An intelligent electrical control system

Abstract

The application discloses an intelligent electrical control system, comprising an electrical data acquisition module, an electrical hierarchical correlation set construction module, an electrical operation domain division module, an electrical anomaly determination module and an electrical control module. The application belongs to the field of intelligent control, and specifically refers to an intelligent electrical control system. The scheme defines strong and weak correlation electrical sample sets, calculates sample adaptive correlation thresholds, and avoids misjudgment. Based on domain density division, the electrical data local working condition collaborative correlation is strengthened. The electrical tolerance coefficient and normal domain accommodation coefficient are introduced to calculate the abnormal matching degree, so that the edge normal sample is avoided to be misjudged as abnormal. Through comprehensive calculation of local normal correlation and global operation difference of the abnormal candidate domain electrical sample, misjudgment caused by single dimension index determination is avoided. The normal domain electrical sample is combined with the local correlation strength and the operation stability to select the core reference sample, so that the core normal electrical sample screening deviation is avoided, and the electrical control effect is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control, specifically to an intelligent electrical control system. Background Technology

[0002] Electrical control systems are automated systems that collect operating parameters of electrical equipment through various sensors, control devices, and actuators. They then monitor, logically judge, and automatically adjust the equipment's start-up, shutdown, operating status, and parameter output according to preset rules to ensure the safe, stable, and efficient operation of the electrical system. However, general electrical control systems suffer from problems such as wasted high-quality normal electrical operating data or failure to identify anomalies, and coarse processing of peripheral electrical operating data, leading to poor electrical control reliability. Furthermore, general electrical control systems suffer from high false positive and false negative rates in anomaly identification, and poor alignment between monitoring results and actual industrial operating needs, resulting in poor electrical control performance. Summary of the Invention

[0003] To address the aforementioned issues and overcome the shortcomings of existing technologies, this invention provides an intelligent electrical control system. Addressing the problems of wasted high-quality normal electrical operation data or failed anomaly identification in general electrical control systems, and the coarse processing of edge electrical operation data, leading to poor electrical control reliability, this solution defines strong and weakly correlated electrical sample sets and calculates an adaptive correlation threshold for the samples to avoid mistakenly classifying normal electrical samples in the weakly correlated electrical sample set as anomalies. It also divides the normal domain and anomaly candidate domain based on the normal domain density to strengthen the local condition-based collaborative correlation between electrical operation data. Furthermore, it introduces an electrical tolerance coefficient and a normal domain tolerance coefficient to calculate the anomaly matching degree, avoiding misclassification of normal electrical samples as anomalies due to a single electrical condition. This approach addresses the issue of slightly deviating air quality indicators from the baseline, which could lead to the misclassification of normal samples as abnormal. This improves the reliability of electrical control. To address the common problems in general electrical control systems, such as high false positive and false negative rates in anomaly identification and poor alignment between monitoring results and actual industrial operational needs, resulting in poor electrical control performance, this solution comprehensively calculates the local normal correlation and global operational differences of electrical samples in the anomaly candidate domain. This avoids misjudgments caused by single-dimensional indicator judgments and retains valid electrical operation data with slightly fluctuating parameters but normal operating status. Furthermore, for electrical samples in the normal domain, core baseline samples are selected based on local correlation strength and operational stability, avoiding bias in the selection of core normal electrical samples and thus improving electrical control effectiveness.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides an intelligent electrical control system, including an electrical data acquisition module, an electrical hierarchical association set construction module, an electrical operation domain division module, an electrical anomaly determination module, and an electrical control module;

[0005] The electrical data acquisition module acquires electrical operation data to obtain an electrical sample set;

[0006] The electrical hierarchical association set construction module defines strongly associated electrical sample sets and weakly associated electrical sample sets based on feature similarity, calculates the adaptive association threshold of samples, and obtains the local association strength.

[0007] The electrical operation domain partitioning module calculates the normal domain density based on the local correlation strength and partitions the normal domain and the abnormal candidate domain.

[0008] The electrical anomaly determination module filters core electrical samples in the normal domain and performs matching verification on electrical samples in the candidate abnormal domain, thereby determining the abnormal electrical samples.

[0009] The electrical control module implements electrical control based on the final determined set of abnormal electrical samples.

[0010] Furthermore, the electrical data acquisition module acquires electrical operating data, including core electrical parameters, equipment status parameters, and operating condition related parameters; the acquired data is preprocessed to obtain an electrical sample set.

