Electrical equipment fault monitoring and location system based on data analysis

Abstract

The present invention discloses an electrical equipment fault monitoring and positioning system based on data analysis, which relates to the field of electrical equipment monitoring technology. In order to solve the problems in the prior art that still exist in terms of the comprehensiveness of data processing, the intelligence of fault detection, and the accuracy of fault positioning, the present invention dynamically adjusts the sensor acquisition frequency according to the equipment operating status to ensure the validity and pertinence of the data, uses time domain and frequency domain analysis combined with network model training to improve the accuracy and reliability of fault feature extraction, calculates similarity and partition positioning, quickly and accurately determines the known fault type, and analyzes and predicts unknown fault features at the same time, and updates the determined unknown fault information to the database. Combined with the fault type, equipment topology and structural information, the fault propagation path and impact range are simulated, the fault location range is determined and evaluated and optimized, providing accurate information for fault repair, reducing equipment maintenance costs and downtime.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment monitoring, and in particular to an electrical equipment fault monitoring and positioning system based on data analysis. Background Art

[0002] Traditional electrical equipment fault monitoring and positioning methods mostly rely on manual inspections, which are inefficient and difficult to detect potential faults; for example, Chinese patent publication number CN104833894A discloses a fault monitoring system and fault monitoring method for electrical equipment, including an intelligent temperature sensing device, a remote processor and an alarm terminal; the intelligent temperature sensing device is connected to the remote processor via remote communication; the remote processor is connected to the alarm terminal; the intelligent temperature sensing device is adsorbed and fixed on the connecting part of the electrical equipment, and is used to collect the temperature signal of the connecting part according to a preset first time interval, and send the temperature signal to the remote processor; the remote processor is used to determine whether the temperature signal is greater than a preset first threshold, and if so, send an alarm signal to the alarm terminal; the alarm terminal is used to sound an alarm when an alarm signal is received.

[0003] However, in the existing technology, faults are judged only by monitoring the temperature signals of the connectors. The monitoring parameters are single and cannot fully reflect the operating status of the electrical equipment. There are still deficiencies in the comprehensiveness of data processing, the intelligence of fault detection and the accuracy of fault location. It is impossible to accurately determine the specific location and cause of the fault, and cannot meet the increasingly complex needs of electrical equipment fault monitoring and location. Summary of the Invention

[0004] The purpose of the present invention is to provide an electrical equipment fault monitoring and positioning system based on data analysis, which collects rich data through multiple types of sensors and combines data processing and analysis to achieve fault detection and positioning, ensure the safe and stable operation of electrical equipment, and solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The electrical equipment fault monitoring and location system based on data analysis includes:

[0007] A data acquisition module is configured to collect multi-source data based on various types of sensors, obtain monitoring data of electrical equipment in real time during operation, and adjust the collection frequency of various types of sensors according to the operating status of the equipment;

[0008] a data processing module configured to pre-process the collected monitoring data, obtain operating data of various types of sensors, correct the acquired monitoring data based on the operating data of various types of sensors, and extract fault characteristics from the corrected monitoring data;

[0009] Among them, the acquired monitoring data is corrected, and the evaluation of sensor operation status correction data is also included, and historical operation data is screened. After power comparison and time slice division, the historical data for building the sensor performance evaluation model is determined by waveform feature comparison and screening;

[0010] A fault detection module is configured to match the extracted fault features with the fault database, determine the known fault type corresponding to the fault features based on the matching results, and analyze unknown fault features and update them into the fault database;

[0011] The fault location module is configured to analyze the propagation path and impact range of the fault in the device based on the fault characteristics and the corresponding fault type, and determine the location of the fault in combination with the device structure information.

[0012] Furthermore, the data processing module includes:

[0013] A data preprocessing unit is configured to identify and eliminate abnormal data based on sensor types, obtain corresponding noise characteristics, remove noise, and standardize monitoring data obtained by sensors of different types;

[0014] a data correction unit configured to evaluate the operating status of each type of sensor, perform error correction on the preprocessed monitoring data based on the operating status of the sensor, and obtain corrected monitoring data;

[0015] The data correction unit further includes screening the historical operation data to obtain the historical operation data that can be used to build the sensor performance evaluation model, and extracting features from the historical operation data that can be used to build the sensor performance evaluation model to obtain data features of the historical operation data as input data for building the sensor performance evaluation model;

[0016] The feature extraction unit is configured to calculate the time domain statistical features and waveform features of the corrected monitoring data, construct a time domain feature vector, convert the time domain monitoring data through Fourier transform, calculate the frequency domain features, construct a frequency domain feature vector, input the time domain feature vector and the frequency domain feature vector into the network model for training, and mine the fault features in the data.

[0017] Furthermore, the data correction sheet obtains the data features of the historical operating data as input data for building the sensor performance evaluation model. The specific execution steps are as follows:

[0018] Real-time monitoring of the average power per unit time during the operation of electrical equipment, where the value range of the unit time is 12 hours to 36 hours;

[0019] Compare the average power corresponding to each unit time with a preset power reference value, and extract the power value corresponding to the average power exceeding the preset power reference value as the target reference power value;

[0020] When the target reference power value is one, the time slice is set using a preset basic time length; wherein the value range of the basic time length is 5 minutes to 10 minutes;

[0021] When there are multiple target reference power values, the target reference power standard deviation is obtained according to the average power value of the multiple target reference power values;

[0022] The time length is set by using the target reference power standard deviation combined with the average power value of multiple target reference power values;

[0023] Dividing the time length corresponding to the historical operation data using the time length to obtain multiple time slices;

[0024] Divide the running time of the historical running data according to the time length to obtain multiple time slices;

[0025] Data is filtered based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model;

[0026] Feature extraction is performed on historical operating data that can be used to build a sensor performance evaluation model, and data features of the historical operating data are obtained as input data for building the sensor performance evaluation model.

[0027] Furthermore, the data preprocessing unit specifically includes:

[0028] Determine the data fluctuation range of different categories of sensor data according to the sensor category, extract abnormal data that does not meet the fluctuation range of the sensor data of this category, perform jump anomaly and missing anomaly detection, remove duplicate labeled data, and eliminate abnormal data;

[0029] Based on the sensor category, the noise characteristics of different types of sensor data after removing abnormal data are obtained, and the filtering parameters are dynamically adjusted according to the real-time changes of the monitoring data to remove high-frequency noise and low-frequency interference;

[0030] Establish standardized data format conversion rules to convert the monitoring data collected by various types of sensors with different coding formats and storage structures into a unified format, and perform dimension normalization on various types of monitoring data through pre-set dimension conversion coefficients.

