An electrical detection method and system based on artificial intelligence

Abstract

The application provides an electrical detection method and system based on artificial intelligence, applied to the technical field of electrical detection, and abnormality is recognized by acquiring real-time data, determining working conditions, matching historical normal data, and comparing difference degrees, effectively solving the problems that normal fluctuations and abnormal states are difficult to distinguish in the prior art, and early hidden faults are difficult to find, and having the advantages that the electrical detection method and system based on artificial intelligence can adapt to a complex dynamic operation environment of a microgrid, effectively process multi-source heterogeneous data, and improve the accuracy and robustness of electrical equipment anomaly detection.

Description

Technical Field

[0001] This application relates to the field of electrical testing technology, and in particular to an electrical testing method and system based on artificial intelligence. Background Technology

[0002] Microgrids, as a new type of power system, consist of distributed power sources, energy storage systems, load and control systems, exhibiting high flexibility and complexity. During the operation of a microgrid, continuous condition monitoring and fault diagnosis of the numerous electrical devices it contains are required to ensure the safe and stable operation of the system.

[0003] However, the distributed nature of microgrid devices leads to scattered data collection points, and data transmission may suffer from delays, packet loss, or inconsistencies, making it difficult to obtain synchronized and complete data. Furthermore, operational data from different devices and sensors often exhibit multi-source heterogeneity, including different data formats, sampling rates, and timestamps, which increases the difficulty of data integration and preprocessing.

[0004] Furthermore, the operation mode of microgrids is influenced by the external environment and internal dispatch strategies, exhibiting complex, varied, and dynamically changing characteristics. Under different operating modes and environmental conditions, the range of normal operating parameters and data patterns of equipment differ significantly. This dynamically changing normal mode makes it difficult for traditional detection methods based on fixed thresholds or static rules to accurately distinguish between normal fluctuations and abnormal states, easily leading to false alarms or missed alarms. In complex and dynamic operating environments, electrical equipment may exhibit early and subtle signs of faults.

[0005] These early faults typically manifest as slight changes in equipment operating parameters or subtle alterations in the relationships between multiple parameters, rather than drastic exceedances of a single parameter. These weak anomalous signals are easily masked by normal data fluctuations, environmental changes, or communication noise, making them difficult to detect through simple parameter monitoring or threshold judgment. Furthermore, the complex inter-coupling relationships between devices within a microgrid mean that an anomaly in one device may be caused by the status of other devices, system control, or external factors, further complicating accurate fault location and cause diagnosis.

[0006] Existing technologies are insufficient in addressing the above challenges. They are unable to effectively process data with delays, packet loss, and heterogeneity in the complex dynamic environment of microgrids, accurately identify early, hidden equipment anomalies that manifest as subtle changes or changes in correlations between multiple devices / parameters, and perform effective fault location and cause diagnosis.

[0007] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0008] In view of the shortcomings of the prior art, this application provides an artificial intelligence-based electrical detection method and system, which has the advantages of being able to adapt to the complex dynamic operating environment of microgrids, effectively processing multi-source heterogeneous data, and improving the accuracy and robustness of electrical equipment anomaly detection.

[0009] In a first aspect, an electrical testing method based on artificial intelligence is provided, the method comprising the steps of:

[0010] S1: Obtain real-time operating status data of electrical equipment in the microgrid, and determine the current real-time operating condition of the microgrid based on the real-time operating status data;

[0011] S2: Based on the real-time operating conditions, match the corresponding historical normal operating status data from the historical normal mode library;

[0012] S3: Compare the real-time operating status data with the historical normal operating status data to determine the degree of difference between the real-time operating status data and the historical normal operating status data;

[0013] S4: Identify electrical equipment anomalies in the microgrid based on the degree of difference.

[0014] This application proposes an artificial intelligence-based electrical detection method that can adapt to the complex and dynamic operating environment of microgrids, effectively process multi-source heterogeneous data, and improve the accuracy and robustness of electrical equipment anomaly detection.

[0015] Furthermore, step S1 includes:

[0016] S11: Obtain the original operating status data of electrical equipment in the microgrid. The original operating status data includes multiple sets of status indicators used to characterize multiple operating condition dimensions. Each set of status indicators comes from multiple distributed electrical devices in the microgrid.

[0017] S12: For at least one operating condition dimension, determine whether there is a conflict between the corresponding status indicators;

[0018] S13: If a conflict exists, then according to the preset device permission level, select the status indicator corresponding to the device with the highest permission from the conflicting status indicators, and use it as the determining indicator for the operating condition dimension.

[0019] S14: Determine the real-time operating condition based on the determination indicators of the operating condition dimension.

[0020] This application proposes an artificial intelligence-based electrical detection method that can effectively handle conflicts in multi-source heterogeneous data and improve the accuracy of real-time operating condition determination.

[0021] Furthermore, step S13 includes:

[0022] S131: For conflicting operating condition dimensions, obtain the preset device permission list;

[0023] S132: Determine the validity of the status indicators of each device in the list according to the priority of the device permission list;

[0024] S133: Select the device with the highest priority in the device permission list and whose status indicator data is determined to be valid, and use the status indicator corresponding to the device as the determining indicator of the operating condition dimension.

[0025] This application proposes an artificial intelligence-based electrical testing method that further refines the conflict handling logic to ensure the reliability of the operating condition determination.

[0026] Furthermore, step S2 includes:

[0027] S21: Based on the real-time operating conditions, obtain multiple adjacent historical normal operating status data from the historical normal mode library;

[0028] S22: Based on the proximity between the real-time operating condition and the historical operating condition corresponding to each of the multiple adjacent historical normal operating condition data, determine the weight of each of the adjacent historical normal operating condition data.

[0029] S23: Generate a normal operation status data reference based on multiple adjacent historical normal operation status data and their respective weights;

[0030] S24: The normal operating status data is used as a reference for historical normal operating status data that is matched with the real-time operating conditions.

[0031] This application proposes an artificial intelligence-based electrical testing method that can dynamically match historical normal data according to real-time operating conditions, thereby improving the accuracy of normal mode references.

[0032] Furthermore, step S22 includes:

[0033] S221: Obtain multiple dimensions constituting the real-time operating condition, and for each dimension, calculate the dimensional deviation between the real-time operating condition and each adjacent historical operating condition.

