mixture model, which can accurately predict the probability of each power equipment fault, thereby providing support for decision-making, and providing a power fault prediction system and method.
[0005] This invention provides a power fault prediction system, comprising: an information acquisition module for acquiring fault events and corresponding historical power data, wherein the historical power data includes historical voltage data, historical current data, and historical power data of each power device; a model building module for establishing a corresponding Gaussian distribution based on each fault event and the corresponding historical power data, and then establishing a Gaussian mixture model; a knowledge graph building module for building a knowledge graph to associate each power device with each functional node; and a result analysis module for inputting the current voltage data, current data, and power data of the power device into the Gaussian mixture model, determining the probability of the current fault event, and determining the corresponding functional node that caused the current power device to fail based on the knowledge graph.
[0006] Furthermore, the specific steps for the model building module to build the Gaussian mixture model are as follows:
[0007] A Gaussian mixture model is established. Based on different fault events, corresponding Gaussian distributions are created, with each Gaussian distribution's data point x serving as input. Data point x represents the corresponding historical power data. The Gaussian mixture model is established as follows:
[0008] P(x)=∑ i w i *n(x|μ i,Σ i )
[0009] Where i is the nth Gaussian distribution, w i The mixture weights are the i-th Gaussian distribution, N(x|μ i ,Σ i Let be the mean μ of the given i-th Gaussian distribution. i The covariance matrix Σ i Given the probability density function, P(x) represents the probability obtained after inputting data point x, and the mean μ of each Gaussian distribution is calculated using the Expectation Maximization algorithm. i , covariance matrix ∑ i and mixed weight w i The calculation steps are as follows:
[0010] If the total number of various information categories in the information acquisition system to be monitored is set to i, then it represents i Gaussian distributions. In the Expectation Maximization algorithm, each mixture weight is initialized to... The calculation involves updating the mixed weights using alternating Expectation and Maximization algorithms until the model converges, yielding the mean, covariance matrix, and mixed weight parameters.
[0011] Furthermore, the knowledge graph building module builds a knowledge graph to associate various power devices and functional nodes. The specific steps are as follows:
[0012] Collect topology and operational data between various power devices and functional nodes, and then use the open-source OpenRefine data cleaning tool to clean and deduplicatize the topology and operational data between various power devices and functional nodes.
[0013] The connection relationships between each power device and its corresponding functional nodes are defined using the Cypher visual editor, forming a knowledge graph visualization data model.
[0014] Based on the knowledge graph visualization data model, functional nodes and corresponding power equipment entities are created in the open-source Amazon Neptune graph database. The connection relationships between entities are established, and then edges are created based on the connection relationships between power equipment and corresponding functional nodes. Natural language text is used to add relationship attributes to the edges to represent the connection relationships between power equipment and corresponding functional nodes, thus obtaining the knowledge graph.
[0015] Furthermore, the result analysis module inputs the current voltage, current, and power data of the power equipment into a Gaussian mixture model to determine the probability of the current fault event. The specific method for this is as follows:
[0016] By inputting the voltage, current, and power data of the current power equipment into the Gaussian mixture model, the output probability P(x) is the failure probability of the power equipment, and the following judgment rules apply:
[0017]
[0018] α is a parameter used to determine the probability of a power equipment failure. When α is 1, the power equipment has failed; when α is -1, the power equipment is normal; when α is 0, the power equipment is marked as abnormal and is monitored.
[0019] Furthermore, the specific method by which the result analysis module determines the corresponding functional node that caused the current power equipment failure based on the knowledge graph is as follows:
[0020] By retrieving the relational attributes of all edges connected to the current power equipment using keywords, the corresponding functional node causing the current power equipment failure can be identified.
[0021] A power fault prediction method, comprising:
[0022] Acquire fault events and corresponding historical power data, including historical voltage data, historical current data, and historical power data of each power device;
[0023] Based on each fault event and the corresponding historical power data, a corresponding Gaussian distribution is established, and then a Gaussian mixture model is established.
[0024] Establish a knowledge graph to connect various power devices and functional nodes;
[0025] The voltage, current, and power data of the current power equipment are input into the Gaussian mixture model to determine the probability of the current fault event, and the corresponding functional node that causes the current power equipment to fail is determined based on the knowledge graph.
