Substation operation and maintenance health analysis method

Abstract

The invention discloses a transformer substation operation and maintenance health analysis method, which comprises the following steps of: transmitting equipment operation data, environment data, inspection data and maintenance records acquired by a sensor to a big data storage platform by adopting a real-time data transmission protocol; the big data storage platform performs data cleaning and preprocessing on the received data, and then performs data storage; the big data storage platform adopts an FP-Growth algorithm to generate data association for the data subjected to data cleaning and preprocessing, and association between devices is generated; relates to the technical field of substation operation and maintenance, mainly discovers the relevance between equipment by analyzing historical data, establishes an equipment fault model by adopting a classification algorithm, and predicts the influence range of equipment faults, compared with a digital twinning method, the method is more suitable for overall risk assessment and trend analysis of a large-scale substation network, and the reliability of the network is improved. And a complex virtual model does not need to be constructed, so that the cost is relatively low.

Description

Technical Field

[0001] The present invention relates to the technical field of substation operation and maintenance, and particularly to a method for analyzing the health of substation operation and maintenance. Background Art

[0002] At present, remote video monitoring is adopted in substations, realizing basic remote image monitoring and remote inspection of the appearance of some equipment. In the prior art, the operation and maintenance of substation equipment mainly rely on inspection robots combined with various monitoring devices to collect and comprehensively analyze and diagnose the operation data of substation equipment.

[0003] Publication No. CN112564272B discloses a method for health control of substations integrating multi-dimensional monitoring and fault information. Through the monitoring and sensing device, the state data of substation equipment can be collected. Through the collector, the state data of the substation can be quickly uploaded to the server. Through the server, the health state of the substation can be controlled, improving the safety of the substation.

[0004] Publication No. CN111597231B discloses an intelligent inspection system for substations based on data mining of multi-source heterogeneous systems. By integrating various inspection systems that operation and maintenance personnel need to pay attention to, operation and maintenance personnel can have a clear understanding of the health state of substation equipment only by paying attention to this system, fundamentally improving the utilization rate of various systems and greatly reducing the workload of operation and maintenance personnel at the same time.

[0005] Publication No. CN117893032B discloses a method and system for analyzing the health of substation operation and maintenance based on digital twin, realizing the global perception of substation equipment, helping operation and maintenance personnel determine the faults and risks of associated equipment, and ensuring the safe and reliable operation of the substation.

[0006] However, the above applications still have the following problems: When diagnosing faults of equipment, Publication No. CN112564272B and Publication No. CN111597231B cannot obtain the equipment that may have potential risks associated with the current equipment. Although Publication No. CN117893032B discloses a method and system for analyzing the health of substation operation and maintenance based on digital twin, by constructing relationship mappings for each virtual device in the model, when diagnosing faults of equipment, the equipment that may have potential risks associated with the current equipment can be obtained through the relationship mappings. However, adopting the digital twin method requires constructing a complex virtual model, and the cost is relatively high. Therefore, the present invention proposes a method for analyzing the health of substation operation and maintenance. Summary of the Invention

[0007] To solve the technical problems in the background art and aiming at the deficiencies of the prior art, the present invention proposes a method for analyzing the health of substation operation and maintenance.

[0008] A substation operation and maintenance health analysis method proposed by the present invention includes the following steps: Using a real-time data transmission protocol, transmit the device operation data, environmental data, inspection data, and maintenance records collected by sensors to a big data storage platform; The big data storage platform performs data cleaning and preprocessing on the received data, and then stores the data; The big data storage platform uses the FP-Growth algorithm to generate device associations for the data after data cleaning and preprocessing, and find the associations between devices; For example, when a transformer of a certain model fails, which parameters of other transformers installed in the same batch and the devices connected to it often appear abnormal at the same time, and infer the potential risks that may exist in the transformers of the same batch and related devices.

[0009] Establish a device failure model. Through learning a large amount of historical failure data and using a classification algorithm, judge whether the device has a failure risk. If the device has a failure risk, send a warning signal, and according to the device associations generated by the FP-Growth algorithm for this device, predict the impact range of the device failure.