[0034] S222: According to the preset dimension calibration rules, the deviations of each dimension are calibrated to generate calibration deviations;

[0035] S223: Combine the calibration deviations for the same nearby historical operating conditions to generate a comprehensive distance characterizing the proximity between the real-time operating condition and the nearby historical operating conditions;

[0036] S224: Based on the comprehensive distance, determine the weight corresponding to the nearby historical normal operation status data.

[0037] Furthermore, step S222 includes:

[0038] S2221: For each dimension constituting the real-time operating condition, determine its statistical distribution characteristics in the historical normal operating condition data;

[0039] S2222: Based on the statistical distribution characteristics determined by the dimension, the dimension deviation is processed to generate the dimensionless calibration deviation.

[0040] Furthermore, step S23 includes:

[0041] S231: The multiple data parameters constituting the normal operating state data are divided into a basic parameter group and a derived parameter group according to a preset physical constraint relationship;

[0042] S232: Weighted aggregation of each parameter in the basic parameter group with the corresponding parameter values ​​in multiple adjacent historical normal operation status data and their weights to generate aggregated basic parameters;

[0043] S233: Based on the physical constraint relationship and the basic aggregation parameters, the parameter values ​​of the derived parameter group are calculated;

[0044] S234: Combine the aggregated basic parameters with the parameter values ​​of the derived parameter group to generate the normal operation status data reference.

[0045] Furthermore, step S3 includes:

[0046] S31: Based on the real-time operating conditions, analyze the historical normal operating status data in the historical normal mode library that are adjacent to the real-time operating conditions to determine the normal deviation boundary that characterizes the normal fluctuation range under the real-time operating conditions.

[0047] S32: Calculate the instantaneous deviation between the real-time operating status data and the historical normal operating status data;

[0048] S33: Compare the instantaneous deviation with the normal deviation boundary, and quantify the portion of the instantaneous deviation that exceeds the normal deviation boundary as the degree of difference.

[0049] Furthermore, step S4 includes:

[0050] S41: Obtain multiple differences generated at consecutive time points to generate a difference time series;

[0051] S42: Determine whether the difference time series meets the preset continuous exceedance condition and whether it meets the preset growth trend condition;

[0052] S43: When the difference time series meets the continuous over-limit condition or the growth trend condition, an electrical equipment anomaly in the microgrid is identified.

[0053] Secondly, an artificial intelligence-based electrical testing system is provided for implementing the method described in any of the above claims, the system comprising:

[0054] Acquisition module: Acquires real-time operating status data of electrical equipment in the microgrid, and determines the current real-time operating condition of the microgrid based on the real-time operating status data;

[0055] Matching module: Based on the real-time operating conditions, it matches the corresponding historical normal operating status data from the historical normal mode library;

[0056] Comparison module: compares the real-time operating status data with the historical normal operating status data to determine the degree of difference between the real-time operating status data and the historical normal operating status data;

[0057] Identification module: Based on the degree of difference, identify abnormalities in electrical equipment in the microgrid.

[0058] Beneficial effects: The electrical detection method and system based on artificial intelligence proposed in this application identify anomalies by acquiring real-time data, determining operating conditions, matching historical normal data, and comparing differences. This effectively solves the problems in the prior art of distinguishing between normal fluctuations and abnormal states and of detecting early hidden faults. It has the advantages of being able to adapt to the complex dynamic operating environment of microgrids, effectively processing multi-source heterogeneous data, and improving the accuracy and robustness of electrical equipment anomaly detection. Attached Figure Description

[0059] Figure 1 This is a flowchart of an artificial intelligence-based electrical testing method proposed in this application.

[0060] Figure 2 This is a structural diagram of an artificial intelligence-based electrical detection system proposed in this application.

[0061] Figure 3 This is an architecture diagram of an artificial intelligence-based electrical testing system proposed in this application.

[0062] Labeling explanation: 201, Acquisition module; 202, Matching module; 203, Comparison module; 204, Recognition module. Detailed Implementation

[0063] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0064] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0065] Please refer to Figure 1 An artificial intelligence-based electrical testing method, comprising the following steps:

[0066] S1: Obtain real-time operating status data of electrical equipment in the microgrid, and determine the current real-time operating condition of the microgrid based on the real-time operating status data;

[0067] S2: Based on real-time operating conditions, match the corresponding historical normal operating status data from the historical normal mode library;

[0068] S3: Compare the real-time operating status data with the historical normal operating status data to determine the degree of difference between the real-time operating status data and the historical normal operating status data;

[0069] S4: Identify electrical equipment anomalies in the microgrid based on the degree of difference.

[0070] Real-time operating status data refers to the set of operating parameters collected by each electrical device in the microgrid at the current moment, such as voltage, current, power, frequency, temperature, operating mode, and state of charge. It can be realized through sensor networks and data acquisition systems, and its main purpose is to provide raw information on the current operating status of the microgrid.

[0071] Real-time operating conditions refer to the overall operating mode or state of a microgrid at the current moment, such as grid-connected mode, off-grid mode, specific load level, specific power generation combination, etc. It is determined based on real-time operating status data, such as by analyzing parameters such as total power flow, grid-connected switch status, and main equipment operating mode. It is mainly to provide the operating background information of the current data and provide context for subsequent normal state comparison.

[0072] The historical normal mode library stores the status data patterns of equipment during normal operation under different historical operating conditions of the microgrid. These patterns can be raw data samples, statistical features, or models trained based on historical data. The main purpose is to provide a reference for the status of equipment under various normal operating conditions.

[0073] Matching the corresponding historical normal operating status data means finding or generating the normal status data that is closest to or most relevant to the current operating condition from the historical normal pattern library based on the current real-time operating condition. This can be done by directly searching for historical data that matches the current operating condition label, or by generating it through a weighted average based on neighboring operating condition data. The main purpose is to obtain a normal status benchmark that matches the current real-time operating condition, overcoming the limitations of static thresholds or models.

[0074] The degree of difference refers to the degree of deviation between real-time operating status data and the matched historical normal operating status data. This can be quantified by calculating the distance between the two, statistical deviation, etc. Its main purpose is to quantify the degree of deviation of the real-time status from the normal state under the current working conditions, as a basis for judging anomalies.

