Intelligent substation equipment state inspection method and system based on artificial intelligence

Abstract

The invention discloses an intelligent substation equipment state inspection method and system based on artificial intelligence, and relates to the technical field of equipment inspection. The method comprises the following steps: carrying out interference signal acquisition on a current substation area to obtain an interference monitoring signal set; extracting interference intensity indexes; performing operation interference influence prediction on the substation equipment set, and outputting an interference influence prediction index set; analyzing the operation attribute information, and outputting an equipment importance index set; and calculating to obtain an interference inspection priority index set, and sending the interference inspection priority index set to an inspection control center for path planning to obtain an inspection planning path. The technical problem that the routing inspection priority setting is not reasonable enough due to the fact that the influence degree of environmental interference on substation equipment is difficult to accurately evaluate in the prior art is solved, and the purposes of quantifying the influence degree of interference on the equipment through the model, reasonably determining the routing inspection priority in combination with the importance of the equipment and improving the routing inspection efficiency are achieved. Therefore, the technical effect of improving the scientificity of inspection decision is improved.

Description

Technical Field

[0001] This invention relates to the field of equipment inspection technology, specifically to an intelligent inspection method and system for the status of substation equipment based on artificial intelligence. Background Technology

[0002] With the continuous expansion of substation scale and the increasing complexity of operating environments, equipment condition inspection plays a crucial role in ensuring the safe operation of the power system. Existing inspection methods typically rely on fixed periodic plans or manual experience to set the inspection sequence, making it difficult to reflect the external interference environment in a timely manner. When substation equipment is affected by electromagnetic interference, environmental noise, communication interference, etc., current technology lacks the ability to quantitatively analyze and predict the degree of interference, and cannot accurately assess the impact of interference on different equipment during operation. This results in a lack of rationality and dynamism in setting inspection priorities. This may not only lead to uneven allocation of inspection resources, but also prevent the timely detection and handling of potential risks to critical equipment, thereby affecting the stable operation of the power grid. Summary of the Invention

[0003] This application provides an intelligent inspection method and system for substation equipment status based on artificial intelligence, which solves the technical problem in the prior art that it is difficult to accurately assess the degree of influence of environmental interference on substation equipment, resulting in unreasonable setting of inspection priorities.

[0004] The first aspect of this application provides an intelligent inspection method for the status of substation equipment based on artificial intelligence, the method comprising: An interference sensing device is installed to collect interference signals in the current substation area, obtaining an interference monitoring signal set. Interference intensity analysis is performed on the interference monitoring signal set to extract interference intensity indices. A set of substation equipment is collected, and the interference intensity indices are input into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set, outputting a set of interference impact prediction indices corresponding to the substation equipment set. The operational attribute information of each device in the substation equipment set is analyzed, outputting a set of equipment importance indices corresponding to the substation equipment set. An interference inspection priority index set is calculated based on the interference impact prediction index set and the equipment importance index set, and this set is sent to the inspection control center for path planning to obtain the inspection planning path.

[0005] A second aspect of this application provides an intelligent inspection system for the status of substation equipment based on artificial intelligence, the system comprising: Signal Acquisition Module: Sets up an interference sensor to acquire interference signals in the current substation area, obtaining an interference monitoring signal set. First Analysis Module: Performs interference intensity analysis on the interference monitoring signal set, extracting interference intensity indices. Prediction Module: Collects a set of substation equipment, inputs the interference intensity indices into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set, and outputs a set of interference impact prediction indices corresponding to the substation equipment set. Second Analysis Module: Analyzes the operational attribute information of each device in the substation equipment set, and outputs a set of equipment importance indices corresponding to the substation equipment set. Path Planning Module: Calculates and obtains a set of interference inspection priority indices based on the interference impact prediction indices and the equipment importance indices, and sends the interference inspection priority indices to the inspection control center for path planning, obtaining the inspection planning path.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, interference sensors are installed to collect interference signals in the current substation area, obtaining an interference monitoring signal set. Next, interference intensity analysis is performed on the interference monitoring signal set to extract interference intensity indices. Then, the substation equipment set is collected, and the interference intensity indices are input into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set, outputting a set of interference impact prediction indices corresponding to the substation equipment set. Further, the operational attribute information of each device in the substation equipment set is analyzed, outputting a set of equipment importance indices corresponding to the substation equipment set. Finally, an interference inspection priority index set is calculated based on the interference impact prediction index set and the equipment importance index set, and sent to the inspection control center for path planning to obtain the inspection planning path. This solves the technical problem in existing technologies where it is difficult to accurately assess the degree of environmental interference affecting substation equipment, leading to unreasonable inspection priority settings. It achieves the technical effect of quantifying the impact of interference on equipment through a model and rationally determining inspection priorities based on equipment importance, thereby improving the scientific nature of inspection decisions. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1A schematic diagram of the intelligent inspection method for substation equipment status based on artificial intelligence provided in this application embodiment; Figure 2 A schematic diagram of the structure of an AI-based intelligent inspection system for substation equipment status provided in this application embodiment.