[0031] Furthermore, the data correction unit specifically includes:

[0032] Obtain historical operating data of various types of sensors during the operation of electrical equipment, including raw data collected by sensors under different working conditions and at different times, and corresponding equipment operating status information, where the equipment operating status information corresponds one-to-one with the historical operating data;

[0033] Extract the data features of historical operation data, use the historical operation data features as input and the sensor measurement error as output to build a sensor performance evaluation model;

[0034] Input the real-time collected sensor operation data into the sensor performance evaluation model for evaluation to obtain the current sensor operation status evaluation result;

[0035] Based on the sensor performance evaluation results, analyze the error characteristics of the sensor data, determine the sensor error type, and select the corresponding error correction strategy from the strategy library;

[0036] The collected monitoring data is error corrected according to the error correction strategy. During the correction process, the correction effect is monitored in real time until the error of the monitoring data is controlled within the preset error threshold, and the corrected monitoring data is obtained.

[0037] Furthermore, data screening is performed based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model. The specific execution steps are as follows:

[0038] Extracting the overall waveform of normal sensor operation of each sensor stored in the database;

[0039] Extracting the sensor waveform of each sensor of the historical operation data of each time slice as a sensor sub-waveform;

[0040] Extracting waveform features of the sensor sub-waveform corresponding to each time slice;

[0041] Comparing the waveform characteristics of the sensor sub-waveform corresponding to each time slice with the waveform characteristics of the overall waveform of the sensor's normal operation, determining the waveform position where the characteristic attributes of the waveform characteristics of the overall waveform of the sensor's normal operation are the same as the waveform characteristic attributes of the sensor sub-waveform corresponding to the time slice, and intercepting the waveform to obtain a reference waveform;

[0042] Compare the sensor sub-waveform corresponding to each time slice with its corresponding reference waveform for similarity, and obtain a similarity value;

[0043] Comparing the similarity value with a preset similarity threshold;

[0044] Retrieve the time slice corresponding to the similarity value lower than the preset similarity threshold as the target time slice;

[0045] The historical operation data contained in the target time slice is extracted as the historical operation data capable of constructing a sensor performance evaluation model.

[0046] Furthermore, the fault detection module includes:

[0047] a fault matching unit configured to calculate the similarity between the fault signature and known fault signature vectors stored in the fault database, quantify the degree of matching between the fault signatures, compare the calculated similarity value with a preset similarity threshold interval, and determine the fault type corresponding to the fault signature;

[0048] The fault matching unit also includes building a partition positioning rule based on the main features of the fault characteristics, dividing the fault database into multiple data subsets, extracting key feature parameters of the fault feature vector, and locating the corresponding fault database subset based on the extraction results;

[0049] a fault analysis unit configured to analyze the fault characteristics of the unidentified fault type based on the judgment result of the fault matching unit, mine the associated feature data with the unknown fault characteristics based on the historical monitoring data, predict the possible fault type corresponding to the unknown fault characteristics based on the associated feature data, and evaluate the confidence level of each prediction result;

[0050] The database updating unit is configured to update the determined unknown fault features and their corresponding fault types and prediction credibility into the fault database according to the preset data format and storage rules based on the prediction confidence evaluation result.

[0051] Furthermore, the fault detection module further includes:

[0052] The data correction unit corrects the monitoring data collected by each sensor during the correction process to a data correction amplitude ratio corresponding to a preset error threshold;

[0053] Comparing the data correction amplitude ratio with a preset amplitude ratio threshold;

[0054] Extracting the data correction amplitude ratios corresponding to the amplitude ratios exceeding a preset threshold value to form a data correction amplitude ratio set;

[0055] Comparing the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set with a preset data number threshold;

[0056] When the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set exceeds a preset data number threshold, adjusting the lower limit value of the preset similarity threshold interval using the data correction amplitude ratios included in the data correction amplitude ratio set to obtain an adjusted lower limit value;

[0057] Each time the lower limit of the preset similarity threshold interval is adjusted, the data correction amplitude ratios contained in the data correction amplitude ratio set are cleared, and the data correction amplitude ratios corresponding to the amplitude ratio threshold that exceeds the preset threshold are re-recorded;

[0058] Real-time monitoring of the lower limit of the similarity threshold interval;

[0059] When the lower limit of the similarity threshold interval is lower than the preset lower limit threshold, an abnormal operation alarm of the device sensor is issued.

[0060] Furthermore, the fault location module includes:

[0061] A fault propagation analysis unit is configured to describe the operating status of the electrical equipment based on the fault type corresponding to the fault characteristics and in combination with the topological structure of the electrical equipment, and simulate the propagation path and impact range of the fault in the electrical equipment;

[0062] a fault location determination unit configured to determine the location range of the fault based on the propagation path and impact range simulation results, combined with equipment structure information, working principles, and common failure modes;

[0063] The positioning result evaluation unit is configured to evaluate the location range of the fault, determine the accuracy and reliability of the location range, and optimize the location range according to the evaluation result to determine the location coordinates of the fault.

[0064] Furthermore, the fault location determination unit further includes:

[0065] Based on the simulation results of the fault propagation path and impact range, the regional scope formed by the abnormal change of physical quantities when the fault occurs is obtained, and the boundary of the fault impact is determined;

[0066] According to the boundary shape characteristics of the fault area, the fault area is equivalent to the standard geometric figure that is closest to the boundary shape;

[0067] Obtain the center position of the standard geometric figure, and use the position where the physical quantity changes the most due to the fault as the critical fault reference point, and obtain the straight-line distance between the center position of the standard geometric figure and the critical fault reference point;

[0068] Based on the calculated straight-line distance between the center position of the standard geometric figure and the critical reference point of the fault, a line segment with a length equal to a preset proportional coefficient of the straight-line distance is intercepted in the direction close to the critical reference point of the fault, and the endpoint position of the line segment close to the critical reference point of the fault is taken as the center position of the fault area;

[0069] Taking the installation position of the sensor as the reference point, obtain the effective monitoring range boundary line of the sensor closest to the fault center position, and calculate the shortest straight-line distance between the fault center position and the effective monitoring range boundary line.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] Rich data is collected by multiple types of sensors to make up for the incomplete monitoring of a single sensor; data preprocessing, multiple feature extraction methods and deep learning technology are used to effectively process and analyze complex data, improving the accuracy of fault feature extraction; known faults are identified through similarity calculation and partition positioning, and unknown fault related features are mined and updated to improve system adaptability. The propagation path and range are simulated by combining multiple information, and the fault location is determined through evaluation and optimization, achieving accurate positioning of the fault; the efficiency and accuracy of electrical equipment fault monitoring and positioning are improved, equipment maintenance costs are reduced, downtime is reduced, and stable operation of electrical equipment is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 This is a module diagram of the electrical equipment fault monitoring and positioning system based on data analysis of the present invention. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] In order to solve the technical problems of incomplete monitoring, inaccurate positioning and low efficiency in existing technologies, please refer to Figure 1 , this embodiment provides the following technical solutions:

[0075] The electrical equipment fault monitoring and location system based on data analysis includes:

[0076] The data acquisition module is configured to collect multi-source data based on various types of sensors, including current sensors, voltage sensors, temperature sensors, and vibration sensors. The module adopts a distributed deployment approach and rationally sets sensor locations based on different components and key locations of electrical equipment to ensure comprehensive collection of operating data from all parts of the equipment, obtain real-time monitoring data of electrical equipment during operation, and adjust the collection frequency of various types of sensors based on the equipment's operating status.