[0026] Furthermore, the specific steps for the model building module to build the Gaussian mixture model are as follows:
[0027] A Gaussian mixture model is established. Based on different fault events, corresponding Gaussian distributions are created, with each Gaussian distribution's data point x serving as input. Data point x represents the corresponding historical power data. The Gaussian mixture model is established as follows:
[0028] P(x)=∑ i w i *N(x|μ i ,∑ i )
[0029] Where i is the nth Gaussian distribution, w i The mixture weights are the i-th Gaussian distribution, N(x|μ i ,∑ i Let be the mean μ of the given i-th Gaussian distribution. i The sum of the covariance matrix ∑ i Given the probability density function, P(x) represents the probability obtained after inputting data point x, and the mean μ of each Gaussian distribution is calculated using the Expectation Maximization algorithm. i , covariance matrix ∑ i and mixed weight w i The calculation steps are as follows:
[0030] If the total number of various information categories in the information acquisition system to be monitored is set to i, then it represents i Gaussian distributions. In the Expectation Maximization algorithm, each mixture weight is initialized to... The calculation involves updating the mixed weights using alternating Expectation and Maximization algorithms until the model converges, yielding the mean, covariance matrix, and mixed weight parameters.
[0031] Furthermore, the knowledge graph building module builds a knowledge graph to associate various power devices and functional nodes. The specific steps are as follows:
[0032] Collect topology and operational data between various power devices and functional nodes, and then use the open-source OpenRefine data cleaning tool to clean and deduplicatize the topology and operational data between various power devices and functional nodes.
[0033] The connection relationships between each power device and its corresponding functional nodes are defined using the Cypher visual editor, forming a knowledge graph visualization data model.
[0034] Based on the knowledge graph visualization data model, functional nodes and corresponding power equipment entities are created in the open-source Amazon Neptune graph database. The connection relationships between entities are established, and then edges are created based on the connection relationships between power equipment and corresponding functional nodes. Natural language text is used to add relationship attributes to the edges to represent the connection relationships between power equipment and corresponding functional nodes, thus obtaining the knowledge graph.
[0035] Furthermore, the result analysis module inputs the current voltage, current, and power data of the power equipment into a Gaussian mixture model to determine the probability of the current fault event. The specific method for this is as follows:
[0036] By inputting the voltage, current, and power data of the current power equipment into the Gaussian mixture model, the output probability P(x) is the failure probability of the power equipment, and the following judgment rules apply:
[0037]
[0038] α is a parameter used to determine the probability of a power equipment failure. When α is 1, the power equipment has failed; when α is -1, the power equipment is normal; when α is 0, the power equipment is marked as abnormal and is monitored.
[0039] The specific method used by the result analysis module to determine the corresponding functional node that caused the current power equipment failure based on the knowledge graph is as follows:
[0040] By retrieving the relational attributes of all edges connected to the current power equipment using keywords, the corresponding functional node causing the current power equipment failure can be identified.
[0041] A computer-readable medium storing a computer program that executes the above-described information interaction method when running.
[0042] The beneficial effects of this invention are as follows:
[0043] 1. By establishing a Gaussian distribution for each fault event and then constructing a Gaussian mixture model, the probability of each power equipment fault can be accurately predicted, thereby providing support for decision-making.
[0044] 2. By establishing a knowledge graph, we can not only predict the probability of a fault occurring, but also accurately locate the functional nodes that cause power equipment to fail. Attached Figure Description
[0045] Figure 1 This is a system block diagram of the present invention; Detailed Implementation
[0046] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.
[0047] A power fault prediction system includes: an information acquisition module for acquiring fault events and corresponding historical power data, wherein the historical power data includes historical voltage data, historical current data, and historical power data of each power device; a model building module for establishing a corresponding Gaussian distribution based on each fault event and the corresponding historical power data, and then establishing a Gaussian mixture model; a knowledge graph building module for building a knowledge graph to associate each power device with each functional node; and a result analysis module for inputting the voltage data, current data, and power data of the current power device into the Gaussian mixture model, determining the probability of the current fault event, and determining the corresponding functional node that caused the current power device to fail based on the knowledge graph.
[0048] in:
[0049] (1) The specific steps for the model building module to build the Gaussian mixture model are as follows:
[0050] A Gaussian mixture model is established. Based on different fault events, corresponding Gaussian distributions are created, with each Gaussian distribution's data point x serving as input. Data point x represents the corresponding historical power data. The Gaussian mixture model is established as follows:
[0051] P(x)=∑ i w i *N(x|μ i ,∑ i )
[0052] Where i is the nth Gaussian distribution, w i The mixture weights are the i-th Gaussian distribution, N(x|μ i ,∑ i Let be the mean μ of the given i-th Gaussian distribution. i The sum of the covariance matrix ∑ i Given the probability density function, P(x) represents the probability obtained after inputting data point x, and the mean μ of each Gaussian distribution is calculated using the Expectation Maximization algorithm. i , covariance matrix ∑ i and mixed weight w i The calculation steps are as follows:
[0053] If the total number of various information categories in the information acquisition system to be monitored is set to i, then it represents i Gaussian distributions. In the Expectation Maximization algorithm, each mixture weight is initialized to... The calculation involves updating the mixed weights using alternating Expectation and Maximization algorithms until the model converges, yielding the mean, covariance matrix, and mixed weight parameters.