[0009] Explanation of reference numerals in the attached diagram: Signal acquisition module 11, first analysis module 12, prediction module 13, second analysis module 14, path planning module 15. Detailed Implementation

[0010] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0011] Example 1, as Figure 1 As shown, this application provides an intelligent inspection method for the status of substation equipment based on artificial intelligence, wherein the method includes: An interference sensing device is installed, and interference signals are collected in the current substation area based on the interference sensing device to obtain an interference monitoring signal set.

[0012] In this embodiment, an interference sensing device is deployed within the substation area. This device can be an electromagnetic interference detector, a radio frequency interference receiver, an acoustic interference pickup, or other dedicated sensors capable of sensing electrical environmental interference signals. The interference sensing device continuously collects data on the electromagnetic environment, communication signal fluctuations, and noise interference sources within the target area at a preset sampling frequency, obtaining a raw interference signal data stream. During the data collection process, the interference sensing device automatically timestamps and spatially marks the collected signals to ensure accurate differentiation of interference information at different locations and time periods. The raw interference signal data stream undergoes analog-to-digital conversion, bandpass filtering, and denoising to remove sampling errors and background noise interference, resulting in a structured interference monitoring signal set. This interference monitoring signal set is stored in vector or matrix form, and each signal record contains at least time series information, spectral characteristic parameters, amplitude characteristic parameters, and the coordinates of the collection point.

[0013] Furthermore, the interference sensing device is installed in a critical area of ​​the substation, and the method for identifying the critical area of ​​the substation includes: The distribution of equipment in the current substation area is analyzed to obtain a set of equipment distribution locations. Based on the electrical connection relationships of each device and the set of equipment distribution locations, a substation topology network is constructed. Graph centrality analysis is performed on the substation topology network to identify key areas of the substation. Graph centrality analysis includes degree centrality, betweenness centrality, and proximity centrality.

[0014] Preferably, the distribution locations of equipment in the current substation area are collected to obtain a set of distribution locations of each device in a spatial coordinate system. This set can accurately reflect the relative layout of the equipment in the substation. Based on the electrical connection relationships between the devices and the set of device distribution locations, a substation topology network is constructed, where nodes represent equipment units within the substation, and edges represent electrical connections or functional coupling relationships between devices. Graph centrality analysis is performed on the substation topology network, calculating indices such as degree centrality, betweenness centrality, and proximity centrality: degree centrality characterizes the number of direct connections between devices and other devices, reflecting its local importance in the network; betweenness centrality measures the frequency with which a device node is on the shortest path to other nodes, reflecting its importance in controlling electrical transmission paths; proximity centrality measures the average shortest path distance between a device node and other nodes in the network, reflecting its proximity to its global location.

[0015] Degree centrality: ,in, The degree centrality of node v represents the number of edges connected to it, and N represents the total number of nodes in the substation topology. Higher degree centrality indicates that the node is directly connected to more devices and is more critical in the local area.

[0016] Betweenness centrality: ,in, This represents the number of shortest paths from node s to node t. This represents the number of paths that pass through node v in these shortest paths. Higher betweenness centrality indicates that the node plays a more important transit role in information or power transmission.

[0017] Proximity centrality: ,in, represents the shortest path length between node v and node u, and N represents the total number of nodes in the network. Higher proximity centrality indicates that the node can connect with other nodes more quickly and is more centrally located globally.

[0018] By comprehensively analyzing and weighting the three centrality indicators mentioned above, the equipment areas with the highest comprehensive scores in the topology network are identified as key areas of the substation. Deploying interference sensors in these key areas maximizes the collection of interference signal data that most significantly impacts equipment operation, thereby improving the accuracy and effectiveness of subsequent interference impact prediction and inspection priority determination.

[0019] Interference intensity analysis is performed on the interference monitoring signal set to extract interference intensity indices.