[0075] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0076] Real-time operational status data can be collected by sensors deployed on various devices in the microgrid and aggregated to the data processing unit via a communication network.

[0077] The determination of real-time operating conditions can be based on key parameters collected, such as the total bus voltage, frequency, connection status with the main power grid, and operating modes of main power sources and load equipment, and can be made using preset rules or trained classification models.

[0078] The historical normal operating mode library can be stored on a local server or in a cloud database. It contains a set of normal operating data samples collected and cleaned and labeled under different typical operating conditions. When matching historical normal operating status data, the corresponding historical normal data samples under the current operating condition can be directly extracted from the library based on the currently determined operating condition label, or the statistical characteristics of the data under that operating condition can be extracted as a reference.

[0079] When comparing real-time data with historical normal data and calculating the degree of difference, the Mahalanobis distance between real-time data points and historical normal data sample sets can be calculated, or the probability that real-time data points fall outside the distribution of historical normal data can be calculated.

[0080] Anomaly detection can be achieved by setting a difference threshold. When the difference in real-time data exceeds this threshold, it is determined that there is a device anomaly.

[0081] Through the above-described solution, this application effectively addresses the challenges posed by the complex and dynamic changes in the operating conditions of microgrids. By determining the operating conditions in real time and dynamically matching them with historical normal data as a comparison benchmark, the solution overcomes the inaccuracy of traditional static detection methods in dynamic environments, reducing false alarms and missed alarms. By comparing and quantifying the differences based on the normal state matched to the current operating conditions, the solution can capture early abnormal signals that appear as slight deviations in complex contexts, improving the timeliness and accuracy of anomaly detection. This solution provides an electrical detection framework adapted to the dynamic operating environment of microgrids.

[0082] Furthermore, step S1 includes:

[0083] S11: Obtain the raw operating status data of electrical equipment in the microgrid. The raw operating status data includes multiple sets of status indicators used to characterize multiple operating condition dimensions. Each set of status indicators comes from multiple distributed electrical devices in the microgrid.

[0084] S12: For at least one operating condition dimension, determine whether there is a conflict between the corresponding status indicators;

[0085] S13: If a conflict exists, then based on the preset device permission level, select the status indicator corresponding to the device with the highest permission from the conflicting status indicators as the determining indicator for the operating condition dimension.

[0086] S14: Determine real-time operating conditions based on indicators determined by the operating condition dimension.

[0087] Raw operating status data refers to unprocessed or preliminarily processed operating data collected directly from various electrical devices within the microgrid. This data can include various parameters such as voltage, current, power, frequency, temperature, operating mode, and switch status. These data usually include timestamps and device identifiers.

[0088] The operating condition dimension refers to the key aspects that describe the overall or partial operating status of a microgrid, such as grid-connected / off-grid status, charging / discharging status, generation mode, load level, etc., which can be characterized by one or more status indicators.

[0089] Status indicators refer to specific data parameters used to quantify or describe a particular operating condition dimension. They can be a numerical value (such as frequency or power) or a discrete value (such as an operating mode code). Distributed electrical equipment refers to various electrical devices that are distributed in a microgrid, such as photovoltaic inverters, energy storage converters, battery management systems, smart meters, and circuit breakers.

[0090] A conflict refers to a situation where status indicator data from different devices are inconsistent for the same operating condition dimension. For example, one device reports a grid-connected status, while another device reports an off-grid status.

[0091] Device access hierarchy refers to the preset data credibility or priority ranking for different devices within a microgrid. It can be set based on the device's measurement accuracy, reliability, criticality in the system, or the authority of the data source.

[0092] A defined indicator refers to a single, reliable status indicator data that is ultimately selected to characterize a specific operating condition dimension after conflict determination and resolution.

[0093] This method effectively addresses potential data conflicts in distributed microgrid environments by refining the process of acquiring real-time operating status data and determining real-time operating conditions.

[0094] First, raw operational status data from multiple distributed devices is acquired. This data is organized into multiple sets of status indicators to characterize various operational dimensions, clarifying the dispersed and multi-dimensional nature of the data sources. Next, for at least one operational dimension, the system systematically determines whether there are conflicts between its corresponding status indicators. This step proactively identifies data inconsistencies that may be caused by data heterogeneity or transmission issues. When a conflict is detected, the method does not simply discard data or average it. Instead, based on a preset device permission hierarchy, it selects the data corresponding to the device with the highest permission from the conflicting status indicators as the definitive indicator for that dimension. This priority-based conflict resolution mechanism ensures that in cases of data inconsistency, a more reliable or authoritative data source can be trusted, thereby obtaining a more credible single indicator to characterize that operational dimension.

[0095] Ultimately, based on these determined indicators of operational condition dimensions that have undergone conflict resolution or are inherently conflict-free, the current real-time operational condition is accurately determined.

[0096] Through this series of steps, this method can extract accurate and reliable operating condition information from potentially conflicting raw data, providing a solid foundation for subsequent anomaly detection and improving the accuracy and reliability of the entire detection method.

[0097] In some embodiments of this application, step S13 is proposed to select the status indicator corresponding to the device with the highest authority among conflicting status indicators, based on the preset device authority level, as the indicator for determining the operating condition dimension. However, selecting solely based on the device authority level may result in invalid current status indicator data of the selected device, such as sensor failure or communication interruption, thus making it impossible to accurately determine the operating condition dimension and affecting the subsequent detection accuracy.

[0098] To address the above problem, step S13 further includes:

[0099] S131: For conflicting operating condition dimensions, obtain the preset device permission list;

[0100] S132: Determine the validity of the status indicators of each device in the list according to the priority of the device permission list;

[0101] S133: Select the device with the highest priority in the device permission list and whose status indicator data is deemed valid, and use the status indicator corresponding to the device as the determining indicator for the operating condition dimension.

