Sensor optimized deployment method and system for capacitor bank internal fault location

Abstract

The application provides a sensor optimization deployment method and system for capacitor bank internal fault location, relates to the technical field of power equipment monitoring, and comprises the following steps: determining candidate installation points of sensors according to the topological structure of a capacitor bank and physical deployment conditions of the sensors; simulating the response signal characteristics of electrical parameters of each capacitor unit in the capacitor bank at each candidate installation point under a preset fault according to an electrical simulation model, and generating a fault fingerprint library; calculating the difference degree of the response signal characteristics of each candidate installation point under different preset faults and the response characteristics in a normal state, and combining the discrimination degree of the fault type to calculate a fault location value score; calculating the deployment marginal benefit of each candidate installation point according to the fault location value score, the fault risk weight of the candidate installation point and the monitoring redundancy of the selected installation point, and then determining a target installation point that meets a cost constraint condition. By implementing the method, the accuracy and efficiency of fault location can be improved within a limited cost.

Description

Technical Field

[0001] This application relates to the field of power equipment monitoring technology, and in particular to a sensor optimization deployment method and system for locating internal faults in capacitor banks. Background Technology

[0002] As core equipment for reactive power compensation and filtering in power systems, capacitor banks are crucial to the safe and economical operation of the entire power grid. Capacitor banks typically consist of hundreds or thousands of capacitor units connected in series and parallel, resulting in complex structures and variable operating environments. During long-term operation, individual capacitor units may fail due to dielectric aging, internal breakdown, or oil leakage from the casing. Failure to detect and address a single unit's failure in a timely manner can trigger a chain reaction, leading to damage to a wider range of parallel or series capacitor units, or even causing serious accidents such as capacitor bank explosions, posing a significant threat to power grid stability and equipment safety. Therefore, real-time and accurate monitoring of the operating status of each unit within the capacitor bank, and the ability to quickly locate the specific faulty unit when a fault occurs, is a key technical requirement for ensuring the safe operation of the power system.

[0003] In related technologies, a comprehensive monitoring scheme is typically employed, where one or more sensors, such as voltage, current, or temperature sensors, are installed on each individual capacitor cell of the capacitor bank. These sensors, distributed throughout the entire capacitor bank, can collect the electrical or physical state parameters of each cell in real time. When any capacitor cell malfunctions, its corresponding dedicated sensor can directly and quickly detect the parameter change and upload the signal to the monitoring system. By analyzing which sensor reported the abnormal data, the monitoring system can directly and accurately determine which specific capacitor cell the fault occurred in, thus achieving precise location of the faulty cell.

[0004] However, with the continuous expansion of modern power grids, large capacitor banks can contain hundreds or even thousands of capacitor cells. In this context, continuing with a comprehensive coverage approach that assigns a dedicated sensor to each capacitor cell would result in an enormous total number of sensors, leading to extremely high hardware procurement and on-site installation costs. Furthermore, this approach indiscriminately invests monitoring resources in all capacitor cells, but in reality, cells in different locations do not have equal fault risk or importance to the system. Deploying a large amount of monitoring resources to low-risk, low-importance cells results in significant resource waste. Summary of the Invention

[0005] This application provides a sensor optimization deployment method and system for fault location within a capacitor bank, addressing the problem of excessive cost and resource waste associated with installing sensors on each capacitor unit in a capacitor bank for monitoring in related technologies.

[0006] In a first aspect, this application provides a sensor optimization deployment method for locating faults within a capacitor bank, applied to a capacitor bank fault monitoring system, the method comprising: Based on the topological parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensors, candidate installation points for the sensors are determined; Based on the electrical simulation model, the response signal characteristics of electrical parameters of each capacitor unit in the capacitor bank at each candidate installation point under a preset fault are simulated, and a fault fingerprint database including the mapping relationship between the fault source and the response signal characteristics is generated. The difference between the response signal characteristics of each candidate installation point under different preset faults and the response characteristics under normal conditions is calculated based on the fault fingerprint database. The fault location value score is calculated based on the difference and the distinguishability of the fault type. The fault location value score is used to characterize the fault distinguishability of the candidate installation point. Based on the fault location value score, the fault risk weight of the candidate installation point, and the monitoring redundancy of the selected installation point, the deployment marginal benefit of each candidate installation point is calculated. The deployment marginal benefit is used to characterize the monitoring performance improvement brought about by adding candidate installation points on the basis of existing sensor deployment. The fault risk weight is determined according to the historical failure rate and the location importance level. The monitoring redundancy indicates the degree to which the candidate installation point is covered and monitored by the deployed sensors. Based on the order of deployment marginal benefits from high to low, target installation points that meet the preset cost constraints are selected from the candidate installation points to form an optimized sensor deployment scheme.

[0007] By adopting the above technical solution, the system constructs a comprehensive fault fingerprint database through simulation, thereby pre-quantifying the inherent potential of each candidate installation point to distinguish different faults, i.e., a fault location value score. Then, a fault risk weight representing the likelihood of actual faults and a monitoring redundancy factor to avoid duplicate monitoring are introduced, and these three factors are combined to calculate the deployment marginal benefit. Finally, based on the deployment marginal benefit and under cost constraints, selection is made to ensure that each new sensor is preferentially deployed in "high-value areas" with strong fault differentiation capabilities, high potential risks, and the ability to fill existing monitoring blind spots. This achieves optimal allocation of monitoring resources within a limited cost, improving the fault location accuracy and coverage efficiency of the entire monitoring network.

[0008] In some embodiments, the step of determining candidate installation points for the sensor based on the topological parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensor specifically includes: Based on the three-dimensional topology model of the capacitor bank, identify all electrical connection nodes between capacitor units; Based on the physical size limitations and installation space requirements of the sensor, installable nodes that meet the mechanical installation conditions are selected from all the electrical connection nodes; Based on the electromagnetic field distribution simulation results, target nodes located in the preset strong electromagnetic interference area are eliminated from the installable nodes to obtain candidate installation points for the sensor.

[0009] By adopting the above technical solution, the system comprehensively identifies all electrical connection nodes using a three-dimensional topology model, ensuring the completeness of the selection. Next, by considering the physical size of the sensors and installation space, physically infeasible nodes are eliminated, ensuring the engineering feasibility of the solution. Finally, nodes in areas with strong interference are removed based on electromagnetic field simulation, avoiding the risk of signal distortion or damage to the sensors due to environmental interference. This series of steps ensures that the candidate point set on which subsequent optimization calculations are based is not only theoretically optimal but also possesses high stability and high signal quality in actual deployment and operation.

[0010] In some embodiments, after the step of determining candidate installation points for the sensor based on the topology parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensor, the method further includes: Establish an electrical distance matrix between candidate installation points and each capacitor unit based on the spatial distribution of candidate installation points; The sensitivity level of each candidate installation point to faults at different locations is determined based on the electrical distance matrix; A preset number of target candidate installation points are selected sequentially from high to low sensitivity levels for simulation to reduce the amount of simulation computation. The simulation results of the target candidate installation points are used to generate the fault fingerprint database.

