Substation equipment management platform combined with multi-source data fusion modeling

Abstract

The invention discloses a substation equipment management platform combining multi-source data fusion modeling, and relates to the technical field of substation management, and the platform comprises an operation data set obtaining module which is used for obtaining a multi-source real-time operation data set; the operation characteristic determination module is used for determining a plurality of operation characteristics; the twinborn model construction module is used for constructing a three-dimensional digital twinborn model; the running state information acquisition module is used for acquiring running state information; the operation and maintenance analysis module is used for carrying out operation and maintenance analysis and formulating equipment management suggestions; and the management control module is used for performing intelligent management control on the equipment management suggestions. The technical problems of single data dimension, state evaluation lagging and operation and maintenance strategy staticization in traditional substation equipment management in the prior art are solved, dynamic feedback optimization is performed through multi-source real-time data acquisition and digital twin modeling, accurate evaluation of the substation equipment operation state is realized, and the operation and maintenance efficiency is improved. And the equipment management efficiency is obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of substation management, and specifically relates to a substation equipment management platform combined with multi-source data fusion modeling. Background Art

[0002] In the power system, as a key hub connecting power generation and power consumption, the stable operation of substation equipment is of crucial importance. However, current substation equipment management faces many challenges. On the one hand, traditional data collection methods are relatively single, mostly relying on limited sensors to obtain a few key parameters, such as only focusing on electrical quantities like voltage and current, and it is difficult to comprehensively capture the operating state of equipment. This one-sided data collection makes the understanding of equipment operation conditions not deep enough, and many potential fault hazards cannot be detected in time. On the other hand, in terms of operation feature analysis, existing technologies lack efficient data fusion and in-depth analysis capabilities. For multi-source data from different types of sensors, it is difficult to effectively integrate and mine, resulting in only simple statistical analysis and unable to accurately extract the key features of equipment operation. This makes the early warning of equipment faults not timely and accurate enough, increasing the risk of sudden equipment failures. In addition, maintenance work mainly relies on manual experience to formulate maintenance strategies. This method is not only time-consuming and laborious, but also prone to decision-making mistakes due to individual experience differences. Facing the complex and changeable equipment operation conditions, manual troubleshooting and handling of faults are inefficient, and the equipment downtime is long, seriously affecting the stability and reliability of power supply.

[0003] The prior art has technical problems of single data dimension, lagging state assessment, and static maintenance strategies in traditional substation equipment management. Summary of the Invention

[0004] This application provides a substation equipment management platform combined with multi-source data fusion modeling, which is used to solve the technical problems of single data dimension, lagging state assessment, and static maintenance strategies in traditional substation equipment management in the prior art.

[0005] In view of the above problems, this application provides a substation equipment management platform combined with multi-source data fusion modeling.

[0006] This application provides a substation equipment management platform combined with multi-source data fusion modeling, and the platform includes: An operating dataset acquisition module for real-time data collection through multiple data sources to obtain a multi-source real-time operating dataset of multiple substation devices in a target area; an operating feature determination module for fusing the multi-source real-time operating dataset to generate a fused dataset, performing feature analysis based on the fused dataset to determine multiple operating features; a digital twin model construction module for three-dimensional modeling of multiple substation devices based on Gaussian modeling technology to construct three-dimensional digital twin models of multiple substation devices; an operating status information acquisition module for synchronizing the fused dataset to the three-dimensional digital twin models to evaluate the operation of multiple substation devices and obtain device operating status information; an operation and maintenance analysis module for traversing the device operating status information and combining the multiple operating features to perform operation and maintenance analysis on multiple substation devices and formulate device management suggestions; a management control module for performing simulation operation and maintenance based on the device management suggestions to generate simulation operation and maintenance results, dynamically feedback and optimize the device management suggestions according to the simulation operation and maintenance results to generate a device management plan, and executing the device management plan to perform intelligent management and control on substation devices.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: An operating dataset acquisition module for real-time data collection through multiple data sources to obtain a multi-source real-time operating dataset; an operating feature determination module for fusing the multi-source real-time operating dataset to generate a fused dataset, performing feature analysis to determine multiple operating features; a digital twin model construction module for three-dimensional modeling of multiple substation devices to construct three-dimensional digital twin models of multiple substation devices; an operating status information acquisition module for evaluating the operation of multiple substation devices to obtain device operating status information; an operation and maintenance analysis module for traversing the device operating status information and combining the multiple operating features to perform operation and maintenance analysis on multiple substation devices and formulate device management suggestions; a management control module for performing simulation operation and maintenance based on the device management suggestions to generate simulation operation and maintenance results, dynamically feedback and optimize according to the simulation operation and maintenance results to generate a device management plan, and perform intelligent management and control. It achieves the technical effect of accurately evaluating the operating status of substation devices through multi-source real-time data collection and digital twin modeling for dynamic feedback optimization, and significantly improving the device management efficiency. Brief Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings without creative efforts based on these drawings.

[0009] Figure 1 Schematic diagram of the substation equipment management platform combining multi-source data fusion modeling provided by the embodiment of the present application; Figure 2 Flow chart of the operating state information acquisition module in the substation equipment management platform combining multi-source data fusion modeling provided by the embodiment of the present application.

[0010] Explanation of reference numerals: Operating data set acquisition module 10, operating feature determination module 20, twin model construction module 30, operating state information acquisition module 40, operation and maintenance analysis module 50, management and control module 50. Detailed implementation manners

[0011] The present application provides a substation equipment management platform combining multi-source data fusion modeling to solve the technical problems of single data dimension, lagging state evaluation, and static operation and maintenance strategies in traditional substation equipment management in the prior art.

[0012] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0013] Embodiment, as Figure 1 shown, the present application provides a substation equipment management platform combining multi-source data fusion modeling, and the platform includes: An operating data set acquisition module 10, configured to perform real-time data acquisition through multiple data sources to obtain a multi-source real-time operating data set of multiple substation equipment in a target area.