[0020] In this embodiment, interference intensity analysis is performed on the acquired interference monitoring signal set to extract interference intensity indicators that can quantitatively characterize interference features. Specifically, the interference monitoring signal set is first segmented according to time series to ensure that interference features within different time windows can be effectively captured. Then, frequency domain analysis methods (such as Fast Fourier Transform) are used to perform spectral decomposition on the interference signal to extract the amplitude, power spectral density, and peak intensity parameters of each frequency band to identify the main frequency distribution characteristics of the interference signal. Simultaneously, time domain feature extraction methods are used to calculate the mean, variance, root mean square amplitude, peak factor, and volatility of the interference signal to measure its performance in terms of intensity, stability, and suddenness. Finally, the frequency domain parameters and time domain parameters are normalized and weighted to obtain a standardized interference intensity indicator.

[0021] Collect a set of substation equipment, input the interference intensity index into a pre-trained interference-operation impact prediction model to predict the operation interference impact of the substation equipment set, and output a set of interference impact prediction indexes corresponding to the substation equipment set.

[0022] In this embodiment of the application, information is collected on the equipment of the substation to form a substation equipment set. The substation equipment set includes the main primary and secondary equipment in the substation, such as circuit breakers, transformers, disconnect switches, current transformers, voltage transformers, relay protection devices and communication control units, etc. Each device is associated with its operating parameters, functional attributes and topological location.

[0023] The interference intensity index is used as an input variable and fed into the pre-trained interference-operation impact prediction model. The interference-operation impact prediction model is jointly trained based on a large amount of historical interference data and equipment operating status data, and can establish a nonlinear mapping relationship between interference signal characteristics and changes in equipment operating status.

[0024] During prediction, the model infers the impact of operational interference on each device in the equipment set, outputting corresponding quantitative indicators such as voltage fluctuation threshold shift, power transmission stability degradation rate, communication link interruption probability, and relay protection malfunction risk value. By aggregating the operational interference prediction results of all devices, a set of interference impact prediction indicators corresponding to the substation equipment set is obtained. This set of interference impact prediction indicators can intuitively reflect the potential operational risk level of different devices under the current interference environment.

[0025] Furthermore, the method for constructing the interference-operation impact prediction model by inputting the interference intensity index into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set includes: Historical interference data samples and corresponding interference intensity index libraries were collected from time-series records; equipment operation status data samples were collected from synchronous time-series records; the historical interference data samples, interference intensity index libraries, and equipment operation status data samples were sliced ​​into time windows to obtain interference impact training input samples; impact labels were defined by analyzing the parameter change thresholds of the equipment operation status data samples to obtain interference impact training label samples; supervised training of the model was performed using the interference impact training input samples and the interference impact training label samples to obtain a converged interference-operation impact prediction model.

[0026] Specifically, firstly, interference data samples recorded in the substation's operational history are collected over time. These samples include electromagnetic interference signals, power harmonic interference signals, and communication interference signals. A corresponding interference intensity index library is then established using a pre-defined feature extraction algorithm, with indicators including root mean square interference value, power spectral density, peak factor, and burst frequency. Simultaneously, equipment operating status data samples recorded synchronously with the aforementioned interference data samples are collected. These samples include operating parameters such as equipment voltage, current, power factor, number of switching actions, and communication delay. Next, the interference data samples, the interference intensity index library, and the equipment operating status data samples are sliced ​​according to a unified time window to obtain interference impact training input samples, ensuring temporal consistency of the input data. Then, parameter change analysis is performed on the equipment operating status data samples, and operating thresholds are set (e.g., voltage deviation greater than 5%, communication interruption time exceeding 2 seconds). Based on these thresholds, impact labels are defined, resulting in interference impact training label samples. Finally, the training input samples and training label samples are input into the supervised learning framework for model training. Algorithms such as gradient boosting decision trees, random forests, or deep neural networks are used for iterative optimization. The model parameters are updated by backpropagation by minimizing the loss function (such as mean squared error) until convergence to a stable state, thus obtaining an interference-operation impact prediction model that can accurately predict the impact of interference intensity on equipment operation.

[0027] Furthermore, the interference-running impact prediction model is trained under supervised training using the interference-impact training input samples and the interference-impact training label samples to obtain a converged interference-running impact prediction model. The method includes: A gradient boosting decision tree model and a loss function (mean squared error loss function) for supervised training are initialized. The interference-affected training input samples are input into the initialized gradient boosting decision tree model, and threshold samples for predicted parameter changes are obtained based on the initialized model. The mean squared error loss is calculated using the mean squared error loss function on the known parameter change threshold samples and the interference-affected training label samples, and the mean squared error loss data is output. Backpropagation is performed on the initialized gradient boosting decision tree model according to the mean squared error loss data to obtain the descent gradient. The model parameters of the initialized gradient boosting decision tree model are updated based on the descent gradient until a convergent interference-affected prediction model is obtained.