[0010] Preferably, the warning signal is conveyed to the operation and maintenance personnel by means of text messages, e-mails, and pop-up windows on the monitoring system interface.

[0011] Preferably, in establishing the device failure model, the classification algorithm used can be a decision tree algorithm, a random forest algorithm, or a support vector machine algorithm.

[0012] Preferably, in the big data storage platform, a distributed file system and a database (such as Hadoop, HBase) are used for data storage to ensure the security and scalability of the data.

[0013] Preferably, the data collected by the sensors includes parameter data such as oil temperature, oil level, winding temperature, voltage, current, power, number of breakings, operating time, energy storage state, opening and closing position, and contact resistance; The inspection data includes the inspection data of the appearance status, operating sound, and abnormal smell of substation equipment collected by intelligent inspection equipment or manual inspection methods, and also includes the data recording the abnormal situations found and the treatment measures during the inspection process; The maintenance records of the device include the maintenance time, maintenance content, and replaced parts; ensure the accuracy and integrity of the maintenance records for subsequent analysis.

[0014] Preferably, in the big data storage platform, data cleaning and preprocessing technologies are used to remove noise data, outliers, and duplicate data; interpolation filling methods are used for missing data. The data is standardized to have a unified dimension and unit, facilitating subsequent analysis and comparison.

[0015] Preferably, the FP-Growth algorithm is used to generate data associations as follows: Set: Data classification: Equipment operation history data S, environmental history data H, inspection history data X, and maintenance history record W; Wherein: The equipment operation history data S includes operation data S1, S2, S3,..., Sn on each substation equipment; The environmental history data H includes environmental data H1, H2, H3,..., Hn on each substation equipment; The inspection history data X includes inspection data X1, X2, X3,..., Xn on each substation equipment; The maintenance history record W includes maintenance data W1, W2, W3,..., Wn on each substation equipment; Integrate the classified data in the equipment operation history data S, environmental history data H, inspection history data X, and maintenance history record W to form a transaction dataset. In the transaction dataset, the classified data of the equipment operation history data S, environmental history data H, inspection history data X, and maintenance history record W form individual transactions; Perform item formation processing on each transaction in the transaction dataset, where each item is composed of different data classifications to form an item set; for example, an item may be represented as {S2, H3, X1, W4}, {S3, H2}, {X4, W5} Next, scan the transaction dataset once, count the occurrence frequency of each item, and sort all items in descending order of frequency; Create the root node of the FP-Tree, labeled "null"; Scan the transaction dataset again. For each transaction, sort the items in the transaction in descending order of item frequency. Then, add the items in the transaction to the FP-Tree one by one. If an item already exists on the path, increase the count of that item; if it does not exist, create a new branch; Mine frequent item sets from the FP-Tree. First, start from single items and find all frequent items whose support is greater than or equal to the minimum support threshold; For each frequent item, construct its conditional pattern base, that is, the prefix path of all paths containing that item; For each conditional pattern base, construct a conditional FP-Tree and recursively mine frequent item sets; For the mined frequent item sets, generate association rules; the form of the association rule is A→B, where A and B are subsets of the item sets; Calculate the support and confidence of each association rule; the support is the proportion of the number of transactions containing all items in the association rule to the total number of transactions; the confidence is the proportion of the number of transactions containing all items in the association rule to the number of transactions containing the antecedent A; According to the set minimum support threshold and minimum confidence threshold, filter out the required association rules; By analyzing the generated association rules, find the potential associations between devices; For example, if it is found that the association rule {S3, H2}→{X4, W5} has a high support and confidence, it means that when S3 in the device operation data and H2 in the environmental data appear simultaneously, X4 in the inspection data and W5 in the maintenance records are likely to appear; When a certain type of transformer fails, the transaction data related to the transformer can be analyzed to find out which parameters of other transformers installed in the same batch as it and the devices connected to it often appear abnormally at the same time. For example, if it is found that when a certain type of transformer fails, the association rule {S6 (specific operation data of this type of transformer)}→{S7 (specific operation data of other transformers in the same batch), H5 (specific environmental data), X6 (specific inspection data), W7 (specific maintenance record data)} frequently appears, it can be inferred that there are potential associations between these parameters, and when this type of transformer fails, there may be potential risks for transformers in the same batch and related devices.