[0011] By adopting the above technical solution, the system can quickly assess the inherent sensitivity of each candidate point to faults at different locations by establishing an electrical distance matrix. Based on this sensitivity, the system can prioritize the "golden" candidate points that are most sensitive to fault response for time-consuming electrical simulations. This approach avoids unnecessary complex calculations on a large number of low-sensitivity, low-value candidate points, reducing the computational resources and time required to generate the fault fingerprint database. This allows the entire optimization deployment method to maintain high computational efficiency even when dealing with complex capacitor banks consisting of hundreds or thousands of units.

[0012] In some embodiments, before the step of calculating the deployment marginal benefit of each candidate installation point based on the fault location value score, the fault risk weight of the candidate installation point, and the monitoring redundancy of the selected installation points, the method further includes: Based on historical operating data of the capacitor bank, the frequency and type distribution of faults in capacitor units in each region were statistically analyzed. The positional importance level of each capacitor unit is determined based on its electrical connection relationship and functional importance level within the capacitor bank. The fault risk weight within a preset range around each candidate installation point is calculated by weighting the fault occurrence frequency and the location importance level. The degree of overlap between each candidate installation point and the monitoring range of the deployed sensors is calculated based on the monitoring range and sensitivity of the deployed sensors, thus obtaining the monitoring redundancy of each candidate installation point.

[0013] By adopting the above technical solution, the system analyzes historical data and the electrical connection relationships of units. This allows the system to assess not only the actual probability of fault occurrence in different areas but also to identify the importance of key units. The combination of these two factors is quantified into fault risk weights, enabling monitoring resources to be allocated to high-risk areas. Simultaneously, by calculating the overlap of the monitoring ranges of deployed sensors, monitoring redundancy is quantified, effectively avoiding redundant resource investment in areas already adequately covered.

[0014] In some embodiments, the step of calculating the deployment marginal benefit of each candidate installation point based on the fault location value score, the fault risk weight of the candidate installation point, and the monitoring redundancy of the selected installation points specifically includes: The fault location value score is multiplied by the fault risk weight to obtain the weighted fault location value of the candidate installation point; A redundancy penalty coefficient is calculated based on the monitoring redundancy, and the redundancy penalty coefficient is negatively correlated with the monitoring redundancy. The effective monitoring value is obtained by multiplying the weighted fault location value by the redundancy penalty coefficient. The monitoring benefit per unit cost is calculated based on the preset cost-benefit function and the effective monitoring value. By adopting the above technical solution, the system established a deployment marginal benefit calculation model, ensuring the scientific nature of sensor selection. This model first multiplies the inherent fault discrimination capability of candidate points (fault location value score) by the probability of fault occurrence in the real world (fault risk weight), obtaining a weighted fault location value that better reflects actual value. Next, by introducing a penalty coefficient negatively correlated with monitoring redundancy, the value of candidate points in already covered areas is dynamically reduced, accurately characterizing their "marginal" contribution. Finally, the monitoring benefit per unit cost is calculated by combining cost factors.

[0015] In some embodiments, the step of calculating the fault location value score based on the difference and the distinguishability of the fault type specifically includes: Calculate the Euclidean distance between the response signal characteristics of the candidate installation point under each preset fault and the response characteristics under normal conditions to obtain the fault response difference degree; Construct a feature vector space for the response features of each candidate installation point under different fault types, and calculate the cosine similarity between feature vectors of different fault types; The discriminant index between fault types is calculated based on the cosine similarity, and the discriminant index is negatively correlated with the cosine similarity. The fault response difference and the discrimination index are weighted and fused to obtain a comprehensive fault identification capability index; The information gain of candidate installation points for fault source localization is calculated based on information entropy theory, and the information gain is multiplied by the comprehensive fault identification capability index to obtain the fault localization value score.

[0016] By employing the above technical solution, the system calculates the difference between the fault response and the normal state using Euclidean distance, ensuring that the sensor possesses the basic ability to "detect anomalies." Furthermore, by calculating the cosine similarity between different fault feature vectors and converting it into a discriminative index, the system evaluates the sensor's advanced ability to "distinguish types," avoiding fault confusion. Finally, combining information entropy theory, the system multiplies the aforementioned comprehensive identification capability with the location information gain, so that the final "fault location value score" not only represents the strength of fault identification but also reflects its contribution to narrowing the search range for fault sources, thus improving the accuracy of the score.

[0017] In some embodiments, after the step of selecting target installation points that meet preset cost constraints from candidate installation points according to the order of deployment marginal benefits from high to low, to form an optimized sensor deployment scheme, the method further includes: Based on the location and monitoring range of the target installation point in the sensor optimization deployment scheme, calculate the electrical distance from each capacitor unit to the nearest sensor; Capacitor cells with an electrical distance greater than a preset distance threshold are marked as weak monitoring cells, and the proportion of such weak monitoring cells is counted. When the proportion of the number of the monitored weak units exceeds a preset safety threshold, the supplementary installation point that covers the most monitored weak units is selected from the remaining candidate installation points and added to the target installation point set.

[0018] By employing the aforementioned technical solution, the system can intuitively identify "weak monitoring units" located at the edge or in gaps of the monitoring range by calculating the electrical distance from each capacitor unit to the nearest sensor. By comparing their number with a preset safety threshold, the system can quantitatively assess the coverage adequacy of the deployment scheme. Once a monitoring blind spot is detected exceeding the safety range, the system automatically activates a supplementary mechanism, selecting candidate points that can most efficiently cover these weak units to join the scheme. This proactively eliminates monitoring blind spots, effectively avoiding the risk of missed faults due to localized monitoring deficiencies and enhancing the reliability of the entire monitoring system.

[0019] Secondly, this application provides a capacitor bank fault monitoring system, the system comprising: one or more processors and a memory; The memory is coupled to the one or more processors. The memory is used to store computer program code, which includes computer instructions. The one or more processors call the computer instructions so that the system can implement the sensor optimization deployment method for fault location inside a capacitor bank provided in the above embodiments, which will not be described in detail here.

[0020] Thirdly, this application provides a computer-readable storage medium including instructions that, when executed on a capacitor bank fault monitoring system, enable the system to implement a sensor optimization deployment method for fault location within a capacitor bank provided in the above embodiments, which will not be elaborated further here.

[0021] Fourthly, this application provides a computer program product that, when running on a capacitor bank fault monitoring system, enables the system to implement a sensor optimization deployment method for locating internal faults in a capacitor bank, as provided in the above embodiments, which will not be elaborated here.

[0022] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. The system calculates the marginal benefit of deployment for each candidate installation point. This benefit integrates three key dimensions: the candidate point's inherent ability to distinguish different faults, the actual fault risk determined based on historical data and location importance, and the monitoring redundancy of the point relative to the selected sensors. By selecting from high to low marginal benefits within cost constraints, the system ensures that each deployment decision is the optimal increment based on the current layout, thereby maximizing monitoring performance with the highest resource utilization efficiency.

[0023] 2. Through electrical simulation and information theory calculations, a location value score representing the fault differentiation capability of each point was generated; through analysis of historical data and topology, weight values ​​quantifying the fault risk of each point were generated; and through evaluation of physical dimensions and electromagnetic environment, Boolean values ​​and coefficients for screening and penalty were generated. This effect allows the deployment value of each candidate point to be objectively presented in the form of a vector containing multi-dimensional values.