[0028] Preferably, the gradient boosting decision tree model is initialized, and a loss function is set for supervised training, wherein the mean squared error loss function is selected, and its definition is: Where N is the number of samples, To prevent interference from affecting the threshold of true parameter changes in training labeled samples, The threshold for the predicted parameter changes.

[0029] During training, interference-affected training input samples are batch-wise fed into the initialized gradient boosting decision tree model. The model outputs corresponding prediction parameter change threshold samples through weighted ensemble of multiple weak classification trees. Subsequently, the mean squared error loss function is used to calculate the error between the prediction parameter change threshold samples and the known interference-affected training label samples, yielding mean squared error loss data. Based on the mean squared error loss data, a backpropagation step is performed to calculate the descent gradient of the loss function with respect to the model parameters. Then, the parameters of the gradient boosting decision tree are iteratively updated according to the descent gradient, including updating the split node weights and leaf node output values ​​of each weak classification tree to minimize the overall loss function. Through multiple iterations, training terminates when the convergence condition of the mean squared error loss function meets a preset threshold, resulting in the interference-running effect prediction model in a converged state.

[0030] Furthermore, the method of inputting the interference intensity index into a pre-trained interference-operational impact prediction model to predict the operational interference impact on the substation equipment set includes: The interference intensity index is input into a pre-trained interference-operation impact prediction model to obtain the predicted equipment operation status data of each device in the substation equipment set; the predicted equipment operation status data of each device is compared with the initial equipment operation status data to obtain the operating parameter change threshold, including peak offset and stable offset; based on the peak offset and the stable offset, the interference impact prediction index set corresponding to the substation equipment set is obtained.

[0031] When performing operational interference impact prediction, the extracted interference intensity index is first input into a pre-trained interference-operation impact prediction model. This model is jointly trained based on historical interference data and equipment operating status data, and can output predicted equipment operating status data for each device in the substation equipment set under the current interference environment. This predicted operating status data includes key parameters such as voltage, current, power factor, signal delay, and switching frequency. Subsequently, the predicted operating status data for each device is compared with the corresponding initial operating status data to calculate the operating parameter change threshold under interference. Specifically, the operating parameter change threshold includes at least peak offset and stable offset. The peak offset reflects the maximum deviation of operating parameters under short-term sudden interference, and its calculation formula can be expressed as: ,in, This is the peak offset. To predict operating status parameters, These are the initial running status parameters.

[0032] The stability offset is used to reflect the long-term deviation of operating parameters after the continuous effect of disturbance. It can be obtained by calculating the difference between the mean of the predicted state in the steady-state interval and the mean of the initial state.

[0033] Based on peak offset and stable offset, a set of interference impact prediction indicators for each device is constructed. This set of interference impact prediction indicators can quantify the operational risk level of each device in an interference environment.

[0034] Furthermore, methods for pre-trained perturbation-running influence prediction models also include: Historical interference data samples, interference intensity index library, and equipment operation status data samples under different interference signal types are collected; signal impact sensitivity analysis is performed on the equipment operation status data samples, and the signal type-impact sensitivity mapping relationship corresponding to different interference signal types is output; the interference-operation impact prediction model is updated and trained using the signal type-impact sensitivity mapping relationship to obtain the optimized interference-operation impact prediction model.

[0035] Preferably, historical interference data samples are collected under various interference signal types, including but not limited to electromagnetic harmonic interference, radio interference, transient impulse interference, and communication noise interference. A corresponding interference intensity index library is established to characterize the amplitude characteristics, spectral distribution, burst frequency, and power spectral density under different signal types. Simultaneously, equipment operating status data samples synchronized with the aforementioned interference data are collected, including operating parameters such as voltage, current, power, number of protection actions, and communication delay. Based on this, signal impact sensitivity analysis is performed on the equipment operating status data samples. Specifically, for different interference signal types, sensitivity coefficients of equipment operating parameters are calculated, such as the sensitivity of voltage deviation to electromagnetic harmonic interference and the sensitivity of communication delay to radio interference. These are quantified using parameter change rate, standard deviation increment, and normalized fluctuation index, outputting a signal type-impact sensitivity mapping relationship based on different interference signal types. The signal type-impact sensitivity mapping relationship can clearly distinguish the main impact dimensions and sensitivity of different types of interference on equipment operation. Finally, the signal type-impact sensitivity mapping relationship is introduced into the model training process to update and train the pre-trained interference-operation impact prediction model. During training updates, the model's fitting ability on highly sensitive parameter dimensions is enhanced by adjusting the weights of the input features and the weighting coefficients in the loss function. After multiple rounds of iterative optimization, an optimized interference-operation impact prediction model is obtained. This model can not only predict the impact of general interference intensity on equipment operation, but also dynamically adjust the prediction accuracy according to different interference types, thereby significantly improving the model's adaptability and reliability in complex electromagnetic environments.