[0077] A data processing module is configured to pre-process the collected monitoring data, unify the format and dimensions of the data collected by different sensors, and then use principal component analysis to reduce the dimensionality of the data. At the same time, it obtains the operating data of each category of sensors, corrects the acquired monitoring data based on the operating data of each category of sensors, extracts fault features from the corrected monitoring data, analyzes the features in the time domain, and analyzes the features of the data in the frequency domain through Fourier transform, and uses deep learning technology to automatically learn and extract complex features in the data;

[0078] A fault detection module is configured to match the extracted fault features with the fault database, determine the known fault type corresponding to the fault features based on the matching results, and analyze unknown fault features and update them into the fault database;

[0079] The fault location module is configured to analyze the fault propagation path and impact range in the equipment based on the fault characteristics and corresponding fault type, determine the fault location in combination with the equipment structure information, and use intelligent optimization algorithms to optimize the fault location process to improve positioning accuracy. At the same time, it can generate a fault impact assessment report based on the severity and impact range of the fault.

[0080] In this embodiment, the data acquisition module adjusts the acquisition frequency of each type of sensor, specifically including:

[0081] Acquire feedback data from the data processing module and extract characteristic parameters of the monitoring data in the time and frequency domains in real time, such as current amplitude fluctuations, voltage harmonic content, temperature gradient, and vibration main frequency components. This constructs a device operating status feature vector to comprehensively reflect the device operating status.

[0082] The equipment operation status feature vector is input into the trained equipment operation status assessment model for assessment, and the assessment result is compared with the preset assessment level threshold to determine the current operation status level of the electrical equipment, including normal operation, slight abnormality, and severe abnormality;

[0083] When electrical equipment is operating at normal operating levels, the acquisition frequency of each type of sensor is reduced to reduce data redundancy and storage pressure. At the same time, a lower frequency limit is set to ensure the effective collection of key data.

[0084] When the electrical equipment is at a mild abnormality level, the current acquisition frequency of each type of sensor is maintained, and the abnormal data marking mechanism is activated to mark abnormal features in the subsequent collected data;

[0085] When electrical equipment is at a severe abnormality level, the collection frequency of each category of sensors is increased, and the collection frequency of sensors corresponding to key parts of the electrical equipment is adjusted at a high frequency to obtain more detailed operating data.

[0086] In this embodiment, the data acquisition module implements differentiated sensor acquisition frequency adjustment strategies according to different status levels, reduces the frequency when the equipment is operating normally to reduce data redundancy and storage pressure, effectively improves the efficiency and quality of data acquisition, enhances the timeliness and accuracy of fault warnings, and optimizes resource utilization and system performance.

[0087] In this embodiment, the data processing module includes:

[0088] A data preprocessing unit is configured to identify and eliminate abnormal data based on sensor types, obtain corresponding noise characteristics, remove noise, and standardize monitoring data obtained by sensors of different types;

[0089] a data correction unit configured to evaluate the operating status of each type of sensor, perform error correction on the preprocessed monitoring data based on the operating status of the sensor, and obtain corrected monitoring data;

[0090] The feature extraction unit is configured to calculate the time domain statistical features such as the mean, peak, root mean square value, kurtosis, skewness, etc. of the corrected monitoring data, as well as waveform features such as rise time, fall time, and pulse width, to construct a time domain feature vector, and convert the time domain monitoring data through Fourier transform, calculate the frequency domain features such as spectrum amplitude, frequency distribution, and energy spectrum density, to construct a frequency domain feature vector, input the time domain feature vector and the frequency domain feature vector into the network model for training, and mine the fault features in the data.

[0091] In this embodiment, the data preprocessing unit specifically includes:

[0092] Determine the data fluctuation range of different categories of sensor data according to the sensor category, extract abnormal data that does not meet the fluctuation range of the sensor data of this category, perform jump anomaly and missing anomaly detection, remove duplicate labeled data, and eliminate abnormal data;

[0093] Based on the sensor category, the noise characteristics of different types of sensor data after removing abnormal data are obtained. The filtering parameters are dynamically adjusted according to the real-time changes of the monitoring data to remove high-frequency noise and low-frequency interference and retain the effective characteristics of the data.

[0094] Establish standardized data format conversion rules to convert monitoring data collected by various types of sensors with different coding formats and storage structures into a unified format, and perform dimension normalization on various types of monitoring data through pre-set dimension conversion coefficients to make the data comparable.

[0095] In this embodiment, the data correction unit specifically includes:

[0096] Obtain historical operating data of various types of sensors during the operation of electrical equipment, including raw data collected by sensors under different working conditions and at different times, and corresponding equipment operating status information, where the equipment operating status information corresponds one-to-one with the historical operating data;

[0097] Extract the data features of historical operation data, use the historical operation data features as input and the sensor measurement error as output to build a sensor performance evaluation model;

[0098] Input the real-time collected sensor operation data into the sensor performance evaluation model for evaluation to obtain the current sensor operation status evaluation results, including whether the sensor is in normal working condition, whether there are drift, nonlinear errors and other problems, as well as the current measurement accuracy level;

[0099] Based on the sensor performance evaluation results, the error characteristics of the sensor data are analyzed to determine the sensor error type. If the evaluation results show that the sensor data has a systematic deviation, it is determined to be a drift error; if the data shows a nonlinear change trend within the measurement range, it is determined to be a nonlinear error; if the data has random abnormal fluctuations, it is determined to be a random error, and the corresponding error correction strategy is selected from the strategy library. For example, for drift error, a linear compensation strategy based on the calibration curve is adopted; for nonlinear error, a polynomial fitting or piecewise linearization correction method is adopted; for random error, a filtering algorithm is used for smoothing;

[0100] The collected monitoring data is error corrected according to the error correction strategy. During the correction process, the correction effect is monitored in real time until the error of the monitoring data is controlled within the preset error threshold, and the corrected monitoring data is obtained.