[0024] 3. Based on the optimized sensor deployment results, the system proactively calculates the electrical distance from all capacitor cells to the nearest sensor, thereby identifying weak cells with insufficient monitoring coverage. Once the proportion of these cells exceeds a preset safety threshold, the system will automatically initiate a compensation procedure, selecting supplementary points from the remaining candidate points that can most effectively cover these weak areas and adding them to the deployment plan. This ensures that the final deployment plan, while pursuing economic benefits, also meets the stringent safety requirements for comprehensive and reliable monitoring of the entire capacitor bank, proactively eliminating potential monitoring blind spots. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a sensor optimization deployment method for fault location within a capacitor bank, as described in an embodiment of this application. Figure 2 This is another schematic flowchart of a sensor optimization deployment method for fault location within a capacitor bank, as described in an embodiment of this application. Figure 3 This is a schematic diagram of the physical device structure of a capacitor bank fault monitoring system in an embodiment of this application. Detailed Implementation

[0026] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0027] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0028] For ease of understanding, the method provided in this implementation is described in process below. Please refer to [link / reference]. Figure 1 This is a flowchart illustrating a sensor optimization deployment method for locating internal faults in a capacitor bank, as described in an embodiment of this application.

[0029] S101. Based on the topology parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensors, determine the candidate installation points of the sensors.

[0030] Among them, topology parameters refer to the core parameters such as the connection method, arrangement, and distribution of electrical nodes of each capacitor unit in the capacitor bank; sensor physical deployment conditions refer to the physical constraints that the sensor must meet when it is installed, including the size of the installation space, the sensor physical size adaptation requirements, and the electromagnetic environment adaptation conditions; candidate installation points refer to potential locations that are feasible for sensor installation after screening.

[0031] Specifically, the system acquires the core structural information of the capacitor bank to be monitored and the installation requirements of the sensor itself. Then, considering the limitations of the actual installation environment, it filters out suitable locations from all possible sites. It not only ensures that the selected points are physically capable of installing the sensor, but also avoids subsequent sensor malfunctions due to environmental issues, such as avoiding areas with strong electromagnetic interference, to ensure the sensor can accurately acquire signals later.

[0032] Optionally, the system can import the design drawings and structural parameter files of the capacitor bank to be monitored, and use 3D modeling tools to construct a 3D topological model of the capacitor bank to obtain the spatial location of all electrical connection nodes. Based on the obtained parameters such as the physical size and installation space requirements of the sensors, the system matches them one by one with the electrical connection nodes in the 3D model, and selects nodes whose space size meets the installation requirements as installable nodes. The system runs an electromagnetic field distribution simulation program to simulate the electromagnetic field distribution when the capacitor bank is running, identifies areas of strong electromagnetic interference, and removes nodes located in these areas from the installable nodes. The remaining nodes are the candidate installation points.

[0033] S102. Based on the electrical simulation model, simulate the response signal characteristics of electrical parameters of each capacitor unit in the capacitor bank at each candidate installation point under the preset fault, and generate a fault fingerprint database including the mapping relationship between the fault source and the response signal characteristics.

[0034] Among them, the electrical simulation model refers to the simulation model built based on the electrical characteristics and topology of the capacitor bank, which is used to simulate the changes in electrical parameters under different fault conditions; the preset fault refers to the types of faults that the capacitor bank may experience in advance, including electrical abnormal faults caused by dielectric aging, internal breakdown, and oil leakage from the casing; the response signal characteristics of electrical parameters refer to the unique change patterns and characteristics of electrical parameters such as voltage, current, and frequency at the candidate installation point when a fault occurs; and the fault fingerprint database refers to the data set that stores the mapping relationship between the fault source and the corresponding response signal characteristics.

[0035] Specifically, the system uses a pre-built electrical simulation model to simulate the changes in electrical parameters at each candidate installation point when each preset fault occurs, forming unique signal characteristics. In some embodiments, the system can build a high-precision electrical simulation model based on the topological and electrical characteristic parameters of the capacitor bank using professional simulation software, and calibrate the model to ensure that it accurately reflects the actual operating state of the capacitor bank. The system simulates the occurrence process of each preset fault one by one in the simulation model, and for each candidate installation point, it collects the changes in electrical parameters such as voltage and current in real time when the fault occurs, extracting features such as signal amplitude, frequency, and waveform. The system associates and stores the fault type (fault source) with the response signal characteristics of the corresponding candidate installation point, establishes a complete mapping table, and forms a fault fingerprint database.

[0036] S103. Calculate the difference between the response signal characteristics of each candidate installation point under different preset faults and the response characteristics under normal conditions based on the fault fingerprint database, and calculate the fault location value score based on the difference and the distinguishability of the fault type.

[0037] Among them, the normal state response characteristics refer to the stable signal characteristics of electrical parameters at the candidate installation point when the capacitor bank is running without faults; the difference refers to the degree of deviation between the response signal characteristics under fault conditions and the response characteristics under normal conditions; the fault type discrimination refers to the degree of distinguishability between the response signal characteristics corresponding to different fault types at the candidate installation point; the fault location value score is an indicator used to quantify the fault discrimination ability of the candidate installation point. The higher the score, the stronger the ability of the point to identify and distinguish faults.

[0038] Specifically, the system extracts the response signal features of each candidate installation point under different preset faults from the fault fingerprint database, compares them with the response features of the point under normal conditions, and calculates the degree of difference between the two. The greater the degree of difference, the easier it is to detect the fault at that point. Next, the system obtains the difference between the response signal features of the point corresponding to different fault types, which is the discrimination index. The higher the discrimination index, the more accurately the point can identify different types of faults. Finally, the system combines the degree of difference and the discrimination index to obtain a fault location value score. This score comprehensively reflects the candidate installation point's overall ability to detect and identify faults.

[0039] In some embodiments, the system can retrieve the normal state response characteristics and response signal characteristics under all preset faults for each candidate installation point from the fault fingerprint database, calculate the degree of deviation between each fault characteristic and the normal characteristics, and obtain the difference degree; then, classify the response signal characteristics of different fault types corresponding to each candidate installation point, analyze the uniqueness between different fault characteristics, calculate the distinguishability between various fault characteristics, and obtain the discrimination degree; according to the preset weight coefficient, the difference degree and the discrimination degree are weighted and calculated to obtain the fault location value score of each candidate installation point.

[0040] In other embodiments, the system can also calculate the information entropy value of the fault response signal features corresponding to each candidate installation point based on the information entropy theory. The higher the information entropy value, the richer the fault information contained in the signal features of that point. The system performs matching analysis on the response signal features of different fault types, counts the number of fault types that the point can accurately distinguish and the discrimination accuracy, and thus quantifies the discrimination degree. The information gain after the information entropy value is converted is combined with the discrimination degree and the difference degree to obtain the fault location value score.

[0041] S104. Based on the fault location value score, the fault risk weight of the candidate installation points, and the monitoring redundancy of the selected installation points, calculate the deployment marginal benefit of each candidate installation point.