[0011] Furthermore, the electrical hierarchical association set construction module specifically includes:

[0012] The hierarchical association electrical sample set is defined as follows: For electrical sample i, the association degree with all other electrical samples is calculated using cosine similarity. The top k electrical samples with the highest association degree are selected. If there is an electrical sample j, and the top k association degrees of electrical sample j also include electrical sample i, then electrical sample j is a strongly associated electrical sample of electrical sample i, and the strongly associated electrical sample set of electrical sample i is obtained. If electrical sample j is among the top k association degrees of electrical sample i, and the top 2k associated electrical samples of electrical sample j include i, then electrical sample i is a weakly associated electrical sample set of electrical sample j.

[0013] Based on the weighted summation of the correlation degrees of strong and weakly correlated electrical sample sets, an adaptive correlation threshold for the samples is obtained, and the nearest neighbor coefficient is introduced to weight the weakly correlated samples.

[0014] The number of electrical samples with a correlation degree not lower than the adaptive correlation threshold is counted, and the local operational correlation strength of each electrical sample is quantified.

[0015] Furthermore, the electrical operating domain partitioning module specifically includes:

[0016] Normal domain density calculation; quantify the proportion of strongly correlated electrical samples in electrical sample i, where the local correlation strength is not lower than the local correlation strength of electrical sample i, and determine whether electrical sample i is in the core normal region to obtain the abnormal candidate domain division;

[0017] The abnormal matching degree is calculated for abnormal candidate domain samples from four dimensions: local normal correlation, global operational difference, electrical tolerance, and normal domain tolerance.

[0018] Furthermore, the electrical anomaly determination module specifically includes:

[0019] Screening of core electrical samples in the normal domain; calculating the normal coreness of electrical samples in the normal domain, and combining local correlation strength and stability.

[0020] Verify electrical samples in the candidate domain for anomalies; set an anomaly determination threshold; if the anomaly matching degree of an electrical sample in the candidate domain is higher than the anomaly determination threshold, it is determined to be an abnormal electrical sample; otherwise, it is determined to be a marginal normal electrical sample and classified into the normal domain.

[0021] Iterative optimization; recalculate local correlation strength, divide normal domain into abnormal candidate domain, and verify abnormal electrical samples until the convergence condition is met or the maximum number of iterations is reached; convergence condition: in two adjacent iterations, the change ratio of abnormal electrical samples is lower than the convergence threshold; output the final set of abnormal electrical samples.

[0022] Furthermore, the electrical control module extracts the core abnormal parameters of the abnormal samples based on the final set of real abnormal samples, and formulates a hierarchical adaptive control strategy in combination with the operating characteristics and control requirements of the electrical equipment, thereby realizing electrical control.

[0023] The beneficial effects achieved by adopting the above solution are as follows:

[0024] (1) To address the problem that general electrical control systems suffer from wasted high-quality normal electrical operation data benchmarks or failure in anomaly identification, and coarse processing of edge electrical operation data, which leads to poor electrical control reliability, this solution defines strong and weak correlation electrical sample sets and calculates the adaptive correlation threshold of the samples to avoid misjudging normal electrical samples in the weak correlation electrical sample set as anomalies; it divides the normal domain and anomaly candidate domain based on the normal domain density to strengthen the local operating condition collaborative correlation between electrical operation data; it introduces the electrical tolerance coefficient and the normal domain tolerance coefficient to calculate the anomaly matching degree to avoid misjudging edge normal samples as anomalies due to a slight deviation of a single electrical index from the benchmark value; thereby improving the reliability of electrical control.

[0025] (2) In view of the problems of high false positive and false negative rates in general electrical control systems, poor consistency between monitoring results and actual industrial operation needs, and thus poor electrical control effect, this solution comprehensively calculates the local normal correlation and global operation difference of electrical samples in the candidate abnormal domain to avoid misjudgment caused by single-dimensional index judgment and retains effective electrical operation data with slightly fluctuating parameters but normal operating status; and selects core benchmark samples for electrical samples in the normal domain by combining local correlation strength and operation stability to avoid the screening bias of core normal electrical samples, thereby improving the electrical control effect. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating an intelligent electrical control system provided by the present invention.

[0027] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0029] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0030] Example 1, see Figure 1 The present invention provides an intelligent electrical control system, comprising an electrical data acquisition module, an electrical hierarchical association set construction module, an electrical operation domain division module, an electrical anomaly determination module, and an electrical control module;

[0031] The electrical data acquisition module acquires electrical operation data to obtain an electrical sample set; and sends the data to the electrical hierarchical association set construction module.