[0101] Specifically, the data correction unit further includes screening the historical operation data to obtain the historical operation data that can be used to build the sensor performance evaluation model, and extracting features from the historical operation data that can be used to build the sensor performance evaluation model to obtain data features of the historical operation data as input data for building the sensor performance evaluation model. The specific execution steps are as follows:

[0102] Real-time monitoring of the average power per unit time during the operation of electrical equipment, where the value range of the unit time is 12 hours to 36 hours;

[0103] Compare the average power corresponding to each unit time with a preset power reference value, and extract the power value corresponding to the average power exceeding the preset power reference value as the target reference power value;

[0104] When the target reference power value is one, the time slice is set using a preset basic time length; wherein the value range of the basic time length is 5 minutes to 10 minutes;

[0105] When there are multiple target reference power values, the target reference power standard deviation is obtained according to the average power value of the multiple target reference power values;

[0106] The time length is set by using the target reference power standard deviation combined with the average power value of multiple target reference power values;

[0107] The time length is obtained by the following formula:

[0108] ;

[0109] Among them, T represents the length of time, T base Indicates the basic time length; P represents the average power of the target reference power value; σ p represents the target reference power standard deviation; T gp Indicates the average time interval between the unit times of multiple target reference power values; specifically, The ratio of power fluctuation to average power is calculated to represent the relative magnitude of power changes. This ratio reflects the relative stability of the target reference power value. A larger ratio indicates more severe power fluctuations; a smaller ratio indicates more stable power. This serves as an important basis for subsequent calculations to reflect the impact of power fluctuations over time. This part reflects the influence of the distribution of the target reference power value in time on the result. gp The larger the value, the longer the unit time interval, the sparser the power data distribution in time, the smaller the exponential function value, and the smaller the adjustment effect on the final time length; on the contrary, T gp The smaller it is, the denser the power data is distributed in time, the larger the exponential function value is, and the greater the adjustment effect on the time length is. First multiply the results of the previous two parts, and then take the square root. The square root operation is to normalize the result of the product of the two parts so that the final result is within a suitable numerical range. At the same time, it also comprehensively considers the joint influence of the power fluctuation amplitude and time distribution on the result. This formula comprehensively reflects the combined effect of the two factors of power fluctuation and time distribution on the time length adjustment. Through this operation, the influence of the two factors is integrated together to provide a basis for the subsequent adjustment of the basic time length. The above formula is based on the basic time length T baseBased on the target reference power value, the result obtained from the above calculation is used to adjust it. When the power fluctuation is large and the time distribution is dense, the value of the adjustment item is large, which will increase the final time length T. Conversely, when the power fluctuation is small and the time distribution is sparse, the value of the adjustment item is small, and the final time length T is relatively small. The final time length T is the result of a reasonable adjustment to the basic time length after comprehensively considering the fluctuation, average level, and time distribution of the target reference power value. It is more consistent with the time characteristics related to power in actual operation and is used to reasonably divide historical operation data into time periods.

[0110] Dividing the time length corresponding to the historical operation data using the time length to obtain multiple time slices;

[0111] Divide the running time of the historical running data according to the time length to obtain multiple time slices;

[0112] Data is filtered based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model;

[0113] Feature extraction is performed on historical operating data that can be used to build a sensor performance evaluation model, and data features of the historical operating data are obtained as input data for building the sensor performance evaluation model.

[0114] The technical solution described above achieves the following: by real-time monitoring of the average power of electrical equipment per unit time (12-36 hours) and comparing it with a preset power reference value, target reference power values exceeding the reference value are extracted. This eliminates abnormally low power levels, retains data that better reflects normal or typical equipment operating conditions, reduces errors caused by power anomalies, and improves the accuracy of the data used for model construction. Time slices are set based on the target reference power value. When there is only one target reference power value, a preset base time length (5-10 minutes) is used; when there are multiple target reference power values, the time length is set based on the target reference power standard deviation and the average power value. Reasonable time slice division enables more scientific segmentation of historical operating data along the temporal dimension, ensuring data similarity and continuity within each time slice, further improving data accuracy and providing a reliable data foundation for subsequent model construction. By filtering historical operating data based on time slices, it is possible to quickly and accurately identify data that meets the requirements for building sensor performance evaluation models, avoiding blind searches through large amounts of useless data. This significantly reduces data processing workload and time costs, and improves the efficiency of data processing and model construction. Feature extraction is performed on the filtered data, and the resulting data features serve as model input data. Due to the precise initial data screening, the extracted features are more representative and targeted, better reflecting information related to sensor performance, accelerating model training, and improving model building efficiency. The filtered and processed data is more representative of the actual operation of electrical equipment related to sensor performance. Input into the model allows the model to be trained and evaluated based on more realistic and valid data, resulting in more reliable sensor performance evaluation results and reducing evaluation errors caused by data bias. The extracted targeted data features enable the model to better learn the relevant laws of sensor performance, making the model's sensor performance evaluation more accurate and stable, enhancing the model's reliability in practical applications, and providing strong support for model-based decision-making.

[0115] Furthermore, the formula proposed in the above technical solution comprehensively considers multiple factors, including the target reference power value's average power, standard deviation, and average value per unit time interval. This comprehensively reflects the characteristics and temporal distribution of power data. Compared to considering only a single factor, it can more accurately determine the appropriate time duration based on actual power conditions, avoiding inaccurate time duration calculations due to ignoring important factors. By introducing a nonlinear adjustment mechanism through an exponential function, it can more flexibly reflect the impact of the time interval average on the result. For different temporal distribution characteristics, nonlinear adjustment can provide an adjustment amplitude that is more consistent with actual physical processes, rather than a simple linear adjustment. This further improves the alignment of the time duration calculation with actual conditions and enhances accuracy. Based on the ratio of the power standard deviation to the average power, it effectively adapts to varying degrees of power fluctuation. When power fluctuations are large, the calculated time duration is adjusted accordingly to better match frequent power changes. When power fluctuations are small, the time duration is also adjusted appropriately, making the time division more compatible with actual power stability. By calculating the average value per unit time interval and adjusting the exponential function, the time duration calculation can adapt to the varying temporal distribution of the target reference power values. Regardless of whether the power data is densely or sparsely distributed, the appropriate time length can be calculated through the formula to ensure that the time division is compatible with the actual power-time characteristics, thereby improving the applicability of the time length calculation method in different operating scenarios.