[0102] In step S131, the device permission list is pre-configured with the priority order of status indicator data provided by different devices under specific operating conditions. Obtaining this list provides a structured basis for subsequent data selection. Further, in step S132, data validity determination may include, but is not limited to, checking whether the data timestamp is within a reasonable range (e.g., whether it is outdated), whether the data value is within a preset physical range, whether the data format is correct, or whether the device communication status is normal. By sequentially determining these parameters, the availability and reliability of data from devices with higher permissions can be prioritized. Finally, in step S133, the system starts searching from the device with the highest permission until it finds the first device providing valid data and adopts that data.

[0103] As a specific implementation method, when determining the "grid-connected / off-grid mode" operating condition dimension of a microgrid, data conflicts may exist from the main control system, grid-connected inverters, and energy storage converters. A device permission list can be preset, for example: main control system (priority 1), grid-connected inverter (priority 2), and energy storage converter (priority 3). When a data conflict is detected in this dimension, the system first retrieves this list. Then, it sequentially determines the data validity of the status indicators (e.g., mode status words) of each device. First, it checks the data of the main control system, determining whether its timestamp is within the most recent second and whether the mode status word is a preset valid value (e.g., 0 indicates grid connection, 1 indicates off-grid). If the main control system data is valid, the mode status word of the main control system is selected as the determining indicator. If the main control system data is invalid (e.g., communication interruption or data anomaly), the validity of the data of the grid-connected inverter is checked. If the grid-connected inverter data is valid, its mode status word is selected. If the grid-connected inverter data is also invalid, the validity of the data of the energy storage converter is checked, and so on. In this way, even if the data from the device with the highest privileges is compromised, the system can fall back to the next valid data source with the highest privileges, thereby ensuring the accurate determination of the on / offline mode.

[0104] By combining device permissions with data validity assessment and employing a priority-based approach, this solution overcomes the drawback of relying solely on permissions, which may result in the selection of invalid data. This ensures that the data used to determine operational status dimensions is both authoritative and reliable. This approach provides a more solid foundation for subsequent anomaly detection, improving the robustness and accuracy of the entire detection method.

[0105] Furthermore, step S2 includes:

[0106] S21: Based on real-time operating conditions, obtain multiple adjacent historical normal operating status data from the historical normal mode library;

[0107] S22: Determine the weight of each adjacent historical normal operating condition data based on the proximity between the real-time operating condition and the historical operating condition corresponding to each data in multiple adjacent historical normal operating condition data.

[0108] S23: Generate a reference for normal operation status data based on multiple adjacent historical normal operation status data and their respective weights;

[0109] S24: Use the normal operating status data as a reference, and use it as historical normal operating status data to match the real-time operating conditions.

[0110] Among them, real-time operating conditions refer to a set of key parameters or mode identifiers that characterize the current overall operating status of the microgrid. These parameters can be represented by a combination of parameters such as voltage level, power flow direction, grid connection / disconnection status, load level, and power generation output.

[0111] A historical normal mode library refers to a dataset that stores the historical operating status of a microgrid under various known normal operating modes and environmental conditions. It can be stored using a structured database, a time-series database, or a collection of data files.

[0112] Multiple neighboring historical normal operation status data refers to a set of historical data records selected from the historical normal mode library, whose corresponding historical operating conditions and current real-time operating conditions have a high degree of similarity under a certain metric. These data can be obtained using methods based on distance metrics (such as Euclidean distance) or similarity metrics (such as cosine similarity).

[0113] Proximity refers to the numerical value that quantifies the similarity or closeness between a real-time operating condition and a certain historical operating condition. It can be determined by calculating the distance or similarity index between operating condition parameters.

[0114] Weight refers to a numerical value assigned to each neighboring historical normal operation status data to represent its importance or influence in the process of generating normal operation status data references. It can be calculated using a function based on proximity (such as the reciprocal of the distance).

[0115] Normal operating status data reference refers to a synthetic data representation generated by integrating multiple nearby historical normal operating status data and their weights. This representation aims to more comprehensively reflect the typical characteristics and fluctuation range of the normal state under the current real-time operating conditions. It can be generated using weighted average, weighted median or other weighted aggregation statistical methods.

[0116] In a specific embodiment, the method can be implemented as follows: First, acquire the current real-time operating conditions, including key parameters such as the current grid connection status of the microgrid, total load power, total photovoltaic output, and energy storage charging and discharging power. Then, in the historical normal mode library, calculate the Euclidean distance between the current real-time operating conditions and each historical operating condition in the library, and select the K closest historical data records as multiple neighboring historical normal operating state data. Next, based on the Euclidean distance between the real-time operating conditions and these K neighboring historical operating conditions, use the reciprocal of the distance as the weight calculation function; the smaller the distance (i.e., the closer) the historical data, the greater its weight. For example, the weight can be calculated using the reciprocal of the distance. Subsequently, for each parameter in the normal operating state data (e.g., voltage, current, temperature, etc.), calculate a weighted average of its corresponding value in the K neighboring historical normal operating state data according to the calculated weight, generating a reference value for that parameter. Combining the weighted average reference values ​​of all parameters generates the normal operating state data reference. Finally, this generated normal operating status data reference is used as a comparison benchmark under the current real-time operating conditions for calculating the difference between the data and the real-time operating status data.

[0117] Furthermore, step S22 includes:

[0118] S221: Obtain multiple dimensions that constitute the real-time operating condition, and for each dimension, calculate the dimensional deviation between the real-time operating condition and each adjacent historical operating condition.

[0119] S222: Based on the preset dimension calibration rules, calibrate the deviations of each dimension to generate calibration deviations;

[0120] S223: Combine the calibration deviations for the same nearby historical operating conditions to generate a comprehensive distance characterizing the proximity between the real-time operating condition and the nearby historical operating conditions.

[0121] S224: Based on the comprehensive distance, determine the weight corresponding to the nearby historical normal operation status data.

[0122] Among them, the multiple dimensions constituting the real-time operating conditions refer to various parameters or characteristics that jointly describe the operating status of microgrid electrical equipment at a certain moment, such as voltage, current, power, temperature, frequency, operating mode status, and environmental parameters. Data of these dimensions are usually collected and aggregated from distributed devices.

[0123] Dimensional deviation refers to the difference between the value of a real-time operating condition in a specific dimension and the value of a neighboring historical operating condition in the same dimension. This difference can be calculated using methods such as absolute difference or relative difference.