[0014] Specifically, the operation dataset acquisition module 10 collects multi-source real-time operation datasets of multiple substation devices in the target area. At the hardware level, high-performance data acquisition cards and multi-port network adapters are equipped to ensure the ability to connect multiple data sources simultaneously. For sensor data sources, industrial-grade wired or wireless sensor network technologies, such as ZigBee, LoRa, etc., are adopted to establish reliable communication links with various sensors to obtain analog and digital quantity data such as temperature, pressure, vibration, etc., and the A / D conversion module is used to convert analog signals into digital signals. In terms of software, a special data acquisition program is developed, which supports multiple communication protocols, such as Modbus, OPC UA, etc., for data interaction with the automation monitoring system and other intelligent devices. According to the characteristics of different data sources, corresponding data parsing rules and caching mechanisms are set. For data sources with high real-time requirements, such as online monitoring sensors, real-time data push technology is adopted to ensure that data can be quickly transmitted to the module; while for relatively static data sources such as historical maintenance records, through regular data pulling and updating operations, the timeliness of the data is guaranteed. At the same time, in order to ensure the accuracy and integrity of the data, data verification and error correction functions are integrated, and the cyclic redundancy check (CRC) algorithm is used to verify the collected data. Once incorrect data is found, it is immediately processed through data retransmission or error repair algorithms, thus successfully obtaining high-quality multi-source real-time operation datasets.

[0015] The operation feature determination module 20 is used to fuse the multi-source real-time operation datasets to generate a fused dataset, and perform feature analysis based on the fused dataset to determine multiple operation features.

[0016] Specifically, after receiving the multi-source real-time operation dataset from the operation dataset acquisition module, the operation feature determination module 20 first uses an advanced time series analysis algorithm to accurately time-stamp each data point in the dataset, thereby determining the operation timestamp, establishing an accurate time axis for the data, and ensuring the orderliness of subsequent processing. Based on these timestamps, the multi-source data is sorted in order through an alignment algorithm to generate an operation time series data column. On this basis, the autoencoder technology in deep learning is used to reconstruct the data column. The autoencoder encodes and fuses the multi-source real-time operation dataset into a more representative fusion dataset by learning the internal structure and features of the data. A comprehensive feature analysis is carried out on the fusion dataset. Multivariate statistical analysis methods such as principal component analysis (PCA) and singular value decomposition (SVD) are used to reduce the dimension and extract features of the data, mining the key information hidden in the data. At the same time, combined with spectrum analysis technology, the data is analyzed from the frequency domain perspective to obtain operation features such as frequency components and harmonic characteristics. The association rule mining algorithm is also used to analyze the correlation between different variables, further determining multiple operation features that can reflect the operation state of substation equipment, providing a solid basis for subsequent equipment evaluation and operation and maintenance decisions.

[0017] The twin model construction module 30 is used to perform three-dimensional modeling on multiple substation equipment based on Gaussian modeling technology to construct three-dimensional digital twin models of multiple substation equipment.

[0018] Specifically, the twin model construction module 30 comprehensively sorts out various detailed information of substation equipment, including multi-dimensional data such as the physical structure, mechanical parameters, and electrical characteristics of the equipment. Some of these data come from the design documents of the equipment, and some are historical data accumulated during long-term operation. Based on the principle of Gaussian modeling technology, the geometric shape, spatial position relationship, etc. of the equipment are abstracted into the form of Gaussian distribution. Taking a large transformer in a substation as an example, the complex shapes of its key components such as the iron core and windings are decomposed into multiple sub-structures based on Gaussian functions. By accurately calculating the parameters of these sub-structures, such as the mean and covariance, to determine their positions and forms in three-dimensional space, and then gradually constructing the three-dimensional model of the entire transformer. For other equipment such as circuit breakers and instrument transformers, they are also modeled according to their respective structural characteristics and operating principles. During the modeling process, the model is continuously optimized and calibrated by combining actual operation data to ensure that the model can accurately reflect the state changes of the equipment under different working conditions. Finally, three-dimensional digital twin models of multiple substation equipment are successfully constructed. These models not only provide an intuitive interface for the visual management of the equipment, but also lay a solid foundation for subsequent operation evaluation, fault prediction and other links.

[0019] The operating status information acquisition module 40 is used to synchronize the fusion dataset to the three-dimensional digital twin model to evaluate the operation of multiple substation devices and obtain the device operating status information.

[0020] Specifically, the operating status information acquisition module 40 first synchronizes the fusion dataset to the three-dimensional digital twin model completely and accurately through a specially designed data interface and an efficient data transmission protocol. During the synchronization process, data verification and error correction processing will be performed to ensure the reliability of the data. Once the data is successfully imported into the model, a comprehensive evaluation will be carried out for each substation device based on the key operation indicators. These key indicators include electrical parameter indicators, such as the stability of voltage and current, power factor, etc.; mechanical performance indicators, like the vibration amplitude and rotational speed stability of the device; and temperature-related indicators, including the temperature range of key parts of the device. For electrical parameters, professional electrical analysis algorithms are used to calculate the deviation rate between the actual value and the standard value, and a quantitative score is given according to the degree of deviation. In terms of mechanical performance, the health status of the device's mechanical structure is judged by performing spectral analysis on the data collected by vibration sensors, and the corresponding evaluation score is given. For temperature indicators, based on the device's heat dissipation model and material thermal characteristics, it is analyzed whether the temperature is within the normal range to determine the score for this item. By synthesizing the evaluation results in these different aspects, detailed and accurate operating status information of each substation device is finally generated, clearly reflecting the operation quality of the device and providing a key basis for subsequent operation and maintenance decisions.

[0021] The operation and maintenance analysis module 50 is used to traverse the device operating status information and combine the multiple operating characteristics to perform operation and maintenance analysis on multiple substation devices and formulate device management suggestions.

[0022] Specifically, the operation and maintenance analysis module 50 first, in the data integration and preliminary analysis stage, uses the principal component analysis (PCA) algorithm to reduce the dimension of the device operation status information and operation characteristic data, extract key information, remove redundant parts in the data, and improve the efficiency and accuracy of subsequent analysis. For mining the internal correlation of data, the Apriori algorithm is used for association rule mining. By setting appropriate support and confidence thresholds, strong association rules between the device operation status parameters and operation characteristics are found. For example, it is found that when the temperature of a certain component of the device continuously exceeds a specific threshold and the operation duration reaches a certain period, there is a high correlation with the failure risk of this component. In terms of device grouping, the K-Means clustering algorithm is used to divide multiple substation devices into different clusters according to the similarity of the devices in the operation status and feature space. For each cluster, a classification algorithm based on decision trees is adopted. According to historical data and expert experience, a decision tree model is constructed, and corresponding device management suggestions are generated according to the characteristics of the devices in different clusters. For example, for the devices in a certain cluster with a high failure risk, the decision tree model will recommend priority in-depth detection and preventive maintenance, thus providing scientific and accurate guidance for the operation and maintenance management of substation devices.

[0023] The management and control module 60 is used to perform simulation operation and maintenance based on the device management suggestions, generate simulation operation and maintenance results, dynamically feedback and optimize the device management suggestions according to the simulation operation and maintenance results, generate a device management plan, and execute the device management plan to perform intelligent management and control on the substation devices.