[0042] Among them, the fault risk weight refers to the weight value determined based on the historical fault rate and location importance level of the area surrounding the candidate installation point, which is used to reflect the possibility of a fault occurring in the area and the degree of fault impact; the selected installation point refers to the location where the sensor has been determined to be installed in the current deployment plan; the monitoring redundancy refers to the degree to which the candidate installation point is covered by the monitoring range of the selected installation point; the deployment marginal benefit refers to the degree of monitoring performance improvement that can be brought by adding the candidate installation point on the basis of the existing sensor deployment.

[0043] The system obtains a fault location value score for each candidate installation point, and then combines this with the fault risk weight of that point—areas with high fault risk may have greater practical significance for deploying sensors even if the score is slightly lower. Simultaneously, the system also considers monitoring redundancy. If the monitoring range of a point is already fully covered by the selected installation points, deploying another sensor is a redundant investment; the higher the redundancy, the lower the actual benefit. The system comprehensively calculates these three factors to obtain the marginal benefit of deployment. This benefit value directly reflects the performance improvement that adding this candidate installation point to the overall monitoring system based on the current deployment.

[0044] In some embodiments, the system can retrieve the fault location value score and corresponding fault risk weight of each candidate installation point, multiply the two to obtain a weighted fault location value, which reflects the matching degree between the core value and fault risk of the point; calculate the monitoring redundancy based on the overlap of the monitoring range of each candidate installation point with the selected installation point, and determine the redundancy penalty coefficient based on the redundancy, with a larger penalty coefficient for higher redundancy; multiply the weighted fault location value with the redundancy penalty coefficient, and then combine it with a preset cost-benefit function to calculate the monitoring benefit per unit cost, which is used as the deployment marginal benefit of the candidate installation point.

[0045] S105. Select target installation points that meet the preset cost constraints from the candidate installation points according to the order of deployment marginal benefits from high to low, and form a sensor optimization deployment scheme.

[0046] Among them, the preset cost constraints refer to the system's pre-set constraints such as the total budget for sensor deployment and the upper limit of the number of deployable sensors; the target installation point refers to the final installation location selected from the candidate installation points that meets the cost constraints and has high deployment marginal benefits; the sensor optimized deployment scheme refers to a complete deployment plan that includes information such as the location of all target installation points, sensor types, and deployment order.

[0047] Specifically, the system sorts all candidate installation points from highest to lowest based on their marginal benefit. Points with higher marginal benefits show a more significant performance improvement from new deployments and thus have higher priority. The system then sequentially filters installation points according to this sorting, while simultaneously calculating the total deployment cost of the selected points to ensure it does not exceed preset cost constraints. When the selected installation points meet the cost requirements and maximize monitoring performance, these points become the target installation points. Finally, the system integrates the location information of the target installation points, corresponding sensor type suggestions, installation sequence planning, and other information to form a complete sensor optimization deployment plan.

[0048] In some embodiments, the system can sort all candidate installation points from high to low based on their marginal benefits, forming a candidate installation point ranking list; select candidate installation points as target points in sequence according to the ranking order, and accumulate the deployment costs of the target points until the accumulated costs approach or reach the preset cost constraint limit; then perform a final verification on the selected target points to ensure that their monitoring coverage can meet basic monitoring needs and that there are no major monitoring blind spots, and finally determine the target installation points and form a deployment plan.

[0049] In addition, in other embodiments, the system can also calculate the electrical distance from each capacitor unit to the nearest sensor based on the location and monitoring range of the target installation point in the sensor optimization deployment scheme; mark capacitor units with electrical distances greater than a preset distance threshold as weak monitoring units, and count the proportion of weak monitoring units; when the proportion of weak monitoring units exceeds a preset safety threshold, select the supplementary installation point with the most weak monitoring units from the remaining candidate installation points and add it to the target installation point set.

[0050] In the above embodiments, the system constructs a comprehensive fault fingerprint database through simulation, thereby pre-quantifying the inherent potential of each candidate installation point to distinguish different faults, i.e., a fault location value score. Then, a fault risk weight representing the likelihood of actual faults and a monitoring redundancy factor to avoid duplicate monitoring are introduced, and these three factors are combined to calculate the deployment marginal benefit. Finally, based on the deployment marginal benefit and under cost constraints, selection is made to ensure that each new sensor is preferentially deployed in "high-value areas" with strong fault differentiation capabilities, high potential risks, and the ability to fill existing monitoring blind spots. This achieves optimal allocation of monitoring resources within a limited cost, improving the fault location accuracy and coverage efficiency of the entire monitoring network.

[0051] The following provides a more detailed description of the process of the method provided in this implementation. Please refer to [link / reference]. Figure 2 This is another flowchart illustrating a sensor optimization deployment method for locating faults inside a capacitor bank, as described in an embodiment of this application.

[0052] S201. Based on the three-dimensional topology model of the capacitor bank, identify all electrical connection nodes between capacitor units.

[0053] Among them, the three-dimensional topology model refers to a three-dimensional digital model constructed based on the actual size of the capacitor bank, the arrangement of each capacitor unit, and the electrical connection relationship, which is used to intuitively present the spatial structure and electrical relationship of the capacitor bank; the electrical connection node refers to the physical contact point that realizes the electrical connection between capacitor units and between capacitor units and external circuits, and is the hub for the transmission of current and voltage.

[0054] Specifically, the system acquires complete structural information of the capacitor bank to be monitored, and transforms the abstract electrical connections into a visualized spatial model by constructing a three-dimensional topology model, ensuring that no connection point between capacitor units is missed. Subsequently, based on the connection method of each capacitor unit in the model (series and parallel combinations, etc.), all nodes used to achieve electrical conduction are accurately identified.

[0055] Optionally, the system can first receive the capacitor bank design drawings, structural parameter list, and electrical connection schematic diagram imported by the user; then use 3D modeling software (such as SolidWorks or AutoCAD Electrical) to parse the above files, extract the spatial coordinates, dimensional parameters, and connection port information of each capacitor unit; finally, based on the electrical connection relationship, mark the intersection point of all capacitor unit connection ports in the 3D model, which is the electrical connection node.

[0056] S202. Based on the physical size limitations and installation space requirements of the sensor, select the installable nodes that meet the mechanical installation conditions from all electrical connection nodes, and determine the candidate installation points of the sensor by combining the electromagnetic field distribution simulation results.

[0057] Among them, mechanical installation conditions refer to the physical adaptation requirements that the installation location must meet, including spatial size adaptation, installation interface compatibility, and no physical obstruction; installable nodes refer to electrical connection nodes that have the physical feasibility for sensor installation after screening for physical installation conditions; electromagnetic field distribution simulation results refer to the output results of data such as the electromagnetic field strength and distribution range inside and around the capacitor bank during operation through simulation software.

[0058] Specifically, the system matches the physical parameters (size, mounting interface type, etc.) of the sensor to be deployed with the space requirements for installation against the spatial conditions of the identified electrical connection nodes one by one, eliminating nodes with insufficient space, physical obstructions, or incompatible mounting interfaces, thus obtaining the installable nodes. Subsequently, considering that the capacitor bank generates electromagnetic fields during operation, and that strong electromagnetic interference can cause signal distortion in the sensor, the system simulates the electromagnetic field distribution during normal operation of the capacitor bank to identify areas of strong electromagnetic interference. Nodes located within these areas are further eliminated from the installable nodes, and the remaining nodes are the candidate installation points for the sensor.