[0124] Predefined dimension calibration rules refer to predefined functions used to convert the original deviation values ​​of different dimensions into comparable dimensionless values. The purpose is to eliminate the influence of differences in the dimensions and numerical ranges of data of different dimensions. For example, standardization or normalization rules based on statistical methods can be used.

[0125] Calibration bias refers to the dimensional deviation value after processing by preset dimensional calibration rules. These values ​​can be compared between different dimensions.

[0126] Combination refers to the mathematical aggregation of calibration deviation values ​​from multiple dimensions for the same nearby historical operating conditions to obtain a single value. This value comprehensively reflects the overall difference between the two operating conditions in a multi-dimensional space. For example, combination can be achieved by summation, weighted summation, or calculation of multi-dimensional spatial distance.

[0127] The composite distance is a single value obtained after combined calculations. It quantifies the overall proximity between the real-time operating condition and a certain nearby historical operating condition.

[0128] Determining the weight corresponding to the nearby historical normal operating status data based on the comprehensive distance means assigning a weight value to the historical normal operating status data associated with the historical operating condition based on the calculated comprehensive distance value and through a preset mapping relationship. This weight value is usually inversely proportional to the comprehensive distance, with a decrease in distance leading to an increase in weight.

[0129] In a specific embodiment, this solution can be implemented as follows:

[0130] First, obtain multiple dimensions that constitute the real-time operating condition, such as inverter output power, total battery pack voltage, and ambient temperature. For each dimension, calculate the dimensional deviation between the current value of the real-time operating condition and the corresponding value of each adjacent historical operating condition in that dimension; for example, the absolute difference between them can be calculated.

[0131] Next, these dimensional deviations are calibrated according to preset dimensional calibration rules. These calibration rules can employ standardization methods based on the statistical characteristics of historical normal data. For example, Z-score standardization can be applied to the deviation value of each dimension, i.e., subtracting the historical average of the deviation and dividing by the historical standard deviation, thus generating a dimensionless calibration deviation. Then, the calibration deviations of all dimensions for the same nearby historical operating condition are combined. For example, the Euclidean distance of the vector formed by these calibration deviations can be calculated, and this Euclidean distance is used as a comprehensive distance characterizing the proximity between the real-time operating condition and the historical operating condition.

[0132] Finally, based on the calculated comprehensive distance, the weight corresponding to the nearby historical normal operation status data is determined. For example, an inverse proportional function or an exponential decay function can be used to calculate the weight, such as the weight being proportional to the reciprocal of the comprehensive distance, or the weight decreasing exponentially as the comprehensive distance increases, ensuring that a decrease in distance leads to an increase in the weight of historical data.

[0133] Furthermore, step S222 includes:

[0134] S2221: For each dimension that constitutes the real-time operating condition, determine its statistical distribution characteristics in the historical normal operating condition data;

[0135] S2222: Based on the statistical distribution characteristics determined by the dimension, the dimension bias is processed to generate a dimensionless calibration bias.

[0136] Statistical distribution characteristics refer to attributes that describe the distribution state of a data set. These may include, for example, the mean, median, standard deviation, variance, maximum value, minimum value, quantiles, or specific probability distribution model parameters. Determining statistical distribution characteristics can be achieved by performing statistical analysis on historical normal operating data.

[0137] Dimensionless calibration deviation refers to the deviation value that has been processed and no longer has the original physical dimensions, and whose numerical range has been unified or standardized.

[0138] In one embodiment, for each dimension constituting the real-time operating condition, a set of historical normal operating state data corresponding to the current real-time operating condition type can first be obtained from a historical normal mode library. Then, statistical analysis is performed on the data for each dimension in this set, calculating the mean (μ) and standard deviation (σ) of that dimension, and using these mean and standard deviation as the statistical distribution characteristics of that dimension. Next, for the original deviation (Δx) between the real-time operating condition and a neighboring historical operating condition in a certain dimension, the Z-score standardization method can be used to process it based on the mean (μ) and standard deviation (σ) of that dimension, generating the calibration deviation (z) for that dimension. The calculation formula can be z = (Δx - μ) / σ. In this way, the original deviation is converted into a dimensionless value in units of standard deviation, making the deviations of different dimensions comparable.

[0139] Furthermore, step S23 includes:

[0140] S231: The multiple data parameters constituting the normal operating status data are divided into a basic parameter group and a derived parameter group according to the preset physical constraint relationship;

[0141] S232: Take each parameter in the basic parameter group and the corresponding parameter value in multiple adjacent historical normal operation status data, and their weights, and perform weighted aggregation to generate aggregated basic parameters;

[0142] S233: Based on the physical constraint relationship and the basic aggregation parameters, the parameter values ​​of the derived parameter group are calculated;

[0143] S234: Combine the parameter values ​​of the aggregated basic parameters and the derived parameter groups to generate a reference for normal operating status data.

[0144] Among them, physical constraints refer to the physical laws or engineering principles that govern the data parameters of microgrid operation status, such as Ohm's law, Kirchhoff's law, and power conservation. These can be represented by mathematical formulas, physical models, or preset rule sets.

[0145] The basic parameter set refers to the set of parameters that serve as inputs or independent variables in physical constraints. These parameters can be directly measured or used as the basis for calculations, such as voltage, current, and frequency.

[0146] A derived parameter set refers to a set of parameters that serve as output or dependent variables in physical constraints. These parameters can be constructed using power, impedance, energy, and other parameters calculated from the basic parameters.

[0147] Weighted aggregation refers to the process of linearly or non-linearly combining multiple data values ​​according to preset weights. It can be achieved by weighted average, weighted summation, and other methods.

[0148] Aggregated basic parameters refer to the set of parameter values ​​obtained by weighting and aggregating the corresponding values ​​of each parameter in the basic parameter group from multiple nearby historical normal operation status data.

[0149] The reason why this application can generate physically consistent and reasonable normal operation status data references is that the method does not simply perform weighted aggregation of all parameters, but fully considers the physical constraint relationships between parameters.

[0150] First, by dividing the parameters into basic parameter groups and derived parameter groups based on the preset physical constraints, the dependencies between the parameters are clarified.