[0116] Specifically, data screening is performed based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model. The specific execution steps are as follows:

[0117] Extracting the overall waveform of normal sensor operation of each sensor stored in the database;

[0118] Extracting the sensor waveform of each sensor of the historical operation data of each time slice as a sensor sub-waveform;

[0119] Extracting waveform features of the sensor sub-waveform corresponding to each time slice; wherein the waveform features include but are not limited to rising edges, falling edges, etc.;

[0120] Comparing the waveform characteristics of the sensor sub-waveform corresponding to each time slice with the waveform characteristics of the overall waveform of the sensor's normal operation, determining the waveform position where the characteristic attributes of the waveform characteristics of the overall waveform of the sensor's normal operation are the same as the waveform characteristic attributes of the sensor sub-waveform corresponding to the time slice, and intercepting the waveform to obtain a reference waveform;

[0121] Compare the sensor sub-waveform corresponding to each time slice with its corresponding reference waveform for similarity, and obtain a similarity value;

[0122] Comparing the similarity value with a preset similarity threshold;

[0123] Retrieve the time slice corresponding to the similarity value lower than the preset similarity threshold as the target time slice;

[0124] The historical operation data contained in the target time slice is extracted as the historical operation data capable of constructing a sensor performance evaluation model.

[0125] The technical solution achieves the following: By extracting the overall waveform of the sensor during normal operation and the sensor sub-waveforms for each time slice and comparing their waveform features, time slices with characteristic attributes that differ from the normal waveform can be accurately identified. Time slices with similarity values below a preset threshold are selected as target time slices, and their historical operating data is extracted. This ensures that the selected data is specific to abnormal sensor operating conditions or those of research value, eliminating a large amount of normal and redundant data, significantly improving the data validity used for model construction. Waveform features such as rising and falling edges are specifically considered, as these often reflect key changes in sensor operating conditions. By comparing these key features to filter data, the acquired data better captures the core information about sensor performance changes, further improving the data's effectiveness for sensor performance evaluation. The filtered target time slice data, which differs from normal operating conditions, provides a rich set of abnormal or special operating condition samples for the sensor performance evaluation model. Building a model based on this data enables the model to learn more effective information, improving the accuracy and reliability of the sensor performance evaluation and preventing the model from being overwhelmed by a large amount of normal data and failing to accurately capture performance variation patterns. Incorporating a variety of abnormal state data helps the model learn a wider range of sensor performance variation patterns, enhancing the model's generalization capabilities. This enables the model to more accurately evaluate and predict performance when faced with new and similar abnormal operating conditions, improving the model's practicality in real-world applications. By pre-screening valid data, the model avoids incorporating large amounts of normal data into the model building process, reducing unnecessary computation and data processing time. During the model training phase, there's no need to calculate massive amounts of useless data, significantly improving the efficiency of model building and evaluation and shortening the development cycle. When using the constructed model for sensor performance evaluation, since the model is trained on valid data, it can more quickly and accurately identify potential sensor performance issues, improving evaluation efficiency and facilitating timely maintenance measures to ensure the normal operation of sensors and related equipment.

[0126] In this embodiment, the data processing module effectively removes data interference, corrects errors, and deeply extracts fault features from multiple dimensions, providing a solid data foundation for electrical equipment fault monitoring and positioning, ensuring stable equipment operation, and realizing all-round management and in-depth mining of monitoring data, effectively improving data quality and the accuracy of fault feature extraction.

[0127] In this embodiment, the fault detection module includes:

[0128] a fault matching unit configured to calculate the similarity between the fault signature and known fault signature vectors stored in the fault database, quantify the degree of matching between the fault signatures, compare the calculated similarity value with a preset similarity threshold interval, and determine the fault type corresponding to the fault signature;

[0129] The fault matching unit also includes building partitioning and positioning rules based on the main features of the fault characteristics, such as partitioning based on current amplitude fluctuations, partitioning based on voltage harmonic content, etc., dividing the fault database into multiple data subsets, extracting key characteristic parameters of the fault feature vector, such as current amplitude fluctuations, voltage harmonic content, temperature change gradient, vibration main frequency component, etc., and locating the corresponding fault database subset based on the extraction results;

[0130] a fault analysis unit configured to analyze the fault characteristics of the unidentified fault type based on the judgment result of the fault matching unit, mine the associated feature data with the unknown fault characteristics based on the historical monitoring data, predict the possible fault type corresponding to the unknown fault characteristics based on the associated feature data, and evaluate the confidence level of each prediction result;

[0131] The database updating unit is configured to update the determined unknown fault features and their corresponding fault types and prediction credibility into the fault database according to the preset data format and storage rules based on the prediction confidence evaluation result.

[0132] Specifically, the fault detection module also includes:

[0133] The data correction unit corrects the monitoring data collected by each sensor during the correction process to a data correction amplitude ratio corresponding to a preset error threshold;

[0134] Comparing the data correction amplitude ratio with a preset amplitude ratio threshold;

[0135] Extracting the data correction amplitude ratios corresponding to the amplitude ratios exceeding a preset threshold value to form a data correction amplitude ratio set;

[0136] Comparing the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set with a preset data number threshold;

[0137] When the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set exceeds a preset data number threshold, adjusting the lower limit value of the preset similarity threshold interval using the data correction amplitude ratios included in the data correction amplitude ratio set to obtain an adjusted lower limit value;

[0138] The adjusted lower limit value is obtained by the following formula:

[0139] ;

[0140] Where X represents the lower limit after adjustment; X0 represents the lower limit before adjustment; n represents the number of data correction amplitude ratios contained in the data correction amplitude ratio set; σ x Indicates the standard deviation of the data correction amplitude ratio contained in the data correction amplitude ratio set; X i represents the corresponding value of the i-th data correction amplitude ratio in the data correction amplitude ratio set; X p represents the average value of the data correction amplitude ratio contained in the data correction amplitude ratio set; specifically, This represents the average across all data points, reflecting the combined discreteness and relative average characteristics of the entire set of data correction amplitude ratios. This component reflects the overall fluctuation characteristics of the data correction amplitude ratio set and the distribution of the data relative to the average. Larger values indicate more dramatic fluctuations in the data correction amplitude ratios and more significant differences in the data relative to the average, indicating more complex and unstable deviations in the sensor-collected data. The range of the sine function sin() is between [−1,1]. Multiply the value reflecting the data characteristics calculated above by This is then used as the independent variable of the sine function. This utilizes the nonlinear characteristics of the sine function to map the data eigenvalues to a new interval, thereby achieving nonlinear control over the adjustment amplitude of the lower limit. Through the sine function transformation, the eigenvalues of the data correction amplitude ratio set are converted to a value between [−1, 1], which represents the direction and relative amplitude of the lower limit adjustment based on the data characteristics. When this value is close to 1, it means that the lower limit needs to be adjusted significantly based on the data; when it is close to -1, it may indicate the opposite direction or a very small adjustment. The above formula uses the lower limit value X0 before the overall adjustment as the basis, and adjusts it using the sine function value obtained from the above calculation. When the sin() value is large, the 1−sin() value is small, and the adjusted lower limit value X will be significantly lower than X0; when the sin() value is small, the adjustment amplitude is small. X is the lower limit value adjusted based on the characteristics of the data correction amplitude ratio set. It comprehensively considers factors such as the number of data correction amplitude ratios, standard deviation, each data value, and average value, and realizes dynamic adjustment of the lower limit value through nonlinear transformation to adapt to the actual situation of sensor data deviation.