[0032] The electrical hierarchical association set construction module defines strongly associated electrical sample sets and weakly associated electrical sample sets based on feature similarity, calculates the adaptive association threshold of samples to obtain the local association strength, and sends the data to the electrical operating domain partitioning module.

[0033] The electrical operation domain segmentation module calculates the normal domain density based on the local correlation strength, divides the normal domain and the abnormal candidate domain, and sends the data to the electrical anomaly determination module.

[0034] The electrical anomaly determination module filters core electrical samples in the normal domain, performs matching and verification on electrical samples in the candidate abnormal domain, and thus determines the abnormal electrical samples; and sends the data to the electrical control module.

[0035] The electrical control module implements electrical control based on the final determined set of abnormal electrical samples.

[0036] Example 2, see Figure 1 This embodiment is based on the above embodiment. The electrical data acquisition module acquires electrical operation data; it collects core operating parameters under electrical control scenarios, including core electrical parameters, equipment status parameters, and operating condition related parameters; the core electrical parameters include voltage, current, power, frequency, and temperature rise; the equipment status parameters include insulation resistance, grounding resistance, switch opening and closing status, and contactor operation count; the operating condition related parameters include load rate, grid harmonic content, voltage fluctuation, and ambient temperature and humidity; the collected data is preprocessed, including outlier removal (3σ criterion) and data standardization (Z-score standardization), to obtain an electrical sample set.

[0037] Example 3, see Figure 1 This embodiment is based on the above embodiment. The electrical hierarchical association set construction module utilizes the local strong correlation characteristics of electrical data (highly similar parameters of samples under the same operating conditions / equipment groups). It calculates the correlation degree between samples through cosine similarity and then divides the sample sets into strong and weak correlation sets through bidirectional nearest neighbor verification. Strongly correlated samples represent stable and normally operating samples under the same operating conditions, while weakly correlated samples represent marginal normal samples with small fluctuations in operating conditions. The hierarchical design accurately characterizes the local correlation features of electrical samples. Specifically, it includes:

[0038] The hierarchical association set of electrical samples is defined as follows: For electrical sample i, the cosine similarity is used to calculate the association degree with all other electrical samples. The top k (values ​​5-15) electrical samples with the highest association degree are selected. If there is an electrical sample j, and the top k association degrees of electrical sample j also include electrical sample i, then electrical sample j is a strongly associated electrical sample of electrical sample i. This yields the strongly associated electrical sample set of electrical sample i. Let represent electrical samples operating stably under the same conditions; if electrical sample j is among the k most correlated electrical samples of electrical sample i, and the first 2k correlated electrical samples of electrical sample j include i, then electrical sample i is a set of weakly correlated electrical samples of electrical sample j. This indicates an edge-normal electrical sample representing a slight fluctuation in operating conditions;

[0039] Based on the weighted summation of the correlation degrees of strongly and weakly correlated electrical sample sets, an adaptive correlation threshold for the samples is obtained. This represents the minimum correlation standard for an electrical sample to be in normal operating condition. Strongly correlated samples have a weight of 1 (core normal samples), while weakly correlated samples are weighted by a nearest neighbor coefficient (marginal normal samples). This adapts to the local operating characteristics of different electrical samples and is expressed as: ; ;in, It is similarity; and These are the standardized data for electrical operation data i and electrical operation data j, respectively. and ... It is the nearest neighbor coefficient (values ​​range from 0.3 to 0.7).

[0040] The number of electrical samples with a correlation degree not lower than the adaptive correlation threshold is counted, and the local operational correlation strength of each electrical sample is quantified as follows: ;in, It is the local correlation strength of electrical operation data i; It is an indicator function; n is the total number of electrical samples, and t is the electrical sample index; It is standardized data of electrical operation data t; it reflects the clustering degree of electrical samples in local operating conditions. Electrical samples that are stable and operating normally have high correlation strength, while abnormal samples that are faulty or drifting have significantly lower correlation strength.