[0151] Next, each parameter in the basic parameter group is weighted and aggregated. This step fully utilizes the effective information from multiple neighboring historical normal data sets, and the weights reflect the degree of matching between these historical data and real-time operating conditions, thereby generating aggregated values ​​that represent the normal state of the basic parameters under the current operating conditions. Crucially, for the parameter values ​​in the derived parameter group, this application does not directly perform weighted aggregation, but rather calculates them based on the already aggregated basic parameters according to preset physical constraints. Because the derived parameters are calculated using physical laws based on the aggregated basic parameters, it ensures that the generated derived parameter values ​​satisfy physical constraints with the aggregated basic parameters, thus guaranteeing the physical consistency and rationality of the entire normal operating state data reference.

[0152] Finally, the aggregated basic parameters are combined with the calculated derived parameter values ​​to form the final reference data. This approach avoids the physical inconsistencies that may result from directly weighting and aggregating all parameters independently, enabling the generated reference data to more accurately represent the normal operating status under current real-time conditions and providing a more reliable benchmark for subsequent difference calculations and anomaly identification.

[0153] By combining this with the aforementioned step of matching historical normal operating status data based on real-time operating conditions, this application can generate a reference that reflects both historical normal state characteristics and is physically consistent, thereby improving the accuracy of anomaly detection.

[0154] Based on the above principles, a specific implementation of this application can be carried out as follows. Assume that the microgrid's operating status data includes three parameters: voltage (V), current (I), and power (P), and the known physical constraint relationship is P = V * I. In step S231, according to the physical constraint relationship P = V * I, voltage (V) and current (I) can be divided into a basic parameter group, and power (P) can be divided into a derived parameter group. In step S232, multiple adjacent historical normal operating status data are obtained, for example, data 1. Data 2 ..., data N and their respective weights , ... The voltage and current in the basic parameter group are weighted and aggregated to generate aggregated basic parameters: aggregated voltage. : Aggregating current : In step S233, based on the physical constraint relationship P = V * I and the aggregation fundamental parameters... , The power parameter values ​​in the derived parameter set are calculated: power calculation In step S234, the aggregation basic parameters are... With the calculated derived parameter values Combine and generate normal operating status data reference .

[0155] Furthermore, step S3 includes:

[0156] S31: Based on the real-time operating conditions, analyze the historical normal operating status data in the historical normal mode library that are adjacent to the real-time operating conditions to determine the normal deviation boundary that characterizes the normal fluctuation range under the real-time operating conditions.

[0157] S32: Calculate the instantaneous deviation between real-time operating status data and historical normal operating status data;

[0158] S33: Compare the instantaneous deviation with the normal deviation boundary, and quantify the portion of the instantaneous deviation that exceeds the normal deviation boundary as the degree of difference.

[0159] This scheme is a specific implementation that compares real-time operating status data with historical normal operating status data to determine the degree of difference. By introducing the concept of normal deviation boundary, it solves the problem that simple comparison is difficult to distinguish between normal fluctuations and abnormal changes in complex dynamic operating environments, thus improving the accuracy of difference measurement.

[0160] Specifically, based on the current real-time operating conditions, by analyzing historical normal operating status data adjacent to this operating condition in the historical normal mode database, the normal fluctuation range of electrical equipment operating status data parameters under this specific operating condition is dynamically determined, thereby establishing a normal deviation boundary characterizing the normal fluctuation range. The key to this process is that the boundary determination is based on real-time operating conditions and historical normal data analysis, enabling it to adapt to the dynamic changes in microgrid operating modes and environment.

[0161] At the same time, the instantaneous deviation between the current real-time operating status data and the historical normal operating status data matched or generated based on the real-time operating conditions is calculated. This is a preliminary measure of the difference between the current actual state and the ideal normal state.

[0162] Subsequently, the calculated instantaneous deviation is compared with the determined normal deviation boundary. Only when the instantaneous deviation exceeds this normal deviation boundary is an anomaly considered to exist, and the degree of exceeding the boundary is quantified as the final difference. This comparison and quantification method effectively filters out deviations within the normal fluctuation range, avoiding misjudging normal fluctuations as anomalies, thereby reducing the false alarm rate. At the same time, it can highlight those abnormal signals that exceed the normal range, providing a more reliable basis for subsequent anomaly identification.

[0163] By dynamically adjusting the normal deviation boundary using real-time operating conditions and nearby historical data, this scheme can more accurately reflect the normal fluctuation characteristics under the current operating conditions, making the calculation of the difference more precise, thereby effectively distinguishing normal fluctuations from early abnormal signals in complex dynamic environments.

[0164] In some embodiments described above, this application proposes identifying electrical equipment anomalies in a microgrid based on the difference between real-time operating status data and historical normal operating status data. Specifically, this anomaly identification based on difference involves comparing the instantaneous difference with a preset threshold; when the instantaneous difference exceeds the threshold, it is determined to be an anomaly. This allows for rapid response to sudden, severe anomalies. However, in its implementation, relying solely on the instantaneous difference at a single moment for anomaly judgment is insufficient to effectively distinguish between transient data deviations caused by normal fluctuations, environmental changes, communication noise, or data acquisition problems, and early, hidden, or slowly developing anomalies originating from the equipment itself. Especially for early fault indications that manifest as weak changes or subtle shifts in correlations in the data, rather than drastic exceedances of a single parameter, the instantaneous difference may be insufficient as a reliable basis for judgment, easily leading to false alarms or missed alarms, and hindering effective intervention in the early stages of the problem. Therefore, an anomaly identification method that better considers the characteristics of difference changing over time is needed to improve the detection capability of persistent or developing anomalies and reduce false alarms.

[0165] Therefore, further, step S4 includes:

[0166] S41: Obtain multiple dissimilarity values ​​generated at consecutive time points to generate a dissimilarity time series;

[0167] S42: Determine whether the difference time series meets the preset continuous exceedance condition and whether it meets the preset growth trend condition;

[0168] S43: When the time series of differences meets the conditions of continuous over-limit or growth trend, anomalies in electrical equipment in the microgrid are identified.

[0169] A difference time series refers to a sequence of multiple differences obtained at consecutive time points, arranged in chronological order. This can be achieved by storing the calculated difference values ​​in a queue, list, or database and associating them with timestamps.