[0016] Preferably, when using the FP-Growth algorithm to generate data associations, utilize a parallel computing framework, Hadoop MapReduce or Spark, to parallelize the execution of the FP-Growth algorithm, divide the data into multiple partitions, perform frequent item set mining simultaneously on different computing nodes, and finally merge the results. This can greatly improve the running speed of the algorithm, especially for large-scale data sets.

[0017] Preferably, when using the FP-Growth algorithm to generate data associations, for the construction of the FP-Tree and the process of mining frequent item sets, adopt multi-threading technology to parallelize different tasks on a single computing node to improve the efficiency of the algorithm.

[0018] Preferably, when the FP-Growth algorithm is used to generate data association, the algorithm operation process is divided into the initial stage, the intermediate stage, and the in-depth stage. The minimum support threshold of the initial stage is set to I, the minimum support threshold of the intermediate stage is set to O, and the minimum support threshold of the in-depth stage is set to P, where I>O>P. The purpose of this design is to set a relatively high minimum support at the beginning and quickly find some more obvious frequent item sets. As the mining deepens, more potential association rules are mined by reducing the minimum support.

[0019] A terminal includes a processor and a storage medium; the storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the above-mentioned substation operation and maintenance health analysis method.

[0020] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned substation operation and maintenance health analysis method.

[0021] In the present invention, the proposed substation operation and maintenance health analysis method has the following beneficial technical effects: 1. This application generates data associations and associations between devices by using the FP-Growth algorithm; a classification algorithm is used to establish a device failure model. If there is a risk of failure in the equipment, a warning signal is issued, and the device association generated by the FP-Growth algorithm for the device is used to generate associations between devices according to the FP-Growth algorithm, so as to obtain devices that may have potential risks associated with the current device, thereby predicting the impact range of the device failure. Compared with the digital twin method, this application mainly discovers the correlation between devices by analyzing historical data, and establishes a device failure model by using a classification algorithm, so as to predict the impact range of the equipment failure. Compared with the digital twin method, it is better at processing large-scale data, and can dig out the laws and trends hidden in the data. It is suitable for the overall risk assessment and trend analysis of large-scale substation networks, and mainly relies on existing data acquisition equipment and data analysis technology, does not require the construction of complex virtual models, and has relatively low costs.

[0022] 2. The FP-Growth algorithm is used to generate device associations, which can find potential associations between devices. The parallel computing framework and multi-threading technology greatly improve the running speed and efficiency of the FP-Growth algorithm, enabling it to cope with the analysis needs of large-scale data sets. The FP-Growth algorithm operation process is divided into different stages and the minimum support threshold is adjusted, which can find more comprehensive potential association rules at different mining depths, improving the accuracy and comprehensiveness of association mining.

[0023] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. Description of the Drawings

[0024] Figure 1 It is a flowchart of the method of the present invention; Figure 2 It is a schematic flow diagram of the present invention using the FP-Growth algorithm. Detailed Embodiments

[0025] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the drawings, where the same or similar symbols represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0026] As Figure 1 - Figure 2 shown, a substation operation and maintenance health analysis method, a substation operation and maintenance health analysis method, includes the following steps: Adopt a real-time data transmission protocol to transmit the device operation data, environmental data, inspection data, and maintenance records collected by sensors to a big data storage platform; The big data storage platform performs data cleaning and preprocessing on the received data, and then stores the data; The big data storage platform uses the FP-Growth algorithm to generate device associations for the data after data cleaning and preprocessing, and finds the associations between devices; For example, when a transformer of a certain model fails, which parameters of other transformers installed in the same batch and the devices connected to it often appear abnormal at the same time, and infer the potential risks that may exist in the transformers of the same batch and the associated devices.

[0027] Establish a device fault model. Through learning a large amount of historical fault data, use a classification algorithm to judge whether the device has a fault risk. If the device has a fault risk, send a warning signal, and predict the impact range of the device fault according to the device associations generated by the FP-Growth algorithm for the device.