[0024] Specifically, in the simulation operation and maintenance stage, the management and control module 60 uses the Monte Carlo simulation algorithm to perform a large number of simulations on the operation of the device under different random working conditions, sets different input parameters according to the device management suggestions, such as the device maintenance time interval, spare part replacement strategy, etc., and obtains the probability distribution of the device performance indicators through multiple simulation runs, so as to generate simulation operation and maintenance results. When comparing and analyzing the simulation operation and maintenance results with the expected goals, the mean square error (MSE) algorithm is used to calculate the deviation degree between the two, and the effectiveness of the device management suggestions is judged according to the deviation size. If the deviation exceeds the set threshold, the optimization process based on the genetic algorithm is started. In the genetic algorithm, the key parameters in the device management suggestions are encoded as chromosomes, and better parameter combinations are continuously searched through selection, crossover, and mutation operations to reduce the MSE value. After multiple rounds of iterative optimization, the optimal device management plan is determined, and real-time control algorithms such as the proportional-integral-differential (PID) algorithm are used to accurately adjust the operation parameters of the device according to the feedback information of the device operation status, realize the intelligent management and control of the substation devices, and ensure that the device operation performance reaches the best state.

[0025] In a possible implementation manner, the operation characteristic determination module 20 further includes: A running timestamp determination unit for performing running time series analysis on multiple substation devices according to the multiple data sources to determine the running timestamps.

[0026] A running time series data column generation unit for aligning the multi-source real-time running data set according to the running timestamps to generate a running time series data column.

[0027] An autoencoder construction unit for traversing the running time series data column to reconstruct the multi-source real-time running data set and construct an autoencoder.

[0028] An encoding fusion unit for encoding and fusing the multi-source real-time running data set through the autoencoder to generate the fusion data set.

[0029] Specifically, for the massive data of multiple substation devices provided by multiple data sources, a time series analysis algorithm is adopted. By identifying the time marker information in the data or based on the sequence law of data generation, the running timestamp corresponding to each data point is accurately determined, so that it is arranged in an orderly manner in the time dimension.

[0030] Based on the determined running timestamps, next, the multi-source real-time running data set is aligned to generate a running time series data column. Taking a certain data source as a reference, usually a data source with higher data integrity and accuracy is selected. Using the timestamp sequence of the reference data source as a clue, the matching timestamps are searched in other data sources. For the exactly matching timestamps, the corresponding data is arranged in sequence according to the preset rules (such as data source importance, data type, etc.) and inserted into the corresponding position of the running time series data column. If there are slight differences in timestamps but within the acceptable error range, linear interpolation is used for data fitting. Through calculation and verification, and comparison with the preset value, each data is verified. If the verification fails, it indicates that the data may be incorrect or missing, and it is processed according to the predetermined strategy, such as replacing it with adjacent correct data or issuing an exception alarm for manual processing, and finally a qualified running time series data column is generated.

[0031] Traverse the running time series data column, and for the data at each time point, extract the corresponding information from multiple data sources to form feature vectors. For example, extract the corresponding values from data sources such as voltage, current, and power to form feature vectors with multiple dimensions. These feature vectors will be used as the input data for the autoencoder. When constructing the autoencoder, first build the encoder part. The encoder consists of multiple layers of neural networks. The number of neurons in the input layer is the same as the dimension of the feature vector and is responsible for receiving the input data. As the network depth increases, the number of neurons gradually decreases to achieve data compression and extract key features. For example, use a fully connected layer, multiply the input data by the weight matrix and add the bias term, and then process it through an activation function (such as the ReLU function) to obtain the compressed feature representation. The decoder corresponds to the encoder and is also composed of multiple layers of neural networks. The number of neurons in its input layer is the same as the dimension of the feature representation output by the encoder, and the number of neurons in the output layer is the same as the dimension of the original input feature vector. The decoder gradually restores the compressed features output by the encoder to a data form similar to the original input through reverse operations. Similarly, use fully connected layers, weight matrices, bias terms, and activation functions (such as linear activation functions) to try to reconstruct the original feature vector. During the whole process, by continuously adjusting the parameters (weights and biases) of the encoder and decoder, minimize the difference between the reconstructed data and the original input data. This difference is usually measured by loss functions such as mean squared error. Use optimization algorithms such as gradient descent, and based on the feedback of the loss function, iteratively update the parameters so that the autoencoder can learn the internal structure and features of the data, thereby completing the reconstruction of the multi-source real-time running data set and the construction of the autoencoder.

[0032] After the construction of the autoencoder is completed, the autoencoder is used to perform encoding fusion operations on the multi-source real-time operating dataset, thereby generating a fused dataset. Each data sample in the multi-source real-time operating dataset is sequentially input into the encoder part of the constructed autoencoder. The encoder performs complex non-linear transformations on the input data according to the weights and biases obtained during its training, compressing the high-dimensional, multi-source raw data into low-dimensional feature vectors, which contain the key information of the raw data. Subsequently, these encoded feature vectors are passed inside the autoencoder to the decoder part. However, here it is not to decode and restore them to the form of the original data, but to use the feature vectors output by the encoder for fusion. The specific method is to merge the feature vectors obtained by encoding different data sources according to certain rules. For example, simply concatenating the corresponding dimension feature values, or by weighted summation, fusing multiple feature vectors into a unified feature vector. In this way, encoding fusion operations are performed on all data samples in the entire multi-source real-time operating dataset, and finally a series of fused feature vectors are obtained, which constitute the fused dataset. This fused dataset not only integrates the information of multi-source data, but also removes the redundant parts in the data through the encoding process of the autoencoder, making the data more representative and compact, providing a high-quality data basis for subsequent tasks such as device operation analysis and fault diagnosis based on this dataset.

[0033] In a possible implementation manner, the operating feature determination module 20 further includes: A multi-dimensional data matrix acquisition unit, configured to load the fused dataset for multi-dimensional data integration to obtain a multi-dimensional data matrix.

[0034] A multi-domain feature signal acquisition unit, configured to traverse the multi-dimensional data matrix to perform feature mining according to the operating time series data column to obtain multi-domain feature signals, where the multi-domain features include operating frequency domain feature signals and operating time domain feature signals.

[0035] A frequency domain component information determination unit, configured to decompose the operating time domain feature signal according to the operating frequency domain feature signal to determine a plurality of frequency domain component information.

[0036] A correlation analysis unit, configured to traverse the plurality of frequency domain component information to perform correlation analysis with multiple substation devices to obtain multiple device operating cycle features.