[0054] The aforementioned Expectation Maximization algorithm specifically iterates the Gaussian mixture model by alternating between the Expectation algorithm and the Maximization algorithm. The more historical power data there is, the more accurate the Gaussian mixture model becomes.
[0055] (2) The knowledge graph building module builds a knowledge graph to associate various power devices and functional nodes. The specific steps are as follows:
[0056] Collect topology and operational data between various power devices and functional nodes, and then use the open-source OpenRefine data cleaning tool to clean and deduplicatize the topology and operational data between various power devices and functional nodes.
[0057] The connection relationships between each power device and its corresponding functional nodes are defined using the Cypher visual editor, forming a knowledge graph visualization data model.
[0058] Based on the knowledge graph visualization data model, functional nodes and corresponding power equipment entities are created in the open-source Amazon Neptune graph database. The connection relationships between entities are established, and then edges are created based on the connection relationships between power equipment and corresponding functional nodes. Natural language text is used to add relationship attributes to the edges to represent the connection relationships between power equipment and corresponding functional nodes, thus obtaining the knowledge graph.
[0059] The open-source OpenRefine data cleaning tool is a standard data cleaning tool used to remove noise, duplicates, and outliers from data.
[0060] The Amazon Neptune graph database model is developed based on code provided by Amazon, which can assist in the development of knowledge graphs.
[0061] Cypher is a visual editor that uses graphical representations to build large amounts of data. It provides an interface for constructing relationships between functional nodes and electrical devices.
[0062] (3) The result analysis module inputs the voltage, current, and power data of the current power equipment into the Gaussian mixture model to determine the probability of the current fault event. The specific method is as follows:
[0063] By inputting the voltage, current, and power data of the current power equipment into the Gaussian mixture model, the output probability P(x) is the failure probability of the power equipment, and the following judgment rules apply:
[0064]
[0065] α is a parameter used to determine the probability of a power equipment failure. When α is 1, the power equipment has failed; when α is -1, the power equipment is normal; when α is 0, the power equipment is marked as abnormal and is monitored.
[0066] The specific method used by the result analysis module to determine the corresponding functional node that caused the current power equipment failure based on the knowledge graph is as follows:
[0067] By retrieving the relational attributes of all edges connected to the current power equipment using keywords, the corresponding functional node causing the current power equipment failure can be identified.
[0068] The contents not described in detail in this specification are prior art known to those skilled in the art. Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0069] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0070] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0071] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.
Claims
1. A power fault prediction system, characterized in that, include: The information acquisition module is used to acquire fault events and corresponding historical power data, including historical voltage data, historical current data and historical power data of each power device. The model building module establishes a corresponding Gaussian distribution for each fault event and the corresponding historical power data, and then builds a Gaussian mixture model. The knowledge graph building module is used to build a knowledge graph, thereby linking various power devices and functional nodes; The results analysis module is used to input the voltage, current and power data of the current power equipment into the Gaussian mixture model, determine the probability of the current fault event, and determine the corresponding functional node that caused the current power equipment to fail based on the knowledge graph. The specific steps for the model building module to build a Gaussian mixture model are as follows: A Gaussian mixture model is established, and a corresponding Gaussian distribution is created based on different fault events. The data points in each Gaussian distribution... As input, data points Based on the corresponding historical electricity data, the following Gaussian mixture model is established: in, Which Gaussian distribution is it? It is the first A Gaussian distribution of mixed weights, It means that in a given number of... The mean of a Gaussian distribution Covariance Matrix The probability density under the following conditions Input data points The probabilities were then obtained, and the mean of each Gaussian distribution was calculated using the Expectation Maximization algorithm. Covariance matrix and mixed weights The calculation steps are as follows: The total number of various information categories in the information acquisition system that needs to be monitored is set to This means A Gaussian distribution is used, and each mixture weight is initialized to a certain value in the Expectation Maximization algorithm. And substitute them into the calculation; The mixed weights are updated by alternating between the Expectation and Maximization algorithms until the model converges, yielding the mean, covariance matrix, and mixed weight parameters.