[0141] Each time the lower limit of the preset similarity threshold interval is adjusted, the data correction amplitude ratios contained in the data correction amplitude ratio set are cleared, and the data correction amplitude ratios corresponding to the amplitude ratio threshold that exceeds the preset threshold are re-recorded;

[0142] Real-time monitoring of the lower limit of the similarity threshold interval;

[0143] When the lower limit of the similarity threshold interval is lower than the preset lower limit threshold, an abnormal operation alarm of the device sensor is issued.

[0144] The technical effect of the above technical solution is as follows: During the process of correcting sensor monitoring data, the data correction unit calculates the correction amplitude ratio corresponding to correcting the data to a preset error threshold. This ratio reflects the degree of deviation between the sensor-collected data and the accurate data. This data correction amplitude ratio is compared with a preset amplitude ratio threshold. If it exceeds the threshold, it indicates that the sensor-collected data has a significant deviation. These data are then extracted to form a data correction amplitude ratio set. The number of data in the data correction amplitude ratio set is counted and compared with a preset data number threshold. If the number of data in the set exceeds the threshold, it indicates that the sensor has significant data deviation within a certain time period or range, possibly indicating an operational anomaly. When the number of data exceeds the threshold, the data correction amplitude ratios in the data correction amplitude ratio set are used to adjust the lower limit of the preset similarity threshold interval. Generally, the lower limit is appropriately lowered based on the degree of data deviation, making subsequent judgments more rigorous and enabling more sensitive detection of sensor anomalies. After the adjustment, the data correction amplitude ratio set is cleared, and the data correction amplitude ratios that exceed the preset amplitude ratio threshold are re-recorded to continuously monitor sensor data deviation. The lower limit of the similarity threshold interval is monitored in real time. Once the lower limit falls below the preset lower limit, it indicates a high probability of sensor anomaly. A sensor anomaly alarm is issued, notifying relevant personnel to take steps to troubleshoot and resolve the issue. Furthermore, the data correction amplitude ratio reflects the correction amplitude required to correct sensor monitoring data to the preset error threshold. A larger ratio indicates greater deviation in the sensor data. When this ratio is compared with the preset amplitude ratio threshold, exceeding the threshold indicates that the sensor data deviation is outside the normal range. When the number of data correction amplitude ratios exceeding the threshold (i.e., the number of data in the data correction amplitude ratio set) exceeds the preset data number threshold, it indicates that the sensor has severe or frequent data deviation. Based on this determination, if the sensor data deviation is severe, the data correction amplitude ratios in the data correction amplitude ratio set are used to adjust the preset lower limit of the similarity threshold interval. Generally speaking, the larger the data correction amplitude ratio and the larger the number of data in the set, the more severe the deviation in the sensor data. In this case, the lower limit of the similarity threshold interval should be appropriately lowered. This is because when the sensor deviation is large, the originally high similarity threshold may cause some actual abnormal situations to be mistakenly judged as normal. Lowering the lower limit value can make the judgment standard more stringent, making it easier to detect sensor operation abnormalities, and ensuring that the fault detection system can discover problems in a timely manner.

[0145] By comparing the data correction amplitude ratio with a preset amplitude ratio threshold, data with correction amplitude ratios exceeding the threshold are selected and formed into a set. This set is then compared with a preset data number threshold. If the number of data in the set exceeds the threshold, it indicates significant deviation in the sensor data, and the lower limit of the similarity threshold interval needs to be adjusted. This multi-step judgment mechanism accurately detects sensor operating anomalies, avoiding misjudgments due to small data deviations while also addressing large deviations, thereby improving fault detection accuracy. The lower limit of the similarity threshold interval is dynamically adjusted based on the data correction amplitude ratio set, ensuring that the threshold more closely matches the actual sensor operating conditions. As sensor performance changes, the threshold can be adjusted accordingly, ensuring appropriate fault detection criteria and further improving detection accuracy. Real-time monitoring of the lower limit of the similarity threshold interval triggers an alarm if it falls below the preset lower limit. This allows for rapid detection of sensor anomalies, prompting timely notification of relevant personnel, and minimizing fault duration and losses. If the number of data in the data correction amplitude ratio set exceeds the limit, the threshold is adjusted, the set is cleared, and re-recorded. This allows for rapid response to changes in sensor data deviations, ensuring the fault detection system remains sensitive and enabling timely detection of potential faults. Dynamically adjusting the lower limit of the similarity threshold interval accommodates performance fluctuations caused by factors such as sensor aging and environmental changes. The threshold can be appropriately adjusted under different operating conditions to ensure the proper functioning of the fault detection system and enhance system adaptability. By determining and processing sensor data deviations based on the data correction amplitude ratio, the system can flexibly respond to various levels of data deviation, effectively detecting and processing both slow-moving and sudden deviations, making the system adaptable to different types of sensor anomalies.

[0146] Furthermore, the formula proposed in the above technical solution comprehensively considers multiple factors, including the number of data points, standard deviation, individual data values, and average value, within the data correction amplitude ratio set. This comprehensively reflects the data characteristics and distribution. Compared to considering only a single factor, this formula can more accurately adjust the lower limit based on the actual sensor data deviation, ensuring that the adjusted lower limit more closely matches the sensor's actual operating conditions, thereby improving the accuracy of the fault detection threshold. By introducing a nonlinear adjustment mechanism using a sine function, the lower limit can be more flexibly adjusted based on data characteristics. For varying data fluctuations and distributions, nonlinear adjustment can produce an adjustment amplitude that better reflects the actual physical process, rather than a simple linear adjustment. This further improves the alignment of the lower limit adjustment with the actual conditions and enhances accuracy. Based on the standard deviation of the data correction amplitude ratio set and the distribution of the data values, the formula effectively adapts to varying degrees of data fluctuation. When data fluctuations are large, the formula adjusts the lower limit accordingly through correlation calculations, ensuring more stringent sensor anomaly detection. When data fluctuations are small, the lower limit adjustment amplitude is reduced accordingly, allowing the fault detection threshold to adapt to varying data fluctuations. By comprehensively considering multiple data characteristics, this formula can adapt to sensor data deviations under different actual operating conditions. Whether the deviation is caused by factors such as sensor aging, environmental changes, or sudden abnormal deviations, the formula can reasonably adjust the lower limit value, allowing the fault detection system to better adapt to the actual operating conditions of the sensor, thereby improving the system's adaptability and reliability.