[0037] A device operating cycle feature addition unit, configured to add the multiple device operating cycle features to the multiple operating features.

[0038] Specifically, when loading the fused dataset for multi-dimensional data integration to obtain a multi-dimensional data matrix, first, use a file reading algorithm to read the fused dataset from the storage device into memory. Assume that the fused dataset is stored in tabular form, with each row representing a data sample and each column representing a feature from a different data source. Then, initialize an empty multi-dimensional array to store the integrated multi-dimensional data matrix. Next, iterate through each row of data in the fused dataset. For each feature value in a row, process it according to its dimension category. For example, a dimension mapping table can be predefined, which maps different feature names to corresponding dimension indices. Through the dimension mapping table, determine the position in the multi-dimensional array where each feature value should be placed. If a certain feature belongs to the electrical parameter dimension, and the index of this dimension in the multi-dimensional array is 0, and another feature belongs to the environmental parameter dimension with an index of 1, then store the electrical parameter feature value in the 0th column of the corresponding row in the multi-dimensional array, and store the environmental parameter feature value in the 1st column. When placing the feature values, the consistency and compatibility of data types also need to be considered. For numerical data, it can be directly stored; for non-numerical data, such as text labels, encoding conversion is required. For example, using a label encoding algorithm, convert it to a numerical form before storing it in the multi-dimensional array. After iterating through all the row data, through operations such as data classification, placement, and type conversion, this multi-dimensional array forms the required multi-dimensional data matrix, providing a structured data basis for subsequent data processing and analysis.

[0039] After obtaining the multi-dimensional data matrix, start traversing this matrix. Using the running time series data column as a clue, analyze each data point in the matrix in sequence. In the time domain, by observing the variation law of the data over time, and using statistical methods such as calculating the mean, variance, peak value, slope, etc., directly extract the running time domain feature signals from the data sequence. These feature signals can reflect the fluctuations and stability of the device operating state over time. In the frequency domain, use algorithms such as Fourier transform to transform the time domain data into the frequency domain space, analyze the transformed results, and extract information such as the amplitude and phase of different frequency components, thereby obtaining the running frequency domain feature signals. These signals reveal the energy distribution and characteristics of the device operation at different frequencies. Through the comprehensive traversal and analysis of the multi-dimensional data matrix, finally, multi-domain feature signals containing running frequency domain feature signals and running time domain feature signals are successfully obtained, providing a key basis for in-depth understanding of the operation state of substation equipment.

[0040] Based on the obtained running frequency-domain characteristic signals, they are used as the key basis for analyzing the running time-domain characteristic signals. The short-time Fourier transform is adopted, which can segment the running time-domain characteristic signals in the time domain with a certain window length, and then perform Fourier transform on each segment of the signal, thereby converting the time-domain signal into a time-frequency two-dimensional representation. In this two-dimensional representation, the amplitude and phase information corresponding to each frequency are the manifestations of this frequency component in the time-domain signal. By extracting and analyzing parameters such as the amplitude, phase, and frequency value of these frequency components, several frequency-domain component information are determined. These information details the composition structure of the running time-domain characteristic signals at different frequencies, providing fine frequency-dimensional data support for further understanding the operating characteristics of the equipment.

[0041] After obtaining several frequency-domain component information, start traversing this information and perform correlation analysis with multiple substation equipment. For each frequency-domain component information, first identify the main frequency characteristics with higher energy. Since the main frequency characteristics carry the main information of the signal and reflect the core laws of equipment operation, they are used as the key indicators to measure the periodic behavior of equipment operation. For each substation equipment, collect various operation data within the same time span, such as the fluctuation conditions of parameters such as voltage, current, and power. Then, use a dedicated correlation analysis algorithm to calculate the Pearson correlation coefficient to quantify the degree of tightness of the association between the main frequency characteristics and each operation parameter of the equipment. During the analysis process, pay special attention to those data change patterns that show significant correlation with the main frequency characteristics. For example, if it is found that the current value of a certain equipment shows a stable periodic correspondence with the change of the main frequency characteristics within a specific time interval, then the characteristics such as this periodic change time interval and amplitude change can be determined as part of the operation cycle characteristics of this equipment. Through such a detailed analysis of each frequency-domain component information with multiple substation equipment one by one, multiple equipment operation cycle characteristics are mined from a large number of data associations. These characteristics can effectively reveal the periodic laws in the equipment operation process and provide important basis for equipment status monitoring, fault prediction, etc.

[0042] After successfully obtaining multiple equipment operation cycle characteristics, integrate these characteristics into the existing multiple operation characteristics. Traverse each equipment operation cycle characteristic. For each characteristic, according to the pre-set rules or data structures, add it to the appropriate operation characteristic set. For example, if the operation characteristics are stored in the form of a dictionary, where different keys represent different types of operation characteristics, then add the equipment operation cycle characteristics to the list or data structure of the corresponding key-value pairs according to their attributes. In this way, all the equipment operation cycle characteristics are integrated into multiple operation characteristic systems one by one, further enriching and perfecting the content of the operation characteristics, and providing more complete data support for the subsequent comprehensive analysis and evaluation of the operation status of substation equipment.

[0043] In a possible implementation, as Figure 2 shown, the operating status information acquisition module 40 further includes: An operating evaluation index setting unit, configured to synchronize the fusion data set to the three-dimensional digital twin model, map the fusion data set to multiple device operating components through the three-dimensional digital twin model, and set multiple operating evaluation indexes.

[0044] An operating score generation unit, configured to perform an operating evaluation on multiple substation devices according to the multiple operating evaluation indexes, and generate multiple operating scores.

[0045] A status label acquisition unit, configured to divide multiple status levels, and perform matching identification according to the multiple operating scores and the multiple status levels to obtain multiple status labels.

[0046] A dynamic monitoring unit, configured to perform dynamic monitoring on multiple substation devices according to the multiple status labels, and generate the device operating status information.

[0047] Specifically, synchronize the fusion data set to the three-dimensional digital twin model. The three-dimensional digital twin model, as a virtual mapping of real substation devices, accurately simulates the appearance, structure, and operating logic of the devices. After data synchronization, map the fusion data set to multiple device operating components through this model to ensure that each component can obtain the operating data related to it. At the same time, set multiple operating evaluation indexes. These indexes are used to measure the operating conditions of the devices, covering multiple key dimensions such as device performance, stability, and energy consumption, providing a comprehensive and accurate standard for subsequent evaluation.