2. The power fault prediction system according to claim 1, characterized in that: The knowledge graph building module builds a knowledge graph, thereby associating various power devices and functional nodes. The specific steps are as follows: Collect topology and operational data between various power devices and functional nodes, and then use the open-source OpenRefine data cleaning tool to clean and deduplicatize the topology and operational data between various power devices and functional nodes. The connection relationships between each power device and its corresponding functional nodes are defined using the Cypher visual editor, forming a knowledge graph visualization data model. Based on the knowledge graph visualization data model, functional nodes and corresponding power equipment entities are created in the open-source Amazon Neptune graph database. The connection relationships between entities are established, and then edges are created based on the connection relationships between power equipment and corresponding functional nodes. Natural language text is used to add relationship attributes to the edges to represent the connection relationships between power equipment and corresponding functional nodes, thus obtaining the knowledge graph.
3. The power fault prediction system according to claim 1, characterized in that: The result analysis module inputs the voltage, current, and power data of the current power equipment into a Gaussian mixture model to determine the probability of the current fault event. The specific method for this determination is as follows: By inputting the voltage, current, and power data of current power equipment into the Gaussian mixture model, the output probability is... The probability of failure of the power equipment is determined by the following rules: Parameters used to determine the probability of failure of electrical equipment. When the value is 1, the electrical equipment has malfunctioned. When the value is -1, the electrical equipment is normal. When the value is 0, the power equipment is marked as abnormal and is monitored.
4. The power fault prediction system according to claim 3, characterized in that: The specific method used by the result analysis module to determine the corresponding functional node that caused the current power equipment failure based on the knowledge graph is as follows: By retrieving the relational attributes of all edges connected to the current power equipment using keywords, the corresponding functional node causing the current power equipment failure can be identified.
5. A method for predicting power faults, characterized in that, include: Acquire fault events and corresponding historical power data, including historical voltage data, historical current data, and historical power data of each power device; Based on each fault event and the corresponding historical power data, a corresponding Gaussian distribution is established, and then a Gaussian mixture model is established. Establish a knowledge graph to connect various power devices and functional nodes; Input the voltage, current and power data of the current power equipment into the Gaussian mixture model to determine the probability of the current fault event, and determine the corresponding functional node that caused the current power equipment to fail based on the knowledge graph; The specific steps for establishing the Gaussian mixture model are as follows: A Gaussian mixture model is established, and a corresponding Gaussian distribution is created based on different fault events. The data points in each Gaussian distribution... As input, data points Based on the corresponding historical electricity data, the following Gaussian mixture model is established: in, Which Gaussian distribution is it? It is the first A Gaussian distribution of mixed weights, It means that in a given number of... The mean of a Gaussian distribution Covariance Matrix The probability density under the following conditions Input data points The probabilities were then obtained, and the mean of each Gaussian distribution was calculated using the Expectation Maximization algorithm. Covariance matrix and mixed weights The calculation steps are as follows: The total number of various information categories in the information acquisition system that needs to be monitored is set to This means A Gaussian distribution is used, and each mixture weight is initialized to a certain value in the Expectation Maximization algorithm. And substitute them into the calculation; The mixed weights are updated by alternating between the Expectation and Maximization algorithms until the model converges, yielding the mean, covariance matrix, and mixed weight parameters.
6. The power fault prediction method according to claim 5, characterized in that: The specific steps for establishing a knowledge graph to associate various power devices and functional nodes are as follows: Collect topology and operational data between various power devices and functional nodes, and then use the open-source OpenRefine data cleaning tool to clean and deduplicatize the topology and operational data between various power devices and functional nodes. The connection relationships between each power device and its corresponding functional nodes are defined using the Cypher visual editor, forming a knowledge graph visualization data model. Based on the knowledge graph visualization data model, functional nodes and corresponding power equipment entities are created in the open-source Amazon Neptune graph database. The connection relationships between entities are established, and then edges are created based on the connection relationships between power equipment and corresponding functional nodes. Natural language text is used to add relationship attributes to the edges to represent the connection relationships between power equipment and corresponding functional nodes, thus obtaining the knowledge graph.
7. The power fault prediction method according to claim 6, characterized in that: The specific method for inputting the current voltage, current, and power data of the power equipment into a Gaussian mixture model to determine the probability of the current fault event is as follows: By inputting the voltage, current, and power data of current power equipment into the Gaussian mixture model, the output probability is... The probability of failure of the power equipment is determined by the following rules: Parameters used to determine the probability of failure of electrical equipment. When the value is 1, the electrical equipment has malfunctioned. When the value is -1, the electrical equipment is normal. When the value is 0, the power equipment is marked as abnormal, and the power equipment is monitored. The specific method for determining the corresponding functional node that causes the current power equipment to malfunction based on the knowledge graph is as follows: By retrieving the relational attributes of all edges connected to the current power equipment using keywords, the corresponding functional node causing the current power equipment failure can be identified.
8. A computer-readable medium having a computer program stored thereon, the computer program executing the power fault prediction method as described in any one of claims 5-7 when run.