[0041] Example 4, see Figure 1 This embodiment is based on the above embodiment. The electrical operation domain division module quantifies the coreness of electrical samples in a locally strongly correlated set by measuring the density of the normal domain, thereby dividing the core normal domain (electrical samples operating stably under the same conditions) and the abnormal candidate domain (including edge samples with large fluctuations in operating conditions and real abnormal samples with equipment failures). The density of the normal domain is calculated based on the local correlation strength of the strongly correlated sample set, ensuring the locality and reliability of the division basis, and conforming to the local stability and global diversity of the operating characteristics of the electrical system. Specifically, it includes:

[0042] Normal domain density calculation; quantifying the proportion of strongly correlated electrical samples in electrical sample i, where the local correlation strength is not lower than the local correlation strength of electrical sample i, to determine whether electrical sample i is in the core normal region. The normal domain density formula is expressed as: ; The candidate domain partitioning rule is: if If the electrical operation data is within the normal range (core normal data, not yet judged as abnormal); otherwise, the electrical operation data belongs to the abnormal candidate range (further verification is needed to determine if it is abnormal); among which... This is the normal domain density of electrical operation data i; It is a variable used to determine the strength of local associations; It is the local correlation strength of electrical operation data j; This is the threshold for segmentation, ranging from 0.6 to 0.8;

[0043] The anomaly matching degree calculation, for anomaly candidate domain samples, calculates the anomaly matching degree from four dimensions: local normal correlation, global operational difference, electrical withstand capability, and normal domain tolerance capability. This accurately distinguishes between normal samples at the electrical edge (small load changes, minor grid disturbances) and true anomaly samples (equipment failures, parameter drift), and calculates the anomaly matching degree between electrical sample i and the normal domain. (The higher the value, the greater the deviation from the normal cluster, and the higher the probability of anomaly), expressed as: ;in, It is the number of electrical samples that fall into the normal domain in the set of strongly correlated electrical samples of electrical sample i, reflecting the local normal correlation. It is the Euclidean distance between electrical sample i and the center of the normal domain; It is a smooth term (1e) -6 ~1e -4 ); It is the electrical withstand factor, which reflects the ability of electrical equipment to withstand fluctuations in parameters. The closer the value is to 0, the higher the probability of fluctuations exceeding the equipment's tolerance range and the greater the likelihood of an anomaly. ; It is the minimum value of the p-th index in the normal domain; It is the standardized eigenvalue of the p-th index of electrical sample i; These are the indicator weights, ranging from 0.1 to 0.5, and summed to 1. This is the normal domain tolerance coefficient, reflecting the electrical normal domain's ability to tolerate parameter fluctuations. The closer the value is to 0, the more stable the normal domain is, and the higher the probability that the candidate domain sample is an anomaly; if ,but ; It is the standard deviation; It is the standardized value of the p-th index in the normal domain; It is the standardized value of the p-th index of all electrical samples, if ,but .

[0044] By performing the above operations, this solution addresses the problems of wasted high-quality normal electrical operating data benchmarks or failure in anomaly identification in general electrical control systems, as well as the coarse processing of edge electrical operating data, leading to poor electrical control reliability. This solution defines strong and weakly correlated electrical sample sets and calculates adaptive correlation thresholds for the samples to avoid misclassifying normal electrical samples in weakly correlated sets as anomalies. It also divides the normal domain and anomaly candidate domain based on the density of the normal domain, strengthening the local condition-based collaborative correlation between electrical operating data. Furthermore, it introduces electrical tolerance coefficients and normal domain tolerance coefficients to calculate the anomaly matching degree, preventing edge normal samples from being misclassified as anomalies due to slight deviations of a single electrical indicator from the benchmark value. This ultimately improves the reliability of electrical control.

[0045] Example 5, see Figure 1 This embodiment is based on the above embodiment. The electrical anomaly determination module initially identifies (domain partitioning results) anomaly electrical samples that may contain false positives (small parameter shifts caused by temporary power grid disturbances or instantaneous load fluctuations). Therefore, it further filters through core verification and iterative convergence of the normal domain. Core normal electrical samples are selected from the normal domain as a benchmark, and then the electrical samples of the anomaly candidate domain are verified according to the anomaly matching degree. Finally, iterative optimization is used to ensure the stability of the anomaly identification results. Specifically, it includes:

[0046] Screening of core electrical samples in the normal domain; calculation of the normal coreness of electrical samples in the normal domain. Considering both the strength and stability of local correlations, a higher value indicates greater suitability as a benchmark for normal data, expressed as: ;in, The minimum Euclidean distance between electrical sample i and normal domain electrical samples with higher local correlation strength (reflecting electrical operation stability; a lower value indicates smaller parameter fluctuations and more stable operation) is expressed as: Select s (5%~10% of the total number of samples) electrical samples with the highest normality core degree as the normal domain core electrical samples, and use them as the benchmark for anomaly verification.