[0059] Optionally, the system can obtain the physical size parameters (length, width, height, mounting hole spacing, etc.) and installation space standards (minimum installation space volume, heat dissipation space requirements, etc.) of the sensor to be deployed; then retrieve the three-dimensional topology model constructed in step S201, measure the available space size and installation interface matching degree around each electrical connection node; filter out nodes whose available space meets the physical size restrictions and whose installation interfaces are compatible as installable nodes; finally, run electromagnetic field simulation software (such as ANSYS Maxwell), set the rated operating parameters of the capacitor bank, simulate the electromagnetic field distribution, set a strong electromagnetic interference threshold, and eliminate nodes whose electromagnetic field strength exceeds the threshold among the installable nodes to obtain candidate installation points.

[0060] S203. Based on the spatial distribution of candidate installation points, establish the electrical distance matrix between candidate installation points and each capacitor unit, and determine the sensitivity level of each candidate installation point to faults at different locations.

[0061] The electrical distance matrix refers to a dataset of electrical distances between each candidate installation point and each capacitor unit, stored in matrix form. The rows of the matrix correspond to candidate installation points, the columns correspond to capacitor units, and the matrix elements are the electrical distances between the two. Electrical distance is a physical quantity that characterizes the length of the electrical signal transmission path and the degree of signal attenuation between the candidate installation point and the capacitor unit, and is related to the transmission efficiency of fault signals. Sensitivity level refers to the degree of sensitivity of the candidate installation point to the signal changes generated when capacitor units at different locations fail. The higher the level, the easier it is for the point to capture the fault signal at the corresponding location.

[0062] Specifically, the system collects the spatial coordinates of all candidate installation points, and, combined with the spatial location of each capacitor unit and the electrical topology of the capacitor bank, calculates the electrical distance from each candidate installation point to each capacitor unit. This distance considers not only physical spatial distance but also the influence of the conductivity characteristics of the electrical connection path on signal transmission. Subsequently, the electrical distances between all candidate installation points and all capacitor units are organized and stored in matrix form, forming an electrical distance matrix. Based on this matrix, the system analyzes the electrical distance relationship between each candidate installation point and capacitor units at different locations: the closer the electrical distance, the smaller the attenuation during fault signal transmission, and the more sensitive the candidate point is to faults at that location. Based on this, the sensitivity of candidate installation points to faults at different locations is classified into different levels.

[0063] Optionally, the system can also simulate the signal transmission loss from each capacitor unit to each candidate installation point using electromagnetic simulation software, quantify the transmission loss into electrical distance; construct an electrical distance matrix with candidate installation points as rows and capacitor units as columns; and use the analytic hierarchy process (AHP) to divide the sensitivity into four levels: super-level, level one, level two, and level three, based on the influence weight of electrical distance on signal perception. The shortest electrical distance corresponds to super-level sensitivity, and so on, to determine the sensitivity level of each candidate installation point.

[0064] S204. Select a preset number of target candidate installation points in descending order of sensitivity level for simulation to generate a fault fingerprint database that includes the mapping relationship between fault sources and response signal features.

[0065] Among them, the preset quantity refers to the number of candidate installation points for simulation that the system pre-sets. This number is determined based on the simulation computing resources, the scale of the capacitor bank, and the accuracy requirements for fault monitoring. The target candidate installation point refers to the key candidate points selected from all candidate installation points according to their sensitivity level for participation in fault simulation. The fault source refers to the specific location (specific capacitor unit) and fault type (such as dielectric aging, internal breakdown, etc.) that cause the capacitor bank fault. The response signal characteristics refer to the unique changes in electrical parameters (voltage, current, etc.) collected at the target candidate installation point when the fault occurs, including amplitude changes, frequency shifts, waveform distortion, etc.

[0066] Specifically, the system sorts all candidate installation points from highest to lowest sensitivity level. Candidate points with higher sensitivity levels have a stronger ability to capture fault signals, and their simulation results have greater reference value for fault location. Then, the system selects the top N (N being a preset number) of candidate points from the sorted list as target candidate installation points for simulation. The electrical simulation model simulates the changes in electrical parameters at each target candidate installation point when different fault sources (different fault types at different locations) occur. The system extracts the unique response signal features corresponding to each fault source, and associates and stores the fault source information with the response signal features to form a fault fingerprint database.

[0067] S205. Calculate the Euclidean distance between the response signal characteristics of the candidate installation point under each preset fault and the response characteristics under normal conditions to obtain the fault response difference.

[0068] Among them, the fault response difference degree refers to the degree of deviation between the response signal characteristics under fault conditions and the response characteristics under normal conditions, which is calculated by Euclidean distance. The greater the difference degree, the more obvious the fault signal.

[0069] Specifically, the system extracts the response signal features of each candidate installation point under normal conditions from the fault fingerprint database and transforms them into multi-dimensional feature vectors (each dimension corresponds to a signal feature parameter, such as amplitude, frequency, etc.). Then, for each preset fault, the system extracts the fault response signal features corresponding to that candidate installation point, also transforming them into multi-dimensional feature vectors. By calculating the Euclidean distance between these two feature vectors, the magnitude of this distance directly reflects the degree of difference between the signal features of the fault state and the normal state, thus obtaining the fault response difference degree. The greater the difference degree, the more significant the signal change at the candidate installation point when that fault occurs, and the easier it is for the monitoring system to identify the fault.

[0070] Optionally, the system can first retrieve the normal state response signal feature data of each candidate installation point from the fault fingerprint database, extract four core parameters—amplitude, frequency, phase, and waveform distortion rate—and construct a 4-dimensional normal feature vector. For each preset fault, the system can extract the same four parameters of the fault response signal corresponding to the candidate installation point and construct a 4-dimensional fault feature vector. Using the Euclidean distance calculation formula, the straight-line distance between the two vectors is calculated, and this distance value is the fault response difference degree of the candidate installation point under this type of fault. According to the correspondence between candidate installation points and fault types, all difference degree data are organized and stored.

[0071] S206. Construct the feature vector space of each candidate installation point's response features under different fault types, calculate the cosine similarity between feature vectors of different fault types, and then determine the distinguishability index between fault types.

[0072] Among them, the feature vector space refers to a multi-dimensional mathematical space constructed with the response signal features of candidate installation points under different fault types as the dimension, and the response feature of each fault type corresponds to a feature vector in the space; the feature vector is a vector formed by arranging multiple feature parameters (such as amplitude, frequency, waveform distortion rate, etc.) of the fault response signal in a preset order, which is used to quantitatively characterize the fault signal features; the cosine similarity is an index that measures the degree of similarity of vectors by calculating the cosine value of the angle between two feature vectors, and the value range is [-1,1]. The closer the value is to 1, the more similar the two vectors are, and the closer it is to -1, the greater the difference is; the discrimination index is a parameter used to quantify the degree of distinguishability between different fault types, and it is negatively correlated with the cosine similarity, that is, the smaller the cosine similarity, the larger the discrimination index, and the easier it is to distinguish the fault type.