[0170] The preset continuous exceedance condition refers to the condition used to determine whether the difference value in the difference time series continuously exceeds a certain threshold. It can be implemented by setting a time window length and / or a threshold for the number of exceedances. For example, it can be required that within the most recent N time points, the difference value exceeds the preset threshold for M consecutive points or accumulates to more than K points.

[0171] The preset growth trend condition refers to the condition used to determine that the difference time series shows an upward trend. It can be achieved by performing trend analysis on the difference time series, such as calculating the slope of the series, comparing the average values ​​of the first and last parts or segments of the series, or applying moving averages.

[0172] To more clearly illustrate the technical solution of this application, a specific embodiment is provided below: In step S41, the system can calculate the difference degree every second and store the difference degree values ​​of the most recent 60 time points (i.e., the most recent minute) in a circular buffer to form a difference degree time series. In step S42, the continuous exceedance condition can be set as follows: in the difference degree time series of the most recent 60 time points, the difference degree value exceeds the preset threshold A for 10 consecutive time points, or 40 time points out of the most recent 60 time points exceed the threshold A. Simultaneously, the growth trend condition can be set as follows: performing linear regression analysis on the difference degree time series of the most recent 60 time points, if the slope of the regression line is greater than the preset threshold B, then a growth trend is considered to exist. In step S43, when either the continuous exceedance condition or the growth trend condition is met, the system can identify an abnormality in the electrical equipment in the microgrid, for example, generating an alarm event and sending a notification.

[0173] Please refer to Figure 2 , Figure 3 An artificial intelligence-based electrical testing system for implementing any of the above methods, the system comprising:

[0174] Acquisition Module 20: Acquires real-time operating status data of electrical equipment in the microgrid, and determines the current real-time operating condition of the microgrid based on the real-time operating status data;

[0175] Matching module 202: Based on real-time operating conditions, it matches the corresponding historical normal operating status data from the historical normal mode library;

[0176] Comparison module 203: Compares real-time operating status data with historical normal operating status data to determine the degree of difference between real-time operating status data and historical normal operating status data;

[0177] Identification module 204: Identifies electrical equipment anomalies in the microgrid based on the degree of difference.

[0178] Among them, the acquisition module 201 refers to the functional unit responsible for collecting the operating data of microgrid equipment and performing preliminary processing to determine the current operating status of the system. It can be implemented by a data acquisition unit, a data preprocessing unit, an operating condition analysis unit, etc.

[0179] The matching module 202 refers to the functional unit that searches for and selects relevant normal status data from historical data based on the current operating conditions. It can be implemented using a database query unit, an index matching unit, a similarity calculation unit, etc.

[0180] The comparison module 203 is a functional unit that compares real-time data with matched historical normal data and quantifies the degree of deviation. It can be implemented using data analysis units, statistical calculation units, difference measurement units, etc.

[0181] The identification module 204 is a functional unit that determines whether there is a device abnormality based on the data comparison results. It can be implemented using a rule engine, an anomaly detection unit, a machine learning model, etc.

[0182] This system provides a modular architecture for implementing the aforementioned AI-based electrical testing method. The acquisition module 201 is responsible for the data acquisition and operating condition determination steps in the method. It can acquire real-time operating status data of distributed electrical equipment in the microgrid. This data may come from different devices, be heterogeneous, delayed, or lost. Based on this real-time operating status data, this module can process and determine the current real-time operating condition of the microgrid. By accurately acquiring and processing the raw data and extracting key operating condition information, this module provides a reliable input foundation for subsequent testing processes, especially in microgrid environments where data quality is limited. The determined real-time operating condition is then transmitted to the matching module 202.

[0183] The matching module 202 is responsible for implementing the historical normal data matching step in the method. Based on the real-time operating conditions determined by the acquisition module, it searches and matches historical normal operating status data corresponding to the current operating conditions from a pre-established historical normal mode library. Accurate matching of historical normal data is crucial given the complex and ever-changing operating modes of microgrids. This module, by matching based on real-time operating conditions, can find the historical data that best represents the current normal state as a reference, thereby adapting to the dynamically changing operating modes of the microgrid and avoiding misjudgments caused by comparing with static benchmarks.

[0184] The real-time operating status data provided by the acquisition module 201 and the historical normal operating status data provided by the matching module 202 are passed to the comparison module 203. The comparison module 203 is responsible for implementing the data comparison and difference determination steps in the method. It receives the real-time operating status data and the historical normal operating status data, compares them to quantify the deviation of the real-time data from the normal reference, and determines the difference. In the context of early microgrid anomalies potentially manifesting as subtle changes or changes in correlations, simple over-limit judgments are insufficient to detect problems. This module, through refined comparison, can transform the difference between real-time data and historical normal data into a measurable difference index, highlighting potential anomalies and providing a quantitative basis for anomaly identification. The calculated difference is then passed to the identification module 204.

[0185] The identification module 204 is responsible for implementing the anomaly identification step in the method. It determines whether there are anomalies in the electrical equipment of the microgrid based on the difference degree output by the comparison module. When microgrid faults may have persistent or trending characteristics, relying solely on instantaneous difference degree may be insufficient. This module analyzes the dynamic changes in difference degree, such as whether it continuously exceeds limits or shows an increasing trend, to more accurately identify early, hidden anomaly signals, distinguish between normal fluctuations and abnormal development, thereby improving the accuracy and robustness of anomaly detection and enabling timely discovery of equipment problems.

[0186] The entire system breaks down abstract methods and steps into specific, collaborative functional modules, forming a complete data processing and anomaly detection process. This enables the methods to be executed efficiently and accurately in real microgrid environments, overcoming the challenges faced by the methods themselves as abstract descriptions in actual deployment and operation.