[0047] Verification of electrical samples in the candidate domain of anomalies; set an anomaly judgment threshold, with a value of 0.3~0.5; if the anomaly matching degree of the electrical sample in the candidate domain of anomalies is higher than the anomaly judgment threshold, it is judged as an abnormal electrical sample; otherwise, it is judged as a marginal normal electrical sample and classified into the normal domain.

[0048] Iterative optimization; recalculate local correlation strength, divide normal domain into abnormal candidate domain, and verify abnormal electrical samples until the convergence condition is met or the maximum number of iterations (5~15) is reached; convergence condition: in two adjacent iterations, the change ratio of abnormal electrical samples is lower than the convergence threshold (value is 0.01~0.05); output the final set of abnormal electrical samples.

[0049] Example 6, see Figure 1 This embodiment is based on the above embodiment. The electrical control module extracts the core abnormal parameters of the abnormal samples based on the final real abnormal sample set. Combining the operating characteristics and control requirements of the electrical equipment, it formulates a hierarchical adaptive control strategy to achieve electrical control. Mild abnormalities (parameters slightly exceeding the threshold) are handled with soft control (load adjustment, parameter fine-tuning), while severe abnormalities (equipment failure, short circuit, overload) are handled with hard control (emergency shutdown, circuit disconnection). This achieves precise and adaptive control of the electrical system, as shown below: Where U represents electrical control operation; It is a mild abnormality soft control strategy, which adopts load regulation and electrical parameter fine-tuning (3%~5%). It is a moderate anomaly early warning and control strategy, which adopts fault early warning and local loop current limiting; It is a hard control strategy for severe abnormalities, which employs emergency equipment shutdown and fault circuit disconnection; These are the thresholds for classifying the degree of abnormality, with values ​​ranging from 0.5 to 0.7 and from 0.7 to 0.9, respectively. .

[0050] By performing the above operations, this solution addresses the problems of high false positive and false negative rates in anomaly identification and poor alignment between monitoring results and actual industrial operation needs in general electrical control systems, leading to poor electrical control performance. This solution comprehensively calculates the local normal correlation and global operational differences of electrical samples in the anomaly candidate domain, avoiding misjudgments caused by single-dimensional indicator judgments and retaining valid electrical operation data with slightly fluctuating parameters but normal operating status. Furthermore, for electrical samples in the normal domain, core benchmark samples are selected based on local correlation strength and operational stability, avoiding bias in the selection of core normal electrical samples, thereby improving electrical control performance.