[0073] Specifically, for each candidate installation point, the system extracts its response signal features (such as voltage amplitude changes, current frequency shifts, waveform peak values, etc.) under all preset fault types from the fault fingerprint database, and organizes the feature parameters of each fault type into a feature vector. Then, using the dimensions of these feature vectors (e.g., "amplitude-frequency-distortion rate") as coordinate axes, a unique feature vector space is constructed for that candidate installation point, ensuring that each fault type corresponds to a unique vector in the space. Next, the system calculates the cosine similarity between the feature vectors corresponding to any two fault types in this space to determine the degree of similarity: if the angle between the feature vectors of the two faults is small and the cosine similarity is high, it indicates that the signal features are similar and easily confused; conversely, if the angle is large, the features are easily distinguished. Finally, based on the negative correlation between cosine similarity and discriminative power, the cosine similarity is converted into a discriminative power index (e.g., subtracting the cosine similarity from 1). The higher the index value, the stronger the candidate installation point's ability to distinguish between the two fault types.

[0074] In one specific embodiment, the system can extract the response features of a single candidate installation point from the fault fingerprint database under five preset fault types (such as internal breakdown, medium aging, oil leakage short circuit, parameter drift, and loose wiring). For each fault, four parameters are extracted: amplitude change rate, frequency offset, waveform distortion rate, and peak duration, forming five 4-dimensional feature vectors. 2. Using these four parameters as x, y, z, and w axes, a 4-dimensional feature vector space is constructed, and the five feature vectors are mapped into the space. 3. A vector calculation tool (such as Python's NumPy library) is called to calculate the cosine similarity between any two fault feature vectors (such as fault A and fault B, fault A and fault C, etc.). 4. Using the formula "discrimination index = 1 - cosine similarity", each cosine similarity is converted into a corresponding discrimination index, and finally, the set of discrimination indices between all fault types of the candidate installation point is obtained.

[0075] S207. The fault response difference and the discrimination index are weighted and fused to obtain the comprehensive fault identification capability index.

[0076] It is understandable that fault difference measure "whether a fault can be detected" (i.e., whether the difference between a fault and a normal state is obvious), and discrimination measure "whether a fault type can be identified" (i.e., whether the difference between different faults is obvious). Both are key elements of fault identification, but their importance may differ in different scenarios (e.g., in scenarios with high security requirements, "fault detection" has a higher priority, and the weight of difference can be set higher).

[0077] The system assigns reasonable weighting coefficients (e.g., 0.6 for difference and 0.4 for discrimination, or adjusts them based on expert experience) to the capacitor bank's operating scenario (e.g., substation main capacitor bank, distribution network filter capacitor bank) and fault handling requirements. Finally, the system merges the two into a comprehensive fault identification capability index by weighted summation (e.g., comprehensive index = difference × difference weight + discrimination index × discrimination weight). This index simultaneously reflects the candidate point's ability to "detect faults" and "identify faults".

[0078] S208. Calculate the information gain of candidate installation points for fault source localization based on information entropy theory, and multiply the information gain by the comprehensive fault identification capability index to obtain the fault localization value score.

[0079] Among them, information entropy theory refers to the theory of quantifying the uncertainty of random variables through "entropy". In this step, it is used to measure "the uncertainty of fault source location when the candidate installation point is not deployed". Information entropy is a parameter that characterizes the uncertainty of fault source location. The higher the entropy value, the more ambiguous the possible location / type of the fault source is, and the greater the difficulty of location. Information gain refers to the amount of reduction in information entropy of fault source location after deploying the candidate installation point, that is, the degree of uncertainty reduction. The greater the information gain, the greater the contribution of the candidate point to narrowing the fault source range.

[0080] Specifically, the system uses a fault fingerprint database to statistically analyze the probability of each fault source occurring and calculates the uncertainty using the information entropy formula (for example, if there are 10 possible fault sources with equal probabilities, the entropy value is high; if the probability of a certain fault source is much higher than the others, the entropy value is low). Then, it calculates the conditional information entropy after deploying the candidate installation point and classifying the fault sources based on their response signal characteristics: that is, the uncertainty of fault source location when the signal characteristics of the candidate point are known. The information gain is "initial information entropy - conditional information entropy," representing how much location uncertainty the candidate point can reduce. Finally, the information gain (location contribution) is multiplied by the comprehensive fault identification capability index (identification capability) to obtain the fault location value score.

[0081] In one specific embodiment, the system can statistically analyze the occurrence probability of all fault sources (e.g., 20 locations × 5 types = 100 fault sources) from the fault fingerprint database, and calculate the initial information entropy H(X) based on the information entropy formula (X is the fault source variable); for a single candidate installation point, extract its response signal features under all fault sources, classify fault sources with similar features into the same category (e.g., into 5 categories), calculate the probability of each category and the conditional entropy of the fault sources within the category, and sum them to obtain the conditional information entropy H(X|Y) (Y is the signal feature variable of the candidate point); calculate the information gain IG = H(X) - H(X|Y), and normalize IG (mapped to [0,1]); extract the comprehensive fault identification capability index of the candidate point obtained in step S207 (also normalized to [0,1]), and multiply the two to obtain the fault location value score (e.g., IG = 0.8, comprehensive index = 0.9, score = 0.72).

[0082] S209. Based on the historical operating data of the capacitor bank, statistical analysis is conducted on the frequency and type distribution of faults in the capacitor units of each region.

[0083] Historical operating data refers to various data recorded during the past operation of the capacitor bank, including the time of failure, location of failure, type of failure, operating environment parameters (such as temperature and humidity), load conditions, etc.; failure type distribution refers to the proportion of different failure types (such as internal breakdown, dielectric aging, oil leakage, etc.) in a certain area to the total number of failures in that area, which is used to identify the main failure risk types in that area.

[0084] Specifically, the system retrieves historical operating data from the capacitor bank's maintenance database. Then, according to preset area division rules (such as division by series / parallel branches or by physical cabinet), all capacitor units are categorized into their corresponding areas. Next, fault-related data for each area is statistically analyzed: on one hand, the fault occurrence frequency is calculated—the total number of faults in the area within a set time range (e.g., the past 5 years) is counted, and divided by the time length to obtain the fault frequency per unit time (e.g., "0.3 times / year"); on the other hand, the fault type distribution is statistically analyzed—all fault records in the area are classified by fault type, and the percentage of each type of fault in the total number of faults in the area is calculated (e.g., "internal breakdown accounts for 60%, dielectric aging accounts for 30%, and others account for 10%)).

[0085] S210. Determine the positional importance level of each capacitor unit based on its electrical connection relationship and functional importance level within the capacitor bank.

[0086] Electrical connection relationship refers to the connection method and position of capacitor unit in the series and parallel topology of capacitor bank (such as "series branch head unit", "parallel branch core unit", "backup branch unit", etc.), which determines the scope of impact of unit failure on the overall circuit; functional importance level refers to the level set according to the role of capacitor unit in the overall function of capacitor bank (such as reactive power compensation, filtering, voltage stabilization) (such as "core functional unit", "auxiliary functional unit", "backup functional unit"). The higher the level, the greater the impact of the unit on the overall function; positional importance level refers to the importance classification set for the position of capacitor unit based on the combination of electrical connection relationship and functional importance level (such as "special level", "level one", "level two", "level three"). It is used to quantify the degree of impact of unit failure on the system. The higher the level, the more serious the chain reaction that unit failure may cause, and the more important it is to monitor.