[0028] This application generates data associations and associations between devices by using the FP-Growth algorithm; a classification algorithm is used to establish a device failure model. If there is a risk of failure in the equipment, a warning signal is issued, and the device association generated by the FP-Growth algorithm for the device is used to generate associations between devices, so as to obtain devices that may have potential risks associated with the current device, thereby predicting the impact range of the equipment failure. Compared with the digital twin method, this application mainly discovers the correlation between devices by analyzing historical data, and establishes a device failure model by using a classification algorithm, so as to predict the impact range of the equipment failure. Compared with the digital twin method, it is better at processing large-scale data, and can dig out the laws and trends hidden in the data. It is suitable for the overall risk assessment and trend analysis of large-scale substation networks, and mainly relies on existing data acquisition equipment and data analysis technology, does not require the construction of complex virtual models, and has relatively low costs.

[0029] The early warning signal is conveyed to operation and maintenance personnel via SMS, email, and pop-up windows on the monitoring system interface.

[0030] In the big data storage platform, distributed file systems and databases (such as Hadoop and HBase) are used for data storage to ensure data security and scalability.

[0031] In an optional embodiment, in establishing the equipment fault model, the classification algorithm used may be a decision tree algorithm, a random forest algorithm, or a support vector machine algorithm.

[0032] In an optional embodiment, the data collected by the sensor includes parameter data of oil temperature, oil level, winding temperature, voltage, current, power, number of breaking times, action time, energy storage state, opening and closing positions, and contact resistance; Inspection data includes inspection data on the appearance, operation sound, and odor of substation equipment collected by intelligent inspection equipment or manual inspection methods, as well as data recording abnormal conditions and treatment measures found during the inspection process; Equipment maintenance records, including maintenance time, maintenance content, and replaced parts; ensure the accuracy and completeness of maintenance records for subsequent analysis.

[0033] In an optional embodiment, in the big data storage platform, data cleaning and preprocessing techniques are used to remove noise data, outliers and duplicate data; interpolation filling methods are used for missing data. Data is standardized to have a unified dimension and unit to facilitate subsequent analysis and comparison.

[0034] In an optional embodiment, the FP-Growth algorithm is used to generate data association as follows: Settings: Data classification: Equipment operation historical data S, environmental historical data H, inspection historical data X, and maintenance historical records W; Among them: The equipment operation historical data S includes the operation data S1, S2, S3,..., Sn on each substation equipment; The environmental historical data H includes the environmental data H1, H2, H3,..., Hn on each substation equipment; The inspection historical data X includes the inspection data X1, X2, X3,..., Xn on each substation equipment; The maintenance historical records W includes the maintenance data W1, W2, W3,..., Wn on each substation equipment; Integrate the classified data in the equipment operation historical data S, environmental historical data H, inspection historical data X, and maintenance historical records W to form a transaction dataset. In the transaction dataset, the classified data of the equipment operation historical data S, environmental historical data H, inspection historical data X, and maintenance historical records W form individual transactions; Perform itemset processing on each transaction in the transaction dataset. Each item is composed of different data classifications to form an itemset; for example, an item may be represented as {S2, H3, X1, W4}, {S3, H2}, {X4, W5} Next, scan the transaction dataset once, count the occurrence frequency of each item, and sort all items in descending order of frequency; Create the root node of the FP-Tree, labeled "null"; Scan the transaction dataset again. For each transaction, sort the items in the transaction in descending order of item frequency. Then, add the items in the transaction to the FP-Tree one by one. If an item already exists on the path, increase the count of that item; if it does not exist, create a new branch; Mine frequent itemsets from the FP-Tree. First, start with single items and find all frequent items whose support is greater than or equal to the minimum support threshold; For each frequent item, construct its conditional pattern base, that is, the prefix path of all paths containing that item; For each conditional pattern base, construct a conditional FP-Tree and recursively mine frequent itemsets; For the mined frequent itemsets, generate association rules; the form of the association rule is A→B, where A and B are subsets of the itemset; Calculate the support and confidence of each association rule; the support is the ratio of the number of transactions containing all items in the association rule to the total number of transactions; the confidence is the ratio of the number of transactions containing all items in the association rule to the number of transactions containing the antecedent A. Filter out the required association rules according to the set minimum support threshold and minimum confidence threshold. Find out the potential associations between devices by analyzing the generated association rules. For example, if it is found that the association rule {S3, H2} → {X4, W5} has a high support and confidence, it means that when S3 in the device operation data and H2 in the environmental data appear simultaneously, X4 in the inspection data and W5 in the maintenance records are very likely to appear. When a certain type of transformer fails, the transaction data related to the transformer can be analyzed to find out which parameters of other transformers installed in the same batch and the devices connected to it often appear abnormally at the same time. For example, if it is found that when a certain type of transformer fails, the association rule {S6 (specific operation data of this type of transformer)} → {S7 (specific operation data of other transformers in the same batch), H5 (specific environmental data), X6 (specific inspection data), W7 (specific maintenance record data)} frequently appears, it can be inferred that there are potential associations between these parameters, and when this type of transformer fails, there may be potential risks for transformers in the same batch and related devices.