[0048] For multiple substation devices, perform an operating evaluation according to multiple operating evaluation indexes and generate operating scores. First, for each device, extract the data related to each index from the fusion data set. For example, for the device performance index, obtain the output power and efficiency data. The calculation method of the output power score is as follows: if the actual output power is lower than the ideal value, multiply the ratio of the actual value to the ideal value by a certain score value; if it is higher than the ideal value, in addition to the basic score, calculate the score for the excess part according to the ratio. The efficiency score is calculated by multiplying the ratio of the actual value to the ideal value by a certain score value, and the sum of the two is the device performance score. For the stability index, extract the data of the fluctuation frequency and amplitude of the operating parameters. The fluctuation frequency score is calculated by multiplying the ratio of the difference from the ideal value by the corresponding score value, and the same applies to the fluctuation amplitude. The sum of the two is the stability score. For the energy consumption index, obtain the energy consumption per unit time data. If the actual energy consumption is lower than the ideal value, multiply the ratio of the difference by the full score value; if it is higher than the ideal value, multiply the ratio of the difference from the maximum acceptable value by the full score value to obtain the energy consumption score. Finally, according to the pre-set weights of each index, perform a weighted sum of the scores of device performance, stability, energy consumption, etc. to obtain the comprehensive operating score of each device.

[0049] Based on industry standards for equipment operation, historical data, operation and maintenance experience, etc., multiple status levels are divided, such as five levels: excellent, good, qualified, warning, and failure. And corresponding operation score ranges are set for each level. For example, the score range of 90 - 100 is designated as the excellent level range, 75 - 89 as good, 60 - 74 as qualified, 40 - 59 as warning, and below 40 as failure. Then, for the operation score of each substation equipment, it is matched and labeled with the multiple pre - divided status levels. For example, if the operation score of a certain equipment is 85 points, after matching and labeling, the corresponding status label of this equipment is good. In this way, corresponding status labels are matched for multiple substation equipments.

[0050] Based on these status labels, dynamic monitoring is carried out on multiple substation equipments, continuously tracking the changes of the equipment status labels. Once the status label changes, this information is immediately captured. For example, if the original status label of a certain equipment is good, and due to an abnormality during operation, the score drops and the status label becomes warning, this change is quickly identified, and combined with the basic information of the equipment, real - time operation data, historical status records, etc., detailed equipment operation status information is generated. This information not only includes the current status of the equipment, but also covers the time of status change, speculation on the possible reasons for the change, etc., providing strong support for operation and maintenance personnel to timely grasp the equipment operation status and take targeted measures.

[0051] In a possible implementation manner, the operation and maintenance analysis module 50 further includes: An operation failure cycle status data generation unit, which is used to perform operation failure analysis on the equipment operation status information according to the multiple equipment operation cycle characteristics, and generate operation failure cycle status data.

[0052] A cycle status cluster generation unit, which is used to perform clustering analysis according to the operation failure cycle status data, determine multiple cycle status clustering centers, divide the operation failure cycle status data according to the multiple cycle status clustering centers, and generate multiple cycle status clusters.

[0053] A multi - level equipment management unit, which is used to traverse the multiple cycle status clusters to perform management analysis on multiple substation equipments, determine multiple management modes, perform reinforcement learning according to the multiple management modes to obtain management learning results, and perform multi - level equipment management based on the management learning results to formulate the equipment management suggestions.

[0054] Specifically, after obtaining the device operation status information, perform operation fault analysis on it based on multiple device operation cycle characteristics. By combining the device operation status with the operation cycle characteristics, analyze the time points, frequencies at which faults occur, and the correlations between the fault status and cycle characteristics, and generate operation fault cycle status data. These data detail the specific circumstances of fault occurrences during the device operation cycle, such as the distribution of faults within the cycle, the relationship between the fault duration and the cycle, etc.

[0055] Perform clustering analysis on the operation fault cycle status data using the K-Means algorithm. By calculating the distances between data points, determine multiple cycle status clustering centers. These clustering centers represent different types of fault cycle status patterns. Then, divide the operation fault cycle status data according to these clustering centers, grouping similar fault cycle status data into the same category, thereby generating multiple cycle status clusters, and each cycle status cluster reflects a specific device fault cycle pattern.

[0056] Start traversing multiple cycle state clusters, and conduct management analysis on multiple substation equipment based on the equipment failure cycle characteristics represented by each cluster. During the analysis process, according to factors such as the characteristics, importance, and operating environment of the equipment, it is determined whether to adopt a single management mode, that is, to formulate a special management strategy for a single device; or a cluster management mode, to group multiple devices with similar failure cycle characteristics into a group and formulate a unified management plan. For example, for some critical and unique equipment, because their failures have a significant impact and special operating characteristics, a single management mode is adopted to tailor detailed inspection and maintenance plans for them; while for a batch of equipment with the same model, similar operating environment, and similar failure cycle characteristics, a cluster management mode is adopted to improve management efficiency. After determining multiple management modes, reinforcement learning is carried out based on this. In the reinforcement learning process, various management scenarios are simulated, different management operations are continuously tried, and feedback is obtained based on the pre-set reward mechanism. This reward mechanism aims to measure the degree to which the management operation improves the operating status of the equipment. For example, if a certain management operation reduces the occurrence rate of equipment failures and improves operating stability, a higher reward will be given; otherwise, a lower reward will be given. Through continuous trial and error and optimization, we gradually learn which management mode can achieve the best results under what circumstances, thus obtaining management learning results. Based on these management learning results, we further implement multi-level equipment management. From the individual level of equipment, we consider the unique needs and operating conditions of each device, and formulate personalized management suggestions, such as maintenance time and repair methods for specific equipment; at the equipment group level, we rationally allocate resources and arrange unified maintenance activities according to the overall operating characteristics and failure rules of the equipment group; at the level of the entire substation system, we comprehensively consider the mutual influence between each equipment group and the overall operation goals of the system, and carry out macro-resource allocation and management coordination. Through this multi-level equipment management, we finally formulate comprehensive, detailed and targeted equipment management suggestions to ensure efficient and stable operation of substation equipment.

[0057] In a possible implementation, the multi-level device management unit further includes: The change trend analysis unit is used to perform change trend analysis on multiple substation equipment based on the multiple periodic state clusters to obtain multiple periodic change trend data.

[0058] The single device information determination unit is used to associate and map the multiple periodic change trend data, extract the periodic change trend data with the same trend change for combination, determine multiple device groups, extract the periodic change trend data with different trend changes for identification, and determine multiple single device information.

[0059] The single device management mode generating unit is used to generate a single device management mode according to the plurality of single device information in combination with corresponding periodic change trend data.