[0187] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0188] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An electrical testing method based on artificial intelligence, characterized in that, The method includes the following steps: S1: Obtain real-time operating status data of electrical equipment in the microgrid, and determine the current real-time operating condition of the microgrid based on the real-time operating status data; Step S1 includes: S11: Obtain the original operating status data of electrical equipment in the microgrid. The original operating status data includes multiple sets of status indicators used to characterize multiple operating condition dimensions. Each set of status indicators comes from multiple distributed electrical devices in the microgrid. S12: For at least one operating condition dimension, determine whether there is a conflict between the corresponding status indicators; S13: If a conflict exists, then according to the preset device permission level, select the status indicator corresponding to the device with the highest permission from the conflicting status indicators, and use it as the determining indicator for the operating condition dimension. S14: Determine the real-time operating condition based on the determination indicators of the operating condition dimension; S2: Based on the real-time operating conditions, match the corresponding historical normal operating status data from the historical normal mode library; Step S2 includes: S21: Based on the real-time operating conditions, obtain multiple adjacent historical normal operating status data from the historical normal mode library; S22: Based on the proximity between the real-time operating condition and the historical operating condition corresponding to each of the multiple adjacent historical normal operating condition data, determine the weight of each of the adjacent historical normal operating condition data. S23: Generate a normal operation status data reference based on multiple adjacent historical normal operation status data and their respective weights; S24: Use the normal operating status data as a reference, and use it as historical normal operating status data that is matched with the real-time operating conditions; S3: Compare the real-time operating status data with the historical normal operating status data to determine the degree of difference between the real-time operating status data and the historical normal operating status data; S4: Identify electrical equipment anomalies in the microgrid based on the aforementioned degree of difference; Step S4 includes: S41: Obtain multiple differences generated at consecutive time points to generate a difference time series; S42: Determine whether the difference time series meets the preset continuous exceedance condition and whether it meets the preset growth trend condition; S43: When the difference time series meets the continuous over-limit condition or the growth trend condition, an electrical equipment anomaly in the microgrid is identified.

2. The electrical testing method based on artificial intelligence according to claim 1, characterized in that, Step S13 includes: S131: For conflicting operating condition dimensions, obtain the preset device permission list; S132: Determine the validity of the status indicators of each device in the list according to the priority of the device permission list; S133: Select the device with the highest priority in the device permission list and whose status indicator data is determined to be valid, and use the status indicator corresponding to the device as the determining indicator of the operating condition dimension.

3. The electrical testing method based on artificial intelligence according to claim 1, characterized in that, Step S22 includes: S221: Obtain multiple dimensions constituting the real-time operating condition, and for each dimension, calculate the dimensional deviation between the real-time operating condition and each adjacent historical operating condition. S222: According to the preset dimension calibration rules, the deviations of each dimension are calibrated to generate calibration deviations; S223: Combine the calibration deviations for the same nearby historical operating conditions to generate a characterization of the real-time... The combined distance of proximity between the operating condition and the aforementioned adjacent historical operating conditions; S224: Based on the comprehensive distance, determine the weight corresponding to the nearby historical normal operation status data.

4. The electrical testing method based on artificial intelligence according to claim 3, characterized in that, Step S222 includes: S2221: For each dimension constituting the real-time operating condition, determine its statistical distribution characteristics in the historical normal operating condition data; S2222: Based on the statistical distribution characteristics determined by the dimension, the dimension deviation is processed to generate the dimensionless calibration deviation.

5. The electrical testing method based on artificial intelligence according to claim 1, characterized in that, Step S23 includes: S231: The multiple data parameters constituting the normal operating state data are divided into a basic parameter group and a derived parameter group according to a preset physical constraint relationship; S232: Weighted aggregation of each parameter in the basic parameter group with the corresponding parameter values ​​in multiple adjacent historical normal operation status data and their weights to generate aggregated basic parameters; S233: Based on the physical constraint relationship and the basic aggregation parameters, the parameter values ​​of the derived parameter group are calculated; S234: Combine the aggregated basic parameters with the parameter values ​​of the derived parameter group to generate the normal operation status data reference.

6. The electrical testing method based on artificial intelligence according to claim 1, characterized in that, Step S3 includes: S31: Based on the real-time operating conditions, analyze the historical normal operating status data in the historical normal mode library that are adjacent to the real-time operating conditions to determine the normal deviation boundary that characterizes the normal fluctuation range under the real-time operating conditions. S32: Calculate the instantaneous deviation between the real-time operating status data and the historical normal operating status data; S33: Compare the instantaneous deviation with the normal deviation boundary, and quantify the portion of the instantaneous deviation that exceeds the normal deviation boundary as the degree of difference.

7. An electrical testing system based on artificial intelligence, characterized in that, The system for implementing the method according to any one of claims 1-6 comprises: Acquisition module: Acquires real-time operating status data of electrical equipment in the microgrid, and determines the current real-time operating condition of the microgrid based on the real-time operating status data; The acquisition module is also used to acquire the original operating status data of electrical equipment in the microgrid. The original operating status data includes multiple sets of status indicators used to characterize multiple operating condition dimensions, wherein each set of status indicators comes from multiple distributed electrical devices in the microgrid. For at least one operating condition dimension, determine whether there is a conflict between the corresponding status indicators; If a conflict exists, the status indicator corresponding to the device with the highest authority is selected from the conflicting status indicators according to the preset device permission level, and used as the determining indicator for the operating condition dimension. The real-time operating condition is determined based on the indicators of the operating condition dimension. Matching module: Based on the real-time operating conditions, it matches the corresponding historical normal operating status data from the historical normal mode library; The matching module is also used to obtain multiple adjacent historical normal operation status data from the historical normal mode library based on the real-time operating conditions. Based on the proximity between the real-time operating condition and the historical operating condition corresponding to each of the multiple adjacent historical normal operating condition data, the weight of each of the adjacent historical normal operating condition data is determined. Based on multiple adjacent historical normal operation status data and their respective weights, a normal operation status data reference is generated; The normal operating status data is used as a reference for historical normal operating status data that is matched with the real-time operating conditions. Comparison module: compares the real-time operating status data with the historical normal operating status data to determine the degree of difference between the real-time operating status data and the historical normal operating status data; Identification module: Based on the degree of difference, identify abnormalities in electrical equipment within the microgrid; The identification module is also used to acquire multiple differences generated at consecutive time points to generate a difference time series; Determine whether the difference time series meets the preset continuous exceedance condition and whether it meets the preset growth trend condition; When the difference time series meets the continuous over-limit condition or the growth trend condition, an electrical equipment anomaly in the microgrid is identified.

Images