[0036] Furthermore, after outputting the set of interference impact prediction indicators corresponding to the substation equipment set, the method further includes: A quantitative analysis of the correlation impact degree is performed on the equipment operation status data sample to obtain a quantitative index sample of the correlation impact degree; the interference-operation impact prediction model is updated and trained based on the quantitative index sample of the correlation impact degree to obtain an optimized interference-operation impact prediction model.

[0037] Preferably, a parameter correlation matrix is ​​constructed based on the correlation between various operating parameters in the equipment operating status data sample. The matrix elements represent the correlation strength of different operating parameters under interference, such as the correlation coefficient between voltage fluctuation and current offset, and the correlation coefficient between power fluctuation and communication delay. The degree of correlation between equipment operating parameters and interference is obtained by calculating the Pearson correlation coefficient. Further, the parameter correlation matrix is ​​standardized, and correlation pairs exceeding a preset threshold are extracted as high-sensitivity parameter combinations to form a quantitative index sample of correlation influence. This quantitative index sample reflects the collaborative change characteristics of different operating parameters under interference, thereby capturing complex influence patterns that cannot be reflected by a single index. Based on this, the quantitative index sample of correlation influence is used as a new training feature input to the interference-operational impact prediction model for model update training. During the update training process, the feature weights of the model under multi-parameter coupling relationships are adjusted, and a multi-objective loss function is used to jointly optimize the prediction accuracy and correlation fit. After iterative training, an optimized interference-operational impact prediction model is obtained, enabling it not only to predict changes in individual parameters but also to comprehensively consider the interactions between different operating parameters, improving the accuracy and robustness of the prediction results.

[0038] The operational attribute information of each device in the substation equipment set is analyzed, and a set of equipment importance indicators corresponding to the substation equipment set is output.

[0039] Furthermore, the method for outputting the set of equipment importance indicators corresponding to the substation equipment set includes: Based on the operational attribute information of each device in the substation equipment set, multiple evaluation indicators are extracted for each device, including voltage level, topology criticality, transmission power, power supply load, and device function type; the multiple evaluation indicators are weighted by device importance to output the set of device importance indicators corresponding to the substation equipment set.

[0040] First, based on the operational attribute information of each device in the substation equipment set, multiple evaluation indicators are extracted to measure the operational importance of the equipment. These indicators include at least voltage level, topological criticality, transmission power, power supply load, and equipment function type. Voltage level is quantified by classifying the equipment according to its voltage level (e.g., 500kV, 220kV, 110kV). Topological criticality is calculated based on the degree centrality, betweenness centrality, and proximity centrality of the substation topology network to characterize the equipment's connectivity and control role in the network. Transmission power is determined by the equipment's rated power and actual operating power. Power supply load is quantified by the downstream load capacity carried by the equipment. Equipment function type is classified according to the equipment's position in the primary or secondary system; for example, main transformers and busbar equipment have a higher weight than general circuit breakers or monitoring and control devices. Then, these evaluation indicators are standardized to allow comparisons of indicators with different dimensions on a unified scale. Next, combined with a preset weight set, each indicator is weighted and calculated to obtain the overall importance value of the equipment. Finally, the importance calculation results of all equipment are summarized, outputting a set of equipment importance indicators corresponding to the substation equipment set.

[0041] The interference inspection priority index set is calculated and obtained according to the interference impact prediction index set and the equipment importance index set. The interference inspection priority index set is sent to the inspection control center for path planning to obtain the inspection planning path.

[0042] In this embodiment, a weighted calculation is performed between the interference impact prediction index set and the equipment importance index set to generate an interference inspection priority index set. This set comprehensively reflects the operational risk level of each device under the current interference environment and its importance in the power grid. The interference inspection priority index set is sent to the inspection control center, which automatically generates an inspection planning path based on the priority order, the device distribution location, and available inspection resources. This ensures that critical equipment is inspected first, while improving the overall rationality and efficiency of the inspection process.

[0043] Furthermore, the set of interference inspection priority indicators is sent to the inspection control center for path planning to obtain the inspection planning path. The method includes: The inspection control center sorts the substation equipment set according to the size of the interference inspection priority index set and outputs the equipment inspection sequence; it obtains the equipment distribution location set and the inspection resource location set of the substation equipment set; and it performs path planning according to the equipment inspection sequence, combined with the equipment distribution location set and the inspection resource location set, to obtain the inspection planning path.