[0087] Specifically, the system acquires the electrical connection relationships of each capacitor unit: for example, a unit fault in a series branch may cause the entire branch to lose power, a core unit fault in a parallel branch may cause load distribution imbalance, and a backup unit fault has a smaller impact on current operation. Based on this, the structural impact range of a unit fault is initially determined. Subsequently, combined with the overall functional positioning of the capacitor bank (e.g., in a capacitor bank used for reactive power compensation, the unit undertaking the main compensation function is the core functional unit, and the unit undertaking auxiliary compensation is the auxiliary unit), the functional importance level of each unit is set. Finally, a comprehensive evaluation model of "electrical connection relationship + functional importance level" is established: for example, "series first-end unit + core function" corresponds to "special level" positional importance (the greatest impact from a fault), "parallel end unit + auxiliary function" corresponds to "level two" (moderate impact), and "backup unit + backup function" corresponds to "level three" (least impact), thereby determining the positional importance level of each capacitor unit.

[0088] S211. The fault risk weights within a preset range around each candidate installation point are calculated by weighting the fault occurrence frequency and the importance level of the location.

[0089] The preset range refers to the physical or electrical coverage area (such as a physical range with a radius of 1 meter, or the range of 3 adjacent units in the electrical connection relationship) that the system pre-sets centered on the candidate installation point, which is used to define the capacitor unit area that the candidate point can effectively monitor.

[0090] Specifically, the system defines a preset monitoring range for each candidate installation point (e.g., a physical range of 1.5 meters based on the sensor's signal acquisition radius, or a range of two adjacent series units + three parallel units based on electrical signal transmission efficiency), determining the set of capacitor units that the candidate installation point can effectively monitor. Then, the system extracts the fault occurrence frequency of all units within this range from the statistical results of step S209, and extracts the location importance level of each unit from the results of step S210 (converting the level into a quantitative score, such as Special Level = 4, Level 1 = 3, Level 2 = 2, Level 3 = 1). Next, weights are assigned to the fault occurrence frequency and location importance level (e.g., frequency weight 0.4, importance weight 0.6, which can be adjusted according to the operation and maintenance strategy), and the "Unit Risk Score = Fault Occurrence Frequency × 0.4 + Location Importance Score × 0.6" is calculated for each unit. Finally, the average or maximum value of the risk scores of all units within the preset range of the candidate point is taken as the fault risk weight of the candidate installation point—if the units within the range generally have high risk, the candidate point has a high weight and should be prioritized for deployment to cover high-risk areas.

[0091] S212. Calculate the degree of overlap between each candidate installation point and the monitoring range of the deployed sensors based on the monitoring range and sensitivity of the deployed sensors, and obtain the monitoring redundancy of each candidate installation point.

[0092] Among them, deployed sensors refer to sensors whose installation locations have been determined in the current deployment plan (this may be empty in the initial stage and will be gradually added as deployment progresses); monitoring range refers to the physical or electrical range within which the sensor can effectively collect signals and achieve fault monitoring (determined by the sensor model, installation location, and signal attenuation characteristics); sensitivity refers to the minimum signal change that the sensor can detect; the higher the sensitivity, the wider the effective boundary of the monitoring range; overlap refers to the proportion of the intersection area between the monitoring range of the candidate installation point and the monitoring range of the deployed sensors to the monitoring range of the candidate installation point; monitoring redundancy refers to a parameter that quantifies the degree to which the candidate installation point is covered by the deployed sensors, and its value range is usually [0,1]. The closer the value is to 1, the higher the overlap and the stronger the redundancy, and the smaller the improvement in monitoring performance when adding this candidate point.

[0093] Specifically, the system acquires key parameters of deployed sensors, including the installation location, monitoring range (e.g., physical radius of 1.2 meters), and sensitivity (e.g., ability to detect voltage changes of 0.01V) of each sensor. Based on these parameters, it delineates the effective monitoring area of ​​each deployed sensor in a 3D topology model (the higher the sensitivity, the wider the area boundary can be). Then, for each candidate installation point, its own effective monitoring area is also delineated (based on the same parameters of the sensors to be deployed). Next, the intersection area (or number of electrical coverage units) of the candidate installation point's monitoring area and the monitoring areas of all deployed sensors is calculated and divided by the candidate installation point's own monitoring area (or number of coverage units) to obtain the degree of overlap. Finally, the degree of overlap is directly used as the monitoring redundancy—if 90% of the candidate point's monitoring area is already covered by deployed sensors, its redundancy is 0.9, indicating that adding this point will result in extremely low monitoring gain and should be prioritized; if there is no overlap, the redundancy is 0, indicating that this point can fill a monitoring blind spot and should be given priority.

[0094] S213. Multiply the fault location value score by the fault risk weight to obtain the weighted fault location value of the candidate installation point.

[0095] Understandably, the fault location value score is the candidate point's "capability indicator" (whether it can perform location tasks effectively), while the fault risk weight is the "scenario requirement indicator" (whether location work is necessary in the area). Multiplying the two reflects "how much location capability the candidate point can provide in a scenario requiring location." Subsequently, the fault location value score for each candidate installation point (e.g., candidate point E's score is 0.85, normalized to [0,1]) and its corresponding fault risk weight (e.g., E's weight is 2.036; if not normalized, it needs to be standardized to [0,1] first, e.g., 0.7 after standardization) are retrieved. Finally, the standardized score is directly multiplied by the weight to obtain the weighted fault location value—e.g., E's weighted value = 0.85 × 0.7 = 0.595, indicating that the candidate point has a high location value in high-risk areas; if a candidate point has a score of 0.9 but a weight of 0.2, the weighted value = 0.18, indicating strong capability but lacking support from high-risk scenarios, resulting in lower value.

[0096] S214. Based on the preset cost-benefit function and effective monitoring value calculation, the monitoring benefit per unit cost is obtained, and then the deployment marginal benefit of candidate installation points is obtained.

[0097] Specifically, the system calculates the effective monitoring value based on the monitoring redundancy in step S212. This is achieved by setting a redundancy penalty coefficient (e.g., "penalty coefficient = 1 - redundancy", where a redundancy of 0.5 corresponds to a penalty coefficient of 0.5). The weighted fault location value from step S213 is multiplied by the penalty coefficient (e.g., weighted value 0.595 × penalty coefficient 0.5 = 0.2975) to obtain the effective monitoring value. This value has already deducted the ineffective gains from redundancy, making it more closely aligned with the actual added value. Subsequently, the deployment cost of each candidate installation point is obtained (including but not limited to sensor procurement cost, installation cost, and subsequent maintenance cost; for example, the total deployment cost of candidate point H is 20,000 yuan). Then, a preset cost-benefit function is called (usually "unit cost monitoring benefit = effective monitoring value / deployment cost") to calculate the monitoring benefit obtainable per unit cost (e.g., 0.2975 / 2 = 0.14875 / 10,000 yuan). Finally, the monitoring benefit per unit cost is directly taken as the marginal benefit of deployment—the marginal benefit of candidate point H is 0.14875 / 10,000 yuan, which means that every 10,000 yuan invested can obtain 0.14875 effective monitoring value; if the marginal benefit of another candidate point I is 0.2 / 10,000 yuan, its cost performance is higher and it should be deployed first.