[0147] In this embodiment, by calculating similarities and establishing partitioning and location rules, known fault types can be quickly and accurately identified, the database can be partitioned, and the search scope can be narrowed. The fault analysis unit effectively improves fault detection efficiency and enhances the system's ability to respond to unknown faults, making it possible to promptly discover potential faults and updating the database with identified unknown fault information. This continuously enriches and improves the fault database, enhances the system's self-learning capabilities, and ensures that the system can continue to operate effectively even in the face of new faults.

[0148] In this embodiment, the fault location module includes:

[0149] A fault propagation analysis unit is configured to describe the operating status of the electrical equipment based on the fault type corresponding to the fault characteristics and in combination with the topological structure of the electrical equipment, and simulate the propagation path and impact range of the fault in the electrical equipment;

[0150] a fault location determination unit configured to determine the location range of the fault based on the propagation path and impact range simulation results, combined with equipment structure information, working principles, and common failure modes;

[0151] The positioning result evaluation unit is configured to evaluate the location range of the fault, determine the accuracy and reliability of the location range, and optimize the location range according to the evaluation result to determine the location coordinates of the fault.

[0152] In this embodiment, the fault location determination unit further includes:

[0153] Based on the simulation results of the fault propagation path and impact range, the regional scope formed by the abnormal change of physical quantities when the fault occurs is obtained, and the boundary of the fault impact is determined;

[0154] Based on the boundary shape characteristics of the fault area, the fault area is equated to a standard geometric shape that is closest to the boundary shape, such as a circle, rectangle, or polygon. By calculating the distance distribution from each point in the area to the boundary, the standard geometric shape that best fits the area shape is selected.

[0155] Obtain the center of the standard geometric figure and use the location where the fault causes the greatest change in physical quantities (such as the location with the maximum current mutation amplitude, the location with the highest voltage distortion rate, the peak point of abnormal temperature increase, and the location where the vibration acceleration exceeds the threshold most significantly) as the critical fault reference point. Obtain the straight-line distance between the center of the standard geometric figure and the critical fault reference point.

[0156] Based on the calculated straight-line distance between the center of the standard geometric figure and the critical fault reference point, a line segment with a length equal to a preset proportional coefficient times the straight-line distance is intercepted in the direction close to the critical fault reference point. The endpoint of the line segment close to the critical fault reference point is used as the center of the fault area. The preset proportional range is based on the statistical analysis of a large number of electrical equipment failure case data and can effectively reflect the location of the core fault area.

[0157] Taking the installation position of the sensor as the reference point, obtain the effective monitoring range boundary line of the sensor closest to the fault center position, and calculate the shortest straight-line distance between the fault center position and the effective monitoring range boundary line.

[0158] In this embodiment, by calculating the shortest straight-line distance between the boundary line of the effective monitoring range and the center position of the fault, and comparing the changes in the distance between the center position of the fault and the boundary line of the sensor monitoring range at different times, the rationality of the fault propagation path can be verified. If the fault propagation direction does not match the distance change trend, it may be necessary to re-examine the fault propagation model and improve the analysis of the fault propagation path; further, the rationality of the existing sensor layout can be evaluated, and the impact of the fault on the sensor monitoring data can be analyzed to provide key data support for subsequent fault impact assessment and fault location accuracy optimization. The shortest straight-line distance is the distance between the center position of the fault and the key reference point, which can help determine whether the fault is close to the sensor monitoring edge, thereby correcting and optimizing the positioning results, reducing positioning errors, and making the fault location coordinates more accurate.

[0159] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. The electrical equipment fault monitoring and positioning system based on data analysis is characterized by: include: A data acquisition module is configured to collect multi-source data based on various types of sensors, obtain monitoring data of electrical equipment in real time during operation, and adjust the collection frequency of various types of sensors according to the operating status of the equipment; a data processing module configured to pre-process the collected monitoring data, obtain operating data of various types of sensors, correct the acquired monitoring data based on the operating data of various types of sensors, and extract fault characteristics from the corrected monitoring data; Among them, the acquired monitoring data is corrected, and the evaluation of sensor operation status correction data is also included, and historical operation data is screened. After power comparison and time slice division, the historical data for building the sensor performance evaluation model is determined by waveform feature comparison and screening; A fault detection module is configured to match the extracted fault features with the fault database, determine the known fault type corresponding to the fault features based on the matching results, and analyze unknown fault features and update them into the fault database; A fault location module is configured to analyze the fault propagation path and impact range in the device based on the fault characteristics and corresponding fault type, and determine the fault location in combination with the device structure information; The data processing module includes: A data preprocessing unit is configured to identify and eliminate abnormal data based on sensor types, obtain corresponding noise characteristics, remove noise, and standardize monitoring data obtained by sensors of different types; a data correction unit configured to evaluate the operating status of each type of sensor, perform error correction on the preprocessed monitoring data based on the operating status of the sensor, and obtain corrected monitoring data; The data correction unit further includes screening the historical operation data to obtain the historical operation data that can be used to build the sensor performance evaluation model, and extracting features from the historical operation data that can be used to build the sensor performance evaluation model to obtain data features of the historical operation data as input data for building the sensor performance evaluation model; The feature extraction unit is configured to calculate the time domain statistical features and waveform features of the corrected monitoring data, construct a time domain feature vector, convert the time domain monitoring data through Fourier transform, calculate the frequency domain features, construct a frequency domain feature vector, input the time domain feature vector and the frequency domain feature vector into the network model for training, and mine the fault features in the data.

2. The electrical equipment fault monitoring and positioning system based on data analysis according to claim 1, characterized in that: The data correction unit obtains the data features of historical operating data as input data for building the sensor performance evaluation model. The specific execution steps are as follows: Real-time monitoring of the average power per unit time during the operation of electrical equipment, where the value range of the unit time is 12 hours to 36 hours; Compare the average power corresponding to each unit time with a preset power reference value, and extract the power value corresponding to the average power exceeding the preset power reference value as the target reference power value; When the target reference power value is one, the time slice is set using a preset basic time length; wherein the value range of the basic time length is 5 minutes to 10 minutes; When there are multiple target reference power values, the target reference power standard deviation is obtained according to the average power value of the multiple target reference power values; The time length is set by using the target reference power standard deviation combined with the average power value of multiple target reference power values; Dividing the time length corresponding to the historical operation data using the time length to obtain multiple time slices; Divide the running time of the historical running data according to the time length to obtain multiple time slices; Data is filtered based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model; Feature extraction is performed on historical operating data that can be used to build a sensor performance evaluation model, and data features of the historical operating data are obtained as input data for building the sensor performance evaluation model.