[0044] Specifically, the inspection control center sorts the substation equipment set according to the interference inspection priority index of each device, obtaining an inspection sequence from high priority to low priority. This inspection sequence reflects the order of devices that most need to be inspected under the current interference environment. Subsequently, the spatial distribution location set of the substation equipment set and the inspection resource location set are obtained. The equipment distribution location set includes the two-dimensional or three-dimensional coordinate information of each device, and the inspection resource location set includes the initial position and reachability of inspection robots, drones, or manual inspection personnel. Based on this, according to the equipment inspection sequence, high-priority devices are placed as the front-end constraints of path planning. Combining the equipment distribution location set and the inspection resource location set, a path optimization problem is constructed. Specifically, the optimization objectives are to minimize the total travel length of the inspection path, minimize the inspection time, and maximize the coverage. Graph search algorithms (such as the improved Dijkstra algorithm, A* heuristic algorithm) or intelligent optimization algorithms (such as genetic algorithms, ant colony algorithms, particle swarm optimization algorithms) are used for path calculation. Ultimately, the output inspection planning path not only ensures that high-priority equipment is covered first, but also achieves optimal overall inspection efficiency when inspection resources are limited.

[0045] In summary, the embodiments of this application have at least the following technical effects: First, interference sensors are installed to collect interference signals in the current substation area, obtaining an interference monitoring signal set. Next, interference intensity analysis is performed on the interference monitoring signal set to extract interference intensity indices. Then, the substation equipment set is collected, and the interference intensity indices are input into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set, outputting a set of interference impact prediction indices corresponding to the substation equipment set. Further, the operational attribute information of each device in the substation equipment set is analyzed, outputting a set of equipment importance indices corresponding to the substation equipment set. Finally, an interference inspection priority index set is calculated based on the interference impact prediction index set and the equipment importance index set, and sent to the inspection control center for path planning to obtain the inspection planning path. This solves the technical problem in existing technologies where it is difficult to accurately assess the degree of environmental interference affecting substation equipment, leading to unreasonable inspection priority settings. It achieves the technical effect of quantifying the impact of interference on equipment through a model and rationally determining inspection priorities based on equipment importance, thereby improving the scientific nature of inspection decisions.

[0046] Example 2 is based on the same inventive concept as the AI-based intelligent inspection method for substation equipment status in the previous examples, such as... Figure 2 As shown, this application provides an intelligent inspection system for the status of substation equipment based on artificial intelligence, wherein the system includes: Signal acquisition module 11: Sets up an interference sensing device to acquire interference signals in the current substation area based on the interference sensing device, and obtains an interference monitoring signal set; First analysis module 12: Performs interference intensity analysis on the interference monitoring signal set and extracts interference intensity indicators; Prediction module 13: Collects a set of substation equipment, inputs the interference intensity indicators into a pre-trained interference-operation impact prediction model to predict the operation interference impact of the substation equipment set, and outputs a set of interference impact prediction indicators corresponding to the substation equipment set; Second analysis module 14: Analyzes the operation attribute information of each device in the substation equipment set and outputs a set of equipment importance indicators corresponding to the substation equipment set; Path planning module 15: Calculates and obtains an interference inspection priority indicator set according to the interference impact prediction indicator set and the equipment importance indicator set, sends the interference inspection priority indicator set to the inspection control center for path planning, and obtains the inspection planning path.

[0047] Furthermore, the signal acquisition module 11 is used to perform the following methods: The distribution of equipment in the current substation area is analyzed to obtain a set of equipment distribution locations. Based on the electrical connection relationship of each device and the set of equipment distribution locations, a substation topology network is constructed. Graph centrality analysis is performed on the substation topology network to identify key areas of the substation. The graph centrality analysis includes degree centrality, betweenness centrality, and proximity centrality.

[0048] Furthermore, the prediction module 13 is used to perform the following method: Historical interference data samples and corresponding interference intensity index libraries were collected from time-series records; equipment operation status data samples were collected from synchronous time-series records; the historical interference data samples, interference intensity index libraries, and equipment operation status data samples were sliced ​​into time windows to obtain interference impact training input samples; impact labels were defined by analyzing the parameter change thresholds of the equipment operation status data samples to obtain interference impact training label samples; supervised training of the model was performed using the interference impact training input samples and the interference impact training label samples to obtain a converged interference-operation impact prediction model.