[0060] A cluster device management mode generation unit is configured to generate a cluster device management mode according to the multiple device groups and the corresponding periodic change trend data.

[0061] Multiple management mode acquisition units are configured to integrate the single device management mode and the cluster device management mode to obtain the multiple management modes.

[0062] Specifically, for multiple periodic state clusters, a comprehensive change trend analysis is carried out on multiple substation devices to obtain multiple periodic change trend data. Specifically, starting from each periodic state cluster, the operation data of the corresponding devices within the cluster are arranged in a time series. By means of the moving average method in the time series analysis method, the long-term trend, seasonal fluctuations, and random fluctuation components in the data are identified. For the key operation parameters of the devices, such as voltage, current, temperature, etc., observe their numerical changes in different periodic stages respectively. For example, analyze how the voltage gradually rises or falls within a complete cycle and the magnitude of the fluctuations. Through continuous tracking and analysis of these parameters over multiple cycles, capture the laws and patterns of parameter changes, and then generate multiple periodic change trend data that describe in detail the evolution of the device operation state over the cycle. These data provide a key basis for subsequent in-depth understanding of the device operation status and prediction of the future device state.

[0063] After obtaining the multiple periodic change trend data, perform correlation mapping on these data. This means establishing a corresponding relationship among numerous data for easy analysis and comparison. Compare each piece of periodic change trend data with other data one by one. For the periodic change trend data with the same trend changes, that is, the data that are consistent in key features such as change direction, amplitude, and period, select and combine them. Based on these combinations, group the devices with the same trend changes into one group to determine multiple device groups. At the same time, for the periodic change trend data with different trend changes, that is, the data that are significantly different from other data in key features, perform special identification on them. These identifications can accurately correspond to the devices that generate this data. Through these identifications, determine multiple single device information, which includes the device number, type, and its unique periodic change trend characteristics, etc.

[0064] According to the determined multiple single device information and in combination with their respective corresponding periodic change trend data, formulate a dedicated management mode for each single device. For example, arrange inspections at specific times and targeted maintenance measures according to the unique operation trend of the device to form a single device management mode.

[0065] For multiple device groups, the cluster device management mode is also generated in combination with the corresponding periodic change trend data. Given that the trend changes of devices within a group are the same, unified management strategies can be formulated, such as centralized maintenance plans, resource allocation schemes, etc., to improve management efficiency and reduce management costs.

[0066] Finally, the generated single-device management mode and the cluster device management mode are integrated. By comprehensively considering the characteristics of different devices as well as the efficiency and effectiveness of management, the advantages of these two management modes are combined to finally obtain multiple comprehensive, flexible, and highly targeted management modes, providing diverse and effective solution options for substation equipment management.

[0067] In a possible implementation manner, the multi-level device management unit further includes: A management state space generation unit, configured to perform operation and maintenance management state learning on the single-device management mode and the cluster device management mode according to multiple substation devices, and generate a management state space.

[0068] A management action space generation unit, configured to perform operation and maintenance management action learning on the single-device management mode and the cluster device management mode according to multiple substation devices, and generate a management action space.

[0069] A management learning unit, configured to perform associated fusion on the management state space and the management action space to generate a fusion solution space. When the fusion solution space is balanced, a reward function is introduced to perform management learning on the single-device management mode and the cluster device management mode, and generate the management learning result.

[0070] Specifically, a unique identifier is set for each substation device to accurately track and distinguish. With the help of a data acquisition system, various types of data during the operation of the device are collected in real time. For example, through sensors installed on the device, operation parameter data such as voltage, current, and temperature are continuously collected and recorded at fixed time intervals (such as every second). At the same time, using a fault monitoring system, the time of device failure, fault codes (corresponding to different fault types), and the time when the fault is repaired are detailedly recorded. For the device maintenance history, through the maintenance record database, the time of each maintenance, the maintenance content (such as replacing parts, software upgrading, etc.), and the information of the maintenance personnel are obtained. For the single-device management mode, for each device, the collected operation parameter data is normalized to be within the same numerical range for subsequent analysis and comparison. For the fault data, the fault occurrence frequency is counted as the number of times within a certain time period (such as one month), the fault repair time is calculated as the duration from the occurrence of the fault to the completion of the repair, and it is also normalized. For the maintenance history, the maintenance cycle is measured as the number of days since the last maintenance and is also normalized. These normalized data are combined into a feature vector representing the operation and maintenance management state of the device at a certain moment in the single-device management mode. For the cluster device management mode, the data is processed in the same way as above. However, due to cluster management, factors such as the relative position of the device in the cluster and the degree of association with other devices need to be considered additionally. For example, if a device plays a key role in the cluster, it is given a higher weight. By analyzing the communication data and collaborative operation data between devices, the degree of association between devices is determined, and these factors are quantified and incorporated into the feature vector. The feature vectors generated by all substation devices in the single-device management mode and the cluster device management mode are summarized, and using data structures in machine learning, such as multi-dimensional arrays or specialized matrix storage formats, a management state space is constructed. In this space, each vector represents the operation and maintenance management state of a device under a specific management mode, and the relationship between different vectors reflects the differences and similarities in the operation and maintenance states of devices, providing a comprehensive and structured data basis for subsequent analysis and decision-making.

[0071] For multiple substation equipment, first collect the operation and maintenance management action information under the single device and cluster device management modes through equipment operation log records and automation system command capture. Then classify these actions, such as daily inspections, fault repairs, etc., and assign a unique code to each specific action type. For the single device management mode, extract the actions of each device based on the operation records, form an action sequence by time, and then use the label encoding algorithm to convert it into a feature vector; for the cluster device management mode, in addition to focusing on the actions of a single device, the collaborative actions between devices are also analyzed, and the action sequence is also formed and the feature vector is generated. Finally, the feature vectors generated by multiple substation equipment under the two management modes are summarized, and the management action space is constructed in the form of a matrix or tensor to provide a structured data basis for subsequent analysis.

[0072] The previously constructed management state space and management action space are associated and integrated. During the integration process, each specific action in the management action space is carefully matched with the corresponding various equipment states in the management state space in a systematic way. For example, for the action of "equipment inspection", the results produced when the equipment is in different states such as "high-load operation state", "normal operation state" and "fault warning state" will be analyzed. Through a large number of such matching and analysis, the effect of each action in different states is fully understood, and then the fusion solution space is generated, which clearly shows the various possible results of the interaction between actions and states, providing a rich information basis for subsequent management decisions.