[0035] In an optional embodiment, when using the FP-Growth algorithm to generate data associations, the FP-Growth algorithm is parallelized using the parallel computing framework Hadoop MapReduce or Spark. The data is divided into multiple partitions, and frequent itemset mining is performed simultaneously on different computing nodes, and finally the results are merged. This can greatly improve the running speed of the algorithm, especially for large-scale datasets.

[0036] Both Hadoop MapReduce and Spark here are common parallel computing frameworks, mainly used for processing large-scale data.

[0037] In an optional embodiment, when using the FP-Growth algorithm to generate data associations, for the construction of the FP-Tree and the process of frequent itemset mining, multi-threading technology is used to parallelize different tasks on a single computing node to improve the efficiency of the algorithm.

[0038] In an optional embodiment, when using the FP-Growth algorithm to generate data associations, the algorithm running process is divided into an initial stage, an intermediate stage, and a deep stage. Let the minimum support threshold in the initial stage be I, the minimum support threshold in the intermediate stage be O, and the minimum support threshold in the deep stage be P, where I > O > P. The purpose of this design is to set a relatively high minimum support at the beginning to quickly find some obvious frequent item sets. As the mining progresses, by reducing the minimum support, more potential association rules can be mined.

[0039] Using the FP-Growth algorithm to generate device associations can find potential associations between devices. Through the parallel computing framework and multi-threaded technology, the running speed and efficiency of the FP-Growth algorithm are greatly improved, enabling it to meet the analysis requirements of large-scale data sets. Dividing the running process of the FP-Growth algorithm into different stages and adjusting the minimum support threshold can find more comprehensive potential association rules at different mining depths, improving the accuracy and comprehensiveness of association mining.

[0040] A terminal includes a processor and a storage medium; the storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the above-mentioned substation operation and maintenance health analysis method.

[0041] A computer-readable storage medium stores a computer program, and when the program is executed by a processor, it implements the steps of the above-mentioned substation operation and maintenance health analysis method.

[0042] The specific differences between using this application and the digital twin method are as follows: I. Data sources and processing methods This application: The data sources are mainly historical and real-time data such as device monitoring data, environmental data, inspection data, and maintenance records during actual operation. Data processing focuses on data cleaning, integration, and using machine learning and data mining algorithms to analyze large amounts of data to discover patterns and correlations. For example, potential connections between different device parameters are found through association rule mining. There are certain requirements for the accuracy and timeliness of the data, but mainly rely on the acquisition accuracy of devices such as sensors and the timeliness of data transmission.

[0043] Digital Twin Method: The data sources include not only the monitoring data of actual physical devices but also the simulation data of virtual devices in the Digital Twin model. More attention is paid to the interaction between the virtual and the physical in data processing. By feeding the data of physical devices back to the Digital Twin model in real time, the model parameters are continuously adjusted and optimized. At the same time, the model is used for simulation and prediction. For example, simulating device failures in the Digital Twin model and analyzing their impacts on the entire system. Higher requirements are placed on the accuracy and synchronization of data to ensure a high degree of consistency between the Digital Twin model and the actual physical system.