[0051] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0052] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An intelligent electrical control system, characterized in that: The system includes an electrical data acquisition module, an electrical hierarchical association set construction module, an electrical operation domain division module, an electrical anomaly determination module, and an electrical control module; The electrical data acquisition module acquires electrical operation data to obtain an electrical sample set; The electrical hierarchical association set construction module defines strongly associated electrical sample sets and weakly associated electrical sample sets based on feature similarity, calculates the adaptive association threshold of samples, and obtains the local association strength. The electrical operation domain partitioning module calculates the normal domain density based on the local correlation strength and partitions the normal domain and the abnormal candidate domain. The electrical anomaly determination module filters core electrical samples in the normal domain and performs matching verification on electrical samples in the candidate abnormal domain, thereby determining the abnormal electrical samples. The electrical control module implements electrical control based on the final determined set of abnormal electrical samples; The electrical hierarchical association set construction module utilizes the local strong correlation characteristics of electrical data, calculates the correlation degree between samples through cosine similarity, and then divides the sample sets into strong and weak correlations through bidirectional nearest neighbor verification; specifically including: The hierarchical association set of electrical samples is defined as follows: For electrical sample i, the cosine similarity is used to calculate the association degree with all other electrical samples. The top k electrical samples with the highest association degrees are selected. If there exists an electrical sample j, and the top k association degrees of electrical sample j also include electrical sample i, then electrical sample j is a strongly associated electrical sample of electrical sample i. This yields the strongly associated electrical sample set of electrical sample i. Let represent electrical samples operating stably under the same conditions; if electrical sample j is among the k most correlated electrical samples of electrical sample i, and the first 2k correlated electrical samples of electrical sample j include i, then electrical sample i is a set of weakly correlated electrical samples of electrical sample j. This indicates an edge-normal electrical sample representing operating condition fluctuations; Based on the weighted summation of the correlation degrees of strongly and weakly correlated electrical sample sets, an adaptive correlation threshold for the samples is obtained. This represents the minimum correlation standard for an electrical sample to be in normal operating condition. Strongly correlated samples have a weight of 1, while weakly correlated samples are weighted by a nearest neighbor coefficient to adapt to the local operating characteristics of different electrical samples. It is represented as: ; ;in, It is similarity; and These are the standardized data for electrical operation data i and electrical operation data j, respectively. and ... It is the nearest neighbor coefficient; The number of electrical samples with a correlation degree not lower than the adaptive correlation threshold is counted, and the local operational correlation strength of each electrical sample is quantified as follows: ;in, It is the local correlation strength of electrical operation data i; It is an indicator function; n is the total number of electrical samples, and t is the electrical sample index; It is standardized data of electrical operation data t; The electrical operation domain segmentation module quantifies the coreity of electrical samples within locally strongly correlated sets using normal domain density, thereby dividing the domain into core normal domains and anomalous candidate domains. The normal domain density is calculated based on the local correlation strength of the strongly correlated sample set, ensuring the locality and reliability of the segmentation criteria. Specifically, this includes: Normal domain density calculation; quantifying the proportion of strongly correlated electrical samples in electrical sample i, where the local correlation strength is not lower than the local correlation strength of electrical sample i, to determine whether electrical sample i is in the core normal region. The normal domain density formula is expressed as: ; The candidate domain partitioning rule is: if If the electrical operation data is within the normal range, then it belongs to the normal range; otherwise, it belongs to the abnormal candidate range. This is the normal domain density of electrical operation data i; It is a variable used to determine the strength of local associations; It is the local correlation strength of electrical operation data j; This is the dividing threshold, ranging from 0.6 to 0.8; S i It is a set of strongly correlated electrical samples of electrical sample i; It is the local correlation strength of electrical operation data i; The anomaly matching degree calculation, for anomaly candidate domain samples, is performed from four dimensions: local normal correlation, global operational difference, electrical tolerance, and normal domain tolerance. It distinguishes between normal samples at the electrical edge and true anomaly samples, and calculates the anomaly matching degree between electrical sample i and the normal domain. , is represented as: ;in, It is the number of electrical samples that are classified into the normal domain in the set of strongly correlated electrical samples of electrical sample i. It is the Euclidean distance between electrical sample i and the center of the normal domain; It is a smoothing term; It is the electrical withstand factor. ; It is the minimum value of the p-th index in the normal domain; It is the standardized eigenvalue of the p-th index of electrical sample i; These are indicator weights; It is the normal domain capacity factor; if ,but ; It is the standard deviation; It is the standardized value of the p-th index in the normal domain; It is the standardized value of the p-th index of all electrical samples, if ,but ; The electrical anomaly determination module selects core normal electrical samples from the normal domain as a benchmark, then verifies the electrical samples in the anomaly candidate domain according to the anomaly matching degree, and finally ensures the stability of the anomaly identification results through iterative optimization; specifically including: Screening of core electrical samples in the normal domain; calculation of the normal coreness of electrical samples in the normal domain. , is represented as: ;in, The minimum Euclidean distance between electrical sample i and normal domain electrical samples with higher local correlation strength is expressed as: ; Select the s electrical samples with the highest normal core degree as the normal domain core electrical samples, and use them as the benchmark for anomaly verification; Verify electrical samples in the candidate domain for anomalies; set an anomaly determination threshold; if the anomaly matching degree of an electrical sample in the candidate domain is higher than the anomaly determination threshold, it is determined to be an abnormal electrical sample; otherwise, it is determined to be a marginal normal electrical sample and classified into the normal domain. Iterative optimization; recalculate local correlation strength, divide normal domain into abnormal candidate domain, and verify abnormal electrical samples until the convergence condition is met or the maximum number of iterations is reached; convergence condition: in two adjacent iterations, the change ratio of abnormal electrical samples is lower than the convergence threshold; output the final set of abnormal electrical samples.

2. The intelligent electrical control system according to claim 1, characterized in that: The electrical data acquisition module acquires electrical operation data, including core electrical parameters, equipment status parameters, and operating condition-related parameters; it preprocesses the acquired data to obtain an electrical sample set.

3. The intelligent electrical control system according to claim 2, characterized in that: The electrical control module extracts the core abnormal parameters of the abnormal samples based on the final set of real abnormal samples, and formulates a hierarchical adaptive control strategy in combination with the operating characteristics and control requirements of the electrical equipment, thereby realizing electrical control.

Images