[0073] When, after a series of analyses and integrations, the fusion solution space reaches an equilibrium state, this indicates that a relatively stable and coordinated relationship has been formed between the operation and maintenance management actions and the device status under the current operation and maintenance management system. At this time, in order to further optimize the management strategy, a reward function is introduced, which is designed based on a series of preset criteria and goals. For example, if a certain management action can significantly reduce the failure rate of the device, improve the operation stability of the device, or enhance the overall operation efficiency of the device, then this action will obtain a higher reward score according to the reward function; on the contrary, if a certain management action causes more problems with the device or reduces the operation performance of the device, then it will obtain a lower reward score. Through this reward mechanism, continuous management learning is carried out on various management actions in the single-device management mode and the cluster-device management mode. In this process, different management strategies are constantly tried, and these strategies are adjusted and optimized according to the feedback of the reward function. For example, if it is found that a certain maintenance action for a specific device status in the single-device management mode can obtain a high reward, then increase the execution frequency of this action in similar situations; conversely, if a certain resource allocation strategy in the cluster-device management mode results in a low reward score, try to adjust this strategy. After multiple such attempts and optimizations, the management learning results are finally generated, providing extremely valuable references for further improving the operation and maintenance management level of substation equipment, helping to formulate more scientific and efficient management plans, and ensuring the stable and reliable operation of substation equipment.

[0074] In a possible implementation manner, the management control module 60 further includes: A simulation scenario parameter construction unit, configured to activate the simulation platform to model the operation environment of multiple substation devices, construct simulation scenario parameters, map the device management suggestions to the simulation scenario parameters for simulation management, and generate multiple management effects.

[0075] A management effect judgment unit, configured to set an expected management effect threshold according to the multiple state levels, and judge whether the multiple management effects meet the expected management effects.

[0076] An operation and maintenance monitoring unit, configured to generate a positive feedback signal if the multiple management effects meet the expected management effects, and perform continuous operation and maintenance monitoring on multiple substation devices through the positive feedback signal to generate the simulation operation and maintenance results.

[0077] An abnormal operation and maintenance device set determination unit, configured to generate a negative feedback signal if the multiple management effects do not meet the expected management effects, match and identify multiple substation devices through the negative feedback signal to determine an abnormal operation and maintenance device set, and add the abnormal operation and maintenance device set to the simulation operation and maintenance results.

[0078] Specifically, after activating the simulation platform, first collect the physical environment data (such as temperature, humidity, air flow rate) and electrical parameters (such as voltage, current, load characteristics) in the substation with the help of devices such as temperature and humidity sensors and instrument transformers. At the same time, collect the technical parameters and aging information of the devices, and use means such as CFD methods, signal processing technologies, and establishing equivalent circuits and aging models to construct the simulation scenario parameters covering all aspects of device operation. Then, analyze the device management suggestions in detail, convert them into specific operation instructions and parameter adjustment requirements, and establish a mapping relationship table with the simulation scenario parameters. Adjust the parameters in the simulation platform according to this table, such as simulating device maintenance, changing the inspection cycle, etc. Start the simulation and set the time parameters, and monitor the multi-dimensional operation indicators such as device temperature, power loss, and failure rate in real time. Quantitatively analyze according to the set evaluation index system, and finally generate multiple management effect data reflecting the implementation effect of the management suggestions.

[0079] Based on multiple pre-divided state levels, combined with the operation characteristics of substation equipment, industry standards, and operation and maintenance objectives, set the corresponding expected management effect thresholds for each state level. For example, for equipment in the "excellent" state level, it is expected to maintain extremely high stability in various key operation indicators, and specific thresholds such as the equipment failure rate should be lower than 0.1% and the equipment operation efficiency should be maintained above 98% are set; for the "good" state level, the equipment failure rate threshold can be set to be lower than 0.5%, and the operation efficiency is maintained above 95%. After obtaining multiple management effects, compare each indicator in each management effect with the expected management effect thresholds of the corresponding state levels one by one. Taking the management effect of a certain device as an example, if its indicators involve equipment failure rate, operation efficiency, etc., compare the actual failure rate and operation efficiency with the failure rate and operation efficiency thresholds set for the state level to which the device belongs respectively. Judge whether the equipment failure rate is lower than the set failure rate threshold and whether the operation efficiency is higher than the set operation efficiency threshold, etc. Through the comprehensive comparison of all management effect indicators and expected management effect thresholds, obtain the judgment results of whether multiple management effects meet the expected management effects, so as to evaluate the effectiveness of the device management suggestions in the simulation environment.

[0080] When it is determined that the multiple management effects meet the expected management effects, a mechanism for generating a positive feedback signal is immediately triggered. This positive feedback signal, like a positive instruction, quickly initiates a continuous operation and maintenance monitoring process for multiple substation devices. In this process, with the help of monitoring technologies and intelligent devices, various operating parameters of substation devices, such as voltage, current, power, temperature, frequency, etc., are collected and analyzed with high frequency and high precision. By comparing with historical data and preset standard parameters in real time, the changes in the operating state of the devices are analyzed to ensure that the devices are always in a stable and efficient operating range. At the same time, these monitoring data are integrated and summarized to generate detailed and comprehensive simulation operation and maintenance results. These results not only cover the real-time operating state information of the devices, but also include predictions of device performance trends, assessments of potential risks, and further verification of the effectiveness of existing management strategies, etc., providing a solid and reliable basis for subsequent operation and maintenance decisions and effectively ensuring the safe and stable operation of the substation.

[0081] When there are situations where multiple management effects do not meet the expected management effects, an algorithm based on deviation analysis is used to generate a negative feedback signal. By calculating the difference between the actual management effect indicators and the expected management effect threshold, if the difference exceeds the preset tolerance range, the generation of the negative feedback signal is triggered. Then, a clustering analysis algorithm is used to process the operation data of multiple substation devices, and devices with similar operation characteristics and deviation trends are grouped into one category. On this basis, the support vector machine (SVM) algorithm is used to classify and discriminate the data of these clustered devices, accurately identifying those devices with a large difference from the normal operation mode, so as to determine the set of abnormal operation and maintenance devices. Finally, the information of these screened abnormal devices is integrated and added to the simulation operation and maintenance results to ensure that the detailed data of each abnormal device, such as device number, abnormal parameters, time point of abnormality, etc., are completely recorded, providing a comprehensive and accurate basis for subsequent fault troubleshooting and optimization strategy formulation.