[0044] II. Analysis Methods and Capabilities This Application: Mainly discovers the correlations and failure modes between devices by analyzing historical data for prediction. For example, based on the historical failure data of a certain type of transformer, inferring the potential risks of transformers and related devices in the same batch. Good at processing large-scale data, capable of mining the rules and trends hidden in the data, and the analysis results are mainly based on statistical analysis and probability prediction.

[0045] Digital Twin Method: Based on the Digital Twin model, physical simulation and multi-physics field analysis can be carried out, capable of simulating the operating status and failure conditions of devices. Real-time monitoring and prediction can be achieved, with strong dynamic adaptability, and the model can be adjusted in a timely manner according to the changes in the actual system.

[0046] III. Application Scenarios and Advantages This Application: Suitable for the overall risk assessment and trend analysis of large-scale substation networks. For example, by analyzing the historical data of multiple substations, the common failure modes of a certain type of device can be discovered, providing a reference for equipment selection and maintenance strategies. The cost is relatively low, mainly relying on existing data acquisition devices and data analysis technologies, without the need to build complex virtual models. It can quickly process a large amount of data, providing timely warning information for operation and maintenance personnel, helping to arrange maintenance plans in advance and reducing the incidence of equipment failures.

[0047] Digital Twin Method: Particularly suitable for the refined operation and maintenance management of key devices or complex systems. For example, for the core transformer of a large substation, the Digital Twin model can provide more accurate fault diagnosis and prediction, helping operation and maintenance personnel to develop personalized maintenance plans.

[0048] The following is an introduction to the FP-Growth algorithm: The FP-Growth (Frequent Pattern Growth) algorithm is an efficient algorithm for discovering frequent item sets in a dataset. The FP-Growth algorithm is based on the frequent pattern tree (FP-tree) data structure. Its main idea is to compress the dataset into a frequent pattern tree and then discover frequent item sets by mining this tree, without the need to scan the dataset multiple times like traditional association rule mining algorithms.

[0049] 1. Construct the FP-tree: First, scan the dataset once to count the support (frequency of occurrence) of each item. Then, delete the items with support below the set minimum support threshold. Next, sort the remaining items in descending order of support. Finally, scan the dataset again and construct the FP-tree for each transaction according to the sorted item order.

[0050] 2. Mine frequent item sets: Starting from a single path in the FP-tree, recursively construct the conditional pattern base (including the prefix paths and corresponding supports on all paths ending with a certain item). Construct a conditional FP-tree for each conditional pattern base. Repeat this process until the conditional FP-tree is empty or contains only a single path.

[0051] In life, it is applied to discover the association relationships between the goods purchased by customers, such as which goods are often purchased together, so as to help merchants make decisions such as product placement and promotion.

[0052] Meanwhile, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0053] In the embodiments provided by the present invention, it should be understood that the disclosed system or method can be implemented in other ways. For example, the above-described invention embodiments are merely illustrative. For example, the division of modules is only a logical function division, and there can be other division methods in actual implementation.

[0054] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules. They can be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0055] In addition, in each embodiment of the present invention, the functional modules can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of a combination of hardware and software functional modules.

[0056] For those skilled in the operation and maintenance field, it is obvious that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the basic characteristics of the present invention.

[0057] As mentioned above, it is only the preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the technical field of the present invention within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent replacements or changes, should be covered by the protection scope of the present invention.

Claims

1. A substation operation and maintenance health analysis method, characterized in that: The following steps are involved: Use real-time data transmission protocol to transmit equipment operation data, environmental data, inspection data and maintenance records collected by sensors to the big data storage platform; The big data storage platform cleans and preprocesses the received data and then stores the data; The big data storage platform uses the FP-Growth algorithm to generate data associations and associations between devices after data cleaning and preprocessing; A classification algorithm is used to establish an equipment failure model. By learning from a large amount of historical failure data, it is determined whether the equipment has a failure risk. If the equipment has a failure risk, an early warning signal is issued, and the impact range of the equipment failure is predicted based on the equipment association generated by the FP-Growth algorithm.

2. The substation operation and maintenance health analysis method according to claim 1 is characterized in that: In establishing the equipment fault model, the classification algorithm used may be one or more of a decision tree algorithm, a random forest algorithm, and a support vector machine algorithm.