[0049] Furthermore, the prediction module 13 is used to perform the following method: A gradient boosting decision tree model and a loss function (mean squared error loss function) for supervised training are initialized. The interference-affected training input samples are input into the initialized gradient boosting decision tree model, and threshold samples for predicted parameter changes are obtained based on the initialized model. The mean squared error loss is calculated using the mean squared error loss function on the known parameter change threshold samples and the interference-affected training label samples, and the mean squared error loss data is output. Backpropagation is performed on the initialized gradient boosting decision tree model according to the mean squared error loss data to obtain the descent gradient. The model parameters of the initialized gradient boosting decision tree model are updated based on the descent gradient until a convergent interference-affected prediction model is obtained.

[0050] Furthermore, the prediction module 13 is used to perform the following method: The interference intensity index is input into a pre-trained interference-operation impact prediction model to obtain the predicted equipment operation status data of each device in the substation equipment set; the predicted equipment operation status data of each device is compared with the initial equipment operation status data to obtain the operating parameter change threshold, including peak offset and stable offset; based on the peak offset and the stable offset, the interference impact prediction index set corresponding to the substation equipment set is obtained.

[0051] Furthermore, the prediction module 13 is used to perform the following method: Historical interference data samples, interference intensity index library, and equipment operation status data samples under different interference signal types are collected; signal impact sensitivity analysis is performed on the equipment operation status data samples, and the signal type-impact sensitivity mapping relationship corresponding to different interference signal types is output; the interference-operation impact prediction model is updated and trained using the signal type-impact sensitivity mapping relationship to obtain the optimized interference-operation impact prediction model.

[0052] Furthermore, the prediction module 13 is used to perform the following method: A quantitative analysis of the correlation impact degree is performed on the equipment operation status data sample to obtain a quantitative index sample of the correlation impact degree; the interference-operation impact prediction model is updated and trained based on the quantitative index sample of the correlation impact degree to obtain an optimized interference-operation impact prediction model.

[0053] Furthermore, the second analysis module 14 is used to perform the following methods: Based on the operational attribute information of each device in the substation equipment set, multiple evaluation indicators are extracted for each device, including voltage level, topology criticality, transmission power, power supply load, and device function type; the multiple evaluation indicators are weighted by device importance to output the set of device importance indicators corresponding to the substation equipment set.

[0054] Furthermore, the path planning module 15 is used to perform the following methods: The inspection control center sorts the substation equipment set according to the size of the interference inspection priority index set and outputs the equipment inspection sequence; it obtains the equipment distribution location set and the inspection resource location set of the substation equipment set; and it performs path planning according to the equipment inspection sequence, combined with the equipment distribution location set and the inspection resource location set, to obtain the inspection planning path.

[0055] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An intelligent inspection method for the status of substation equipment based on artificial intelligence, characterized in that, The method includes: An interference sensing device is installed, and interference signals are collected in the current substation area based on the interference sensing device to obtain an interference monitoring signal set. The interference monitoring signal set is subjected to interference intensity analysis, and interference intensity indices are extracted; Collect a set of substation equipment, input the interference intensity index into a pre-trained interference-operation impact prediction model to predict the operation interference impact of the substation equipment set, and output a set of interference impact prediction indicators corresponding to the substation equipment set. The operational attribute information of each device in the substation equipment set is analyzed, and a set of equipment importance indicators corresponding to the substation equipment set is output. The interference inspection priority index set is calculated and obtained according to the interference impact prediction index set and the equipment importance index set. The interference inspection priority index set is sent to the inspection control center for path planning to obtain the inspection planning path.

2. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 1, characterized in that, The interference sensing device is installed in a critical area of ​​the substation, and the method for identifying the critical area of ​​the substation includes: Analyze the equipment distribution locations in the current substation area to obtain a set of equipment distribution locations; Construct a substation topology network based on the electrical connection relationships of each device and the set of device distribution locations; Graph centrality analysis is performed on the substation topology network to identify key areas of the substation. The graph centrality analysis includes degree centrality, betweenness centrality, and proximity centrality.

3. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 1, characterized in that, The method for constructing the interference-operation impact prediction model by inputting the interference intensity index into a pre-trained interference-operation impact prediction model to predict the operational interference impact on the substation equipment set includes: Collect historical interference data samples recorded in time series and the corresponding interference intensity index library; Collect samples of equipment operating status data recorded synchronously in time sequence; The historical interference data samples, interference intensity index library and equipment operation status data samples are sliced ​​into time windows to obtain interference impact training input samples. By analyzing the parameter change threshold of the device operating status data sample, the influence label is defined, and interference influence training label samples are obtained; The interference-running impact prediction model is trained under supervision using the interference-impact training input samples and the interference-impact training label samples, and then trained to convergence.

4. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 3, characterized in that, The interference-running impact prediction model is trained under supervised supervision using the interference impact training input samples and the interference impact training label samples to obtain a converged interference-running impact prediction model. The method includes: Initialize the gradient boosting decision tree model and the loss function used for supervised training of the model, wherein the loss function is the mean squared error loss function; The interference-affected training input samples are input into the initial gradient boosting decision tree model, and the predicted parameter change threshold samples are obtained based on the initial gradient boosting decision tree model. The mean squared error loss function is used to calculate the mean squared error loss for the known parameter change threshold samples of the predicted parameter change threshold and the known parameter change threshold of the interference-affected training label samples, and the mean squared error loss data is output. The initial gradient boosting decision tree model is backpropagated using the mean squared error loss data to obtain the descent gradient. The model parameters of the initial gradient boosting decision tree model are updated according to the descent gradient until a disturbance-running impact prediction model that has been trained to convergence is obtained.

5. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 1, characterized in that, The method involves inputting the interference intensity index into a pre-trained interference-operational impact prediction model to predict the operational interference impact on the substation equipment set. The interference intensity index is input into a pre-trained interference-operation impact prediction model to obtain the predicted equipment operation status data of each device in the substation equipment set. By comparing the predicted equipment operating status data of each device with the initial equipment operating status data, the threshold for changes in operating parameters is obtained, including peak offset and stable offset. Based on the peak offset and the stable offset, obtain the set of interference impact prediction indicators corresponding to the substation equipment set.

6. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 3, characterized in that, Pre-trained perturbation-running influences the prediction model; other methods include: Collect historical interference data samples, interference intensity index library, and equipment operation status data samples under different interference signal types; The device operation status data samples are subjected to signal impact sensitivity analysis, and the signal type-impact sensitivity mapping relationship corresponding to different interference signal types is output. The interference-operation impact prediction model is updated and trained using the signal type-impact sensitivity mapping relationship to obtain an optimized interference-operation impact prediction model.

7. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 3, characterized in that, After outputting the set of interference impact prediction indicators corresponding to the substation equipment set, the method further includes: Perform a quantitative analysis of the correlation influence on the equipment operation status data samples to obtain a sample of quantitative indicators of the correlation influence. The interference-operation impact prediction model is updated and trained based on the quantitative index samples of the correlation impact degree to obtain the optimized interference-operation impact prediction model.

8. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 1, characterized in that, The method for outputting the set of equipment importance indicators corresponding to the substation equipment set includes: Based on the operational attribute information of each device in the substation equipment set, multiple evaluation indicators are extracted for each device, including voltage level, topology criticality, transmission power, power supply load, and device function type. The multiple evaluation indicators are weighted according to equipment importance, and the set of equipment importance indicators corresponding to the substation equipment set is output.

9. The intelligent inspection method for substation equipment status based on artificial intelligence as described in claim 1, characterized in that, The set of interference inspection priority indicators is sent to the inspection control center for path planning to obtain the inspection planning path. The method includes: The inspection control center sorts the substation equipment set according to the size of the interference inspection priority index set and outputs the equipment inspection sequence. Obtain the set of equipment distribution locations and the set of inspection resource locations for the substation equipment set; Based on the equipment inspection sequence, and combined with the set of equipment distribution locations and the set of inspection resource locations, a path planning is performed to obtain the inspection planning path.

10. An intelligent inspection system for substation equipment status based on artificial intelligence, characterized in that: The system is used to implement the intelligent substation equipment status inspection method based on artificial intelligence as described in any one of claims 1-9, the system comprising: Signal acquisition module: Equipped with an interference sensor, the module acquires interference signals in the current substation area based on the interference sensor and obtains a set of interference monitoring signals. First analysis module: Performs interference intensity analysis on the interference monitoring signal set and extracts interference intensity indicators; Prediction module: Collects the set of substation equipment, inputs the interference intensity index into the pre-trained interference-operation impact prediction model to predict the operation interference impact of the set of substation equipment, and outputs the set of interference impact prediction indexes corresponding to the set of substation equipment. The second analysis module analyzes the operational attribute information of each device in the substation equipment set and outputs a set of equipment importance indicators corresponding to the substation equipment set. Path planning module: Calculates and obtains the interference inspection priority index set according to the interference impact prediction index set and the equipment importance index set, and sends the interference inspection priority index set to the inspection control center for path planning to obtain the inspection planning path.

Images