[0098] The capacitor bank fault monitoring system of this invention is applied to electronic equipment. Figure 3 A schematic diagram of the architecture of an electronic device suitable for implementing embodiments of the present invention is shown.

[0099] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0100] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions (computer programs), or by instructions (computer programs) controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. The electronic device of this embodiment includes a storage medium and a processor, wherein the storage medium stores multiple instructions that can be loaded by the processor to execute any step of the method provided in the embodiments of the present invention.

[0101] Specifically, the storage medium and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more signal lines. The storage medium stores computer-executable instructions that implement data access control methods, including at least one software functional module that can be stored in the storage medium in the form of software or firmware. The processor executes various functional applications and data processing by running the software program and module stored in the storage medium. The storage medium can be, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The storage medium stores the program, and the processor executes the program after receiving the execution instructions.

[0102] Furthermore, the software programs and modules within the aforementioned storage medium may also include an operating system, which may include various software components and / or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components. The processor may be an integrated circuit chip with signal processing capabilities. The aforementioned processor may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc., which can implement or execute the methods, steps, and logic flowcharts disclosed in this embodiment. The general-purpose processor may be a microprocessor or any conventional processor.

[0103] Since the instructions stored in the storage medium can execute the steps in any of the methods provided in the embodiments of the present invention, the beneficial effects of any of the methods provided in the embodiments of the present invention can be achieved, as detailed in the preceding embodiments, and will not be repeated here.

[0104] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A sensor optimization deployment method for fault location within a capacitor bank, applied to a capacitor bank fault monitoring system, characterized in that, The method includes: Based on the topological parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensors, candidate installation points for the sensors are determined; Based on the electrical simulation model, the response signal characteristics of electrical parameters of each capacitor unit in the capacitor bank at each candidate installation point under a preset fault are simulated, and a fault fingerprint database including the mapping relationship between the fault source and the response signal characteristics is generated. The difference between the response signal characteristics of each candidate installation point under different preset faults and the response characteristics under normal conditions is calculated based on the fault fingerprint database. The fault location value score is calculated based on the difference and the distinguishability of the fault type. The fault location value score is used to characterize the fault distinguishability of the candidate installation point. Based on the fault location value score, the fault risk weight of the candidate installation point, and the monitoring redundancy of the selected installation point, the deployment marginal benefit of each candidate installation point is calculated. The deployment marginal benefit is used to characterize the monitoring performance improvement brought about by adding candidate installation points on the basis of existing sensor deployment. The fault risk weight is determined according to the historical failure rate and the location importance level. The monitoring redundancy indicates the degree to which the candidate installation point is covered and monitored by the deployed sensors. According to the order of deployment marginal benefits from high to low, target installation points that meet the preset cost constraints are selected from the candidate installation points to form a sensor optimization deployment scheme. The sensor optimization deployment scheme includes the position coordinate instructions of the target installation point in the three-dimensional topology of the capacitor bank.

2. The method according to claim 1, characterized in that, The step of determining candidate installation points for sensors based on the topological parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensors specifically includes: Based on the three-dimensional topology model of the capacitor bank, identify all electrical connection nodes between capacitor units; Based on the physical size limitations and installation space requirements of the sensor, installable nodes that meet the mechanical installation conditions are selected from all the electrical connection nodes; Based on the electromagnetic field distribution simulation results, target nodes located in the preset strong electromagnetic interference area are eliminated from the installable nodes to obtain candidate installation points for the sensor.

3. The method according to claim 1, characterized in that, Following the step of determining candidate installation points for the sensors based on the topology parameters of the capacitor bank to be monitored and the physical deployment conditions of the sensors, the method further includes: Establish an electrical distance matrix between candidate installation points and each capacitor unit based on the spatial distribution of candidate installation points; The sensitivity level of each candidate installation point to faults at different locations is determined based on the electrical distance matrix; A preset number of target candidate installation points are selected sequentially from high to low sensitivity levels for simulation to reduce the amount of simulation computation. The simulation results of the target candidate installation points are used to generate the fault fingerprint database.

4. The method according to claim 1, characterized in that, Before the step of calculating the deployment marginal benefit of each candidate installation point based on the fault location value score, the fault risk weight of the candidate installation points, and the monitoring redundancy of the selected installation points, the method further includes: Based on historical operating data of the capacitor bank, the frequency and type distribution of faults in capacitor units in each region were statistically analyzed. The positional importance level of each capacitor unit is determined based on its electrical connection relationship and functional importance level within the capacitor bank. The fault risk weight within a preset range around each candidate installation point is calculated by weighting the fault occurrence frequency and the location importance level. The degree of overlap between each candidate installation point and the monitoring range of the deployed sensors is calculated based on the monitoring range and sensitivity of the deployed sensors, thus obtaining the monitoring redundancy of each candidate installation point.

5. The method according to claim 4, characterized in that, The step of calculating the deployment marginal benefit of each candidate installation point based on the fault location value score, the fault risk weight of the candidate installation points, and the monitoring redundancy of the selected installation points specifically includes: The fault location value score is multiplied by the fault risk weight to obtain the weighted fault location value of the candidate installation point; A redundancy penalty coefficient is calculated based on the monitoring redundancy, and the redundancy penalty coefficient is negatively correlated with the monitoring redundancy. The effective monitoring value is obtained by multiplying the weighted fault location value by the redundancy penalty coefficient. The monitoring benefit per unit cost is calculated based on the preset cost-benefit function and the effective monitoring value. The monitoring benefits per unit cost are used as the marginal benefits of deploying candidate installation points.

6. The method according to claim 1, characterized in that, The step of calculating the fault location value score based on the difference and the distinguishability of the fault type specifically includes: Calculate the Euclidean distance between the response signal characteristics of the candidate installation point under each preset fault and the response characteristics under normal conditions to obtain the fault response difference degree; Construct a feature vector space for the response features of each candidate installation point under different fault types, and calculate the cosine similarity between feature vectors of different fault types; The discriminant index between fault types is calculated based on the cosine similarity, and the discriminant index is negatively correlated with the cosine similarity. The fault response difference and the discrimination index are weighted and fused to obtain a comprehensive fault identification capability index; The information gain of candidate installation points for fault source localization is calculated based on information entropy theory, and the information gain is multiplied by the comprehensive fault identification capability index to obtain the fault localization value score.

7. The method according to claim 1, characterized in that, After the step of selecting target installation points that meet preset cost constraints from candidate installation points according to the order of deployment marginal benefits from high to low, and forming an optimized sensor deployment scheme, the method further includes: Based on the location and monitoring range of the target installation point in the sensor optimization deployment scheme, calculate the electrical distance from each capacitor unit to the nearest sensor; Capacitor cells with an electrical distance greater than a preset distance threshold are marked as weak monitoring cells, and the proportion of such weak monitoring cells is counted. When the proportion of the number of the monitored weak units exceeds a preset safety threshold, the supplementary installation point that covers the most monitored weak units is selected from the remaining candidate installation points and added to the target installation point set.

8. A capacitor bank fault monitoring system, characterized in that, The system includes: one or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is run on the capacitor bank fault monitoring system, it causes the system to perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the capacitor bank fault monitoring system, it causes the system to perform the method as described in any one of claims 1-7.

Images