3. The substation operation and maintenance health analysis method according to claim 1 is characterized in that: The data collected by the sensor include parameter data of oil temperature, oil level, winding temperature, voltage, current, power, breaking times, action time, energy storage status, opening and closing positions, and contact resistance; Inspection data includes inspection data on the appearance, operation sound, and odor of substation equipment collected by intelligent inspection equipment or manual inspection methods, as well as data recording abnormal conditions and treatment measures found during the inspection process; Equipment maintenance records, including maintenance time, maintenance content, and replaced parts.

4. The substation operation and maintenance health analysis method according to claim 1 is characterized in that: In the big data storage platform, data cleaning and preprocessing techniques are used to remove noise data, outliers and duplicate data; interpolation filling methods are used to fill in missing data.

5. The substation operation and maintenance health analysis method according to claim 1 is characterized in that: The FP-Growth algorithm is used to generate data associations as follows: set up: Data classification: equipment operation history data S, environment history data H, inspection history data X and maintenance history data W; in: Equipment operation history data S contains the operation data of each substation equipment S1, S2, S3, ..., Sn; Environmental history data H contains environmental data on each substation equipment H1, H2, H3, ..., Hn; Inspection history data X contains inspection data on each substation equipment X1, X2, X3, ..., Xn; Maintenance history records W contain maintenance data on each substation equipment W1, W2, W3, ..., Wn; The classified data in the equipment operation history data S, the environment history data H, the inspection history data X, and the maintenance history data W are integrated to form a transaction data set. In the transaction data set, the classified data in the equipment operation history data S, the environment history data H, the inspection history data X, and the maintenance history data W form transactions one by one. Each transaction in the transaction data set is processed into items, where each item is composed of different data classifications to form an item set; Next, scan the transaction data set once, count the frequency of occurrence of each item, and sort all items in descending order of frequency; Create the root node of FP-Tree and mark it as "null"; Scan the transaction data set again, and for each transaction, sort the items in the transaction in descending order according to the frequency of the items. Then, add the items in the transaction to the FP-Tree one by one. If an item already exists on the path, increase the count of the item; if not, create a new branch; Mining frequent itemsets from FP-Tree. First, starting from a single item, find all frequent items whose support is greater than or equal to the minimum support threshold; For each frequent item, construct its conditional pattern base, which is the prefix path of all paths containing the item; For each conditional pattern base, a conditional FP-Tree is constructed and frequent itemsets are mined recursively; For the mined frequent item sets, association rules are generated; the form of association rules is A→B, where A and B are subsets of the item sets; Calculate the support and confidence of each association rule; support is the ratio of the number of transactions containing all items in the association rule to the total number of transactions; confidence is the ratio of the number of transactions containing all items in the association rule to the number of transactions containing the antecedent A; According to the set minimum support threshold and minimum confidence threshold, the required association rules are screened out; By analyzing the generated association rules, potential associations between devices can be found.

6. The substation operation and maintenance health analysis method according to claim 5 is characterized in that: When the FP-Growth algorithm is used to generate data associations, the parallel computing framework Hadoop MapReduce or Spark is used to parallelize the FP-Growth algorithm, divide the data into multiple partitions, perform frequent item set mining on different computing nodes simultaneously, and finally merge the results.

7. The substation operation and maintenance health analysis method according to claim 6 is characterized in that: When the FP-Growth algorithm is used to generate data associations, multi-threading technology is used for the construction of FP-Tree and the frequent itemset mining process to execute different tasks in parallel on a single computing node.

8. The substation operation and maintenance health analysis method according to claim 7 is characterized in that: When the FP-Growth algorithm is used to generate data association, the algorithm operation process is divided into the initial stage, the intermediate stage, and the in-depth stage. The minimum support threshold of the initial stage is set to I, the minimum support threshold of the intermediate stage is set to O, and the minimum support threshold of the in-depth stage is set to P, where I>O>P.

9. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the substation operation and maintenance health analysis method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the substation operation and maintenance health analysis method described in any one of claims 1-8 are implemented.

Images