[0082] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0083] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0084] This specification and the accompanying drawings are merely illustrative of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A substation equipment management platform integrating multi-source data fusion modeling, characterized in that, The platform includes: An operation dataset acquisition module, which is used to perform real-time data acquisition through multiple data sources to obtain a multi-source real-time operation dataset of multiple substation devices in a target area; An operation feature determination module, which is used to fuse the multi-source real-time operation dataset to generate a fused dataset, and perform feature analysis based on the fused dataset to determine multiple operation features; A digital twin model construction module, which is used to perform 3D modeling on multiple substation devices based on Gaussian modeling technology to construct 3D digital twin models of multiple substation devices; An operation status information acquisition module, which is used to synchronize the fused dataset to the 3D digital twin model to perform operation evaluation on multiple substation devices to obtain device operation status information; An operation and maintenance analysis module, which is used to traverse the device operation status information and combine the multiple operation features to perform operation and maintenance analysis on multiple substation devices to formulate device management suggestions; A management control module, which is used to perform simulation operation and maintenance based on the device management suggestions to generate simulation operation and maintenance results, dynamically feedback and optimize the device management suggestions according to the simulation operation and maintenance results to generate a device management plan, and execute the device management plan to perform intelligent management and control on substation devices.

2. The substation equipment management platform integrating multi-source data fusion modeling according to claim 1, characterized in that The operation feature determination module further includes: An operation timestamp determination unit, which is used to perform operation timing analysis on multiple substation devices according to the multiple data sources to determine the operation timestamp; An operation time series data column generation unit, which is used to align the multi-source real-time operation dataset according to the operation timestamp to generate an operation time series data column; An autoencoder construction unit, which is used to traverse the operation time series data column to reconstruct the multi-source real-time operation dataset to construct an autoencoder; An encoding fusion unit, which is used to encode and fuse the multi-source real-time operation dataset through the autoencoder to generate the fused dataset.

3. The substation equipment management platform integrating multi-source data fusion modeling according to claim 2, characterized in that, The operation feature determination module further includes: A multi-dimensional data matrix acquisition unit, which is used to load the fused dataset for multi-dimensional data integration to obtain a multi-dimensional data matrix; A multi-domain feature signal acquisition unit, which is used to traverse the multi-dimensional data matrix to perform feature mining according to the operation time series data column to obtain multi-domain feature signals, and the multi-domain features include operation frequency domain feature signals and operation time domain feature signals; A frequency domain component information determination unit, which is used to decompose the operation time domain feature signal according to the operation frequency domain feature signal to determine several frequency domain component information; A correlation analysis unit, which is used to traverse the several frequency domain component information to perform correlation analysis with multiple substation devices to obtain multiple device operation cycle features; A device operation cycle feature addition unit, which is used to add the multiple device operation cycle features to the multiple operation features.

4. The substation equipment management platform integrating multi-source data fusion modeling according to claim 1, characterized in that, The operation status information acquisition module further includes: An operation evaluation index setting unit, which is used to synchronize the fused dataset to the 3D digital twin model, and map the fused dataset to multiple device operation components through the 3D digital twin model to set multiple operation evaluation indexes; An operation scoring generation unit, configured to perform operation evaluation on multiple substation devices according to the multiple operation evaluation indicators, and generate multiple operation scores; A status label acquisition unit, configured to divide multiple status levels, and perform matching identification according to the multiple operation scores and the multiple status levels to obtain multiple status labels; A dynamic monitoring unit, configured to perform dynamic monitoring on multiple substation devices according to the multiple status labels, and generate the device operation status information.

5. The substation equipment management platform integrating multi-source data fusion modeling according to claim 3, characterized in that The operation and maintenance analysis module further includes: An operation fault cycle status data generation unit, configured to perform operation fault analysis on the device operation status information according to the multiple device operation cycle characteristics, and generate operation fault cycle status data; A cycle status cluster generation unit, configured to perform clustering analysis according to the operation fault cycle status data, determine multiple cycle status clustering centers, and divide the operation fault cycle status data according to the multiple cycle status clustering centers to generate multiple cycle status clusters; A multi-level device management unit, configured to traverse the multiple cycle status clusters to perform management analysis on multiple substation devices, determine multiple management modes, perform reinforcement learning according to the multiple management modes to obtain a management learning result, and perform multi-level device management based on the management learning result to formulate the device management suggestion.

6. The substation equipment management platform integrating multi-source data fusion modeling according to claim 5, characterized in that The multi-level device management unit further includes: A change trend analysis unit, configured to perform change trend analysis on multiple substation devices based on the multiple cycle status clusters to obtain multiple cycle change trend data; A single device information determination unit, configured to perform associated mapping on the multiple cycle change trend data, extract the cycle change trend data with the same trend change for combination to determine multiple device groups, and extract and identify the cycle change trend data with different trend changes to determine multiple single device information; A single device management mode generation unit, configured to generate a single device management mode according to the multiple single device information in combination with the corresponding cycle change trend data; A cluster device management mode generation unit, configured to generate a cluster device management mode according to the multiple device groups in combination with the corresponding cycle change trend data; A multiple management mode acquisition unit, configured to integrate the single device management mode and the cluster device management mode to obtain the multiple management modes.

7. The substation equipment management platform integrating multi-source data fusion modeling according to claim 6, characterized in that, The multi-level device management unit further includes: A management status space generation unit, configured to perform operation and maintenance management status learning on the single device management mode and the cluster device management mode according to multiple substation devices to generate a management status space; A management action space generation unit, configured to perform operation and maintenance management action learning on the single device management mode and the cluster device management mode according to multiple substation devices to generate a management action space; A management learning unit, configured to perform associated fusion on the management status space and the management action space to generate a fusion solution space. When the fusion solution space is balanced, introduce a reward function to perform management learning on the single device management mode and the cluster device management mode to generate the management learning result.

8. The substation equipment management platform integrating multi-source data fusion modeling according to claim 4, characterized in that The management control module further includes: A simulation scenario parameter construction unit, which is used to activate the simulation platform to model the operating environment of multiple substation devices, construct simulation scenario parameters, map the device management suggestions to the simulation scenario parameters for simulation management, and generate multiple management effects; A management effect judgment unit, which is used to set an expected management effect threshold according to the multiple state levels, and judge whether the multiple management effects meet the expected management effect; An operation and maintenance monitoring unit, which is used to generate a positive feedback signal if the multiple management effects meet the expected management effect, continuously monitor the operation and maintenance of multiple substation devices through the positive feedback signal, and generate the simulation operation and maintenance result; An abnormal operation and maintenance device set determination unit, which is used to generate a negative feedback signal if the multiple management effects do not meet the expected management effect, match and identify multiple substation devices through the negative feedback signal, determine the abnormal operation and maintenance device set, and add the abnormal operation and maintenance device set to the simulation operation and maintenance result.

Images