3. The electrical equipment fault monitoring and positioning system based on data analysis according to claim 2, characterized in that: The data preprocessing unit specifically includes: Determine the data fluctuation range of different categories of sensor data according to the sensor category, extract abnormal data that does not meet the fluctuation range of the sensor data of this category, perform jump anomaly and missing anomaly detection, remove duplicate labeled data, and eliminate abnormal data; Based on the sensor category, the noise characteristics of different types of sensor data after removing abnormal data are obtained, and the filtering parameters are dynamically adjusted according to the real-time changes of the monitoring data to remove high-frequency noise and low-frequency interference; Establish standardized data format conversion rules to convert the monitoring data collected by various types of sensors with different coding formats and storage structures into a unified format, and perform dimension normalization on various types of monitoring data through pre-set dimension conversion coefficients.

4. The electrical equipment fault monitoring and locating system based on data analysis according to claim 3, characterized in that: The data correction unit specifically includes: Obtain historical operating data of various types of sensors during the operation of electrical equipment, including raw data collected by sensors under different working conditions and at different times, and corresponding equipment operating status information, where the equipment operating status information corresponds one-to-one with the historical operating data; Extract the data features of historical operation data, use the historical operation data features as input and the sensor measurement error as output to build a sensor performance evaluation model; Input the real-time collected sensor operation data into the sensor performance evaluation model for evaluation to obtain the current sensor operation status evaluation result; Based on the sensor performance evaluation results, analyze the error characteristics of the sensor data, determine the sensor error type, and select the corresponding error correction strategy from the strategy library; The collected monitoring data is error corrected according to the error correction strategy. During the correction process, the correction effect is monitored in real time until the error of the monitoring data is controlled within the preset error threshold, and the corrected monitoring data is obtained.

5. The electrical equipment fault monitoring and locating system based on data analysis according to claim 4, characterized in that: Data is filtered based on the historical operating data corresponding to each time slice to obtain historical operating data that can be used to build a sensor performance evaluation model. The specific execution steps are as follows: Extracting the overall waveform of normal sensor operation of each sensor stored in the database; Extracting the sensor waveform of each sensor of the historical operation data of each time slice as a sensor sub-waveform; Extracting waveform features of the sensor sub-waveform corresponding to each time slice; Comparing the waveform characteristics of the sensor sub-waveform corresponding to each time slice with the waveform characteristics of the overall waveform of the sensor's normal operation, determining the waveform position where the characteristic attributes of the waveform characteristics of the overall waveform of the sensor's normal operation are the same as the waveform characteristic attributes of the sensor sub-waveform corresponding to the time slice, and intercepting the waveform to obtain a reference waveform; Compare the sensor sub-waveform corresponding to each time slice with its corresponding reference waveform for similarity, and obtain a similarity value; Comparing the similarity value with a preset similarity threshold; Retrieve the time slice corresponding to the similarity value lower than the preset similarity threshold as the target time slice; The historical operation data contained in the target time slice is extracted as the historical operation data capable of constructing a sensor performance evaluation model.

6. The electrical equipment fault monitoring and locating system based on data analysis according to claim 5, characterized in that: Fault detection module, including: a fault matching unit configured to calculate the similarity between the fault signature and known fault signature vectors stored in the fault database, quantify the degree of matching between the fault signatures, compare the calculated similarity value with a preset similarity threshold interval, and determine the fault type corresponding to the fault signature; The fault matching unit also includes building a partition positioning rule based on the main features of the fault characteristics, dividing the fault database into multiple data subsets, extracting key feature parameters of the fault feature vector, and locating the corresponding fault database subset based on the extraction results; a fault analysis unit configured to analyze the fault characteristics of the unidentified fault type based on the judgment result of the fault matching unit, mine the associated feature data with the unknown fault characteristics based on the historical monitoring data, predict the possible fault type corresponding to the unknown fault characteristics based on the associated feature data, and evaluate the confidence level of each prediction result; The database updating unit is configured to update the determined unknown fault features and their corresponding fault types and prediction credibility into the fault database according to the preset data format and storage rules based on the prediction confidence evaluation result.

7. The electrical equipment fault monitoring and locating system based on data analysis according to claim 6, characterized in that: The fault detection module also includes: The data correction unit corrects the monitoring data collected by each sensor during the correction process to a data correction amplitude ratio corresponding to a preset error threshold; Comparing the data correction amplitude ratio with a preset amplitude ratio threshold; Extracting the data correction amplitude ratios corresponding to the amplitude ratios exceeding a preset threshold value to form a data correction amplitude ratio set; Comparing the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set with a preset data number threshold; When the number of data of the data correction amplitude ratios included in the data correction amplitude ratio set exceeds a preset data number threshold, adjusting the lower limit value of the preset similarity threshold interval using the data correction amplitude ratios included in the data correction amplitude ratio set to obtain an adjusted lower limit value; Each time the lower limit of the preset similarity threshold interval is adjusted, the data correction amplitude ratios contained in the data correction amplitude ratio set are cleared, and the data correction amplitude ratios corresponding to the amplitude ratio threshold that exceeds the preset threshold are re-recorded; Real-time monitoring of the lower limit of the similarity threshold interval; When the lower limit of the similarity threshold interval is lower than the preset lower limit threshold, an abnormal operation alarm of the device sensor is issued.

8. The electrical equipment fault monitoring and locating system based on data analysis according to claim 7, characterized in that: Fault location module, including: A fault propagation analysis unit is configured to describe the operating status of the electrical equipment based on the fault type corresponding to the fault characteristics and in combination with the topological structure of the electrical equipment, and simulate the propagation path and impact range of the fault in the electrical equipment; a fault location determination unit configured to determine the location and range of the fault based on the propagation path and impact range simulation results, combined with equipment structure information, working principles, and common failure modes, and generate a fault impact assessment report; The positioning result evaluation unit is configured to evaluate the location range of the fault, determine the accuracy and reliability of the location range, and optimize the location range according to the evaluation result to determine the location coordinates of the fault.

9. The electrical equipment fault monitoring and locating system based on data analysis according to claim 8, characterized in that: The fault location determination unit further includes: Based on the simulation results of the fault propagation path and impact range, the regional scope formed by the abnormal change of physical quantities when the fault occurs is obtained, and the boundary of the fault impact is determined; According to the boundary shape characteristics of the fault area, the fault area is equivalent to the standard geometric figure that is closest to the boundary shape; Obtain the center position of the standard geometric figure, and use the position where the fault causes the largest change in physical quantity as the fault key reference point, and obtain the straight-line distance between the center position of the standard geometric figure and the fault key reference point; Based on the calculated straight-line distance between the center position of the standard geometric figure and the critical reference point of the fault, a line segment with a length equal to a preset proportional coefficient of the straight-line distance is intercepted in the direction close to the critical reference point of the fault, and the endpoint position of the line segment close to the critical reference point of the fault is taken as the center position of the fault area; Taking the installation position of the sensor as the reference point, obtain the effective monitoring range boundary line of the sensor closest to the fault center position, and calculate the shortest straight-line distance between the fault center position and the effective monitoring range boundary line.

Images