A method and system for evaluating health status of power distribution equipment
Through sensor network acquisition and deep learning algorithms, multi-dimensional data processing is solved, insufficient data fusion and lack of intelligence in the existing distribution equipment health status assessment methods are achieved, more accurate health status assessment and fault diagnosis are achieved, and power is improved System security and maintenance efficiency.
Patent Information
- Application Number
- CN202410979956.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-07-22
AI Technical Summary
The existing distribution equipment health status evaluation method relies on single parameter or multi-parameter data, but has insufficient data fusion and feature extraction, resulting in low reliability and accuracy of evaluation results, and lack of intelligent means to identify fault types and locations in a timely manner. and the reasons, resulting in lagging equipment maintenance and management, increasing the risk of power system operation.
Multi-dimensional operation data is collected through sensor networks, data preprocessing and multi-modal feature extraction are performed, comprehensive health index is calculated using multi-source data fusion technology, and a deep learning algorithm is used to establish a health status evaluation model to generate intelligent early warning information and fault diagnosis. Provide detailed maintenance suggestions.
Improve the accuracy and reliability of the evaluation results, can promptly detect potential operating risks of the equipment, avoid fault expansion, ensure the safe and stable operation of the power system, and improve maintenance efficiency through intelligent fault diagnosis and maintenance suggestions, extend the service life of the equipment, and reduce maintenance cost.
Smart Images

Figure CN118503636B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power system monitoring and maintenance, and in particular to a method and system for evaluating the health status of power distribution equipment. Background Art
[0002] With the continuous increase in electricity demand and the ever-expanding scale of power grids, the health status assessment of distribution equipment has become increasingly important. Distribution equipment undertakes key transmission and distribution tasks in the power system, and its operating status is directly related to the safety and stability of the power system. Traditional methods for evaluating the health status of distribution equipment mainly rely on regular maintenance and manual experience, which have problems such as low efficiency, poor accuracy, and high maintenance costs. In modern power systems, with the development of sensor technology, the Internet of Things, and big data analysis technology, real-time monitoring and intelligent analysis of distribution equipment operating data have become possible, providing a technical basis for achieving more accurate health status assessment.
[0003] However, there are still many technical problems in the practical application of existing technologies. First, traditional health status assessment methods usually rely on a single parameter (such as temperature or current) for assessment, which cannot fully reflect the actual operating status of the equipment. Secondly, even if some advanced assessment methods use multi-parameter data, there are still deficiencies in data fusion and feature extraction, resulting in low reliability and accuracy of the assessment results. In addition, existing technologies lack intelligent means in abnormal warning and fault diagnosis, and cannot identify the type, location and cause of faults in a timely and accurate manner, resulting in delayed equipment maintenance and management, and increasing the risk of power system operation. Summary of the invention
[0004] Based on the above objectives, the present invention provides a method and system for evaluating the health status of power distribution equipment.
[0005] A method for evaluating the health status of power distribution equipment comprises the following steps:
[0006] S1, data collection: collect multi-dimensional operation data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data, and establish a comprehensive power distribution equipment operation data set;
[0007] S2, data preprocessing: preprocess the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and use statistical methods to eliminate abnormal data to ensure the accuracy and consistency of the data;
[0008] S3, multimodal feature extraction: extract multimodal feature parameters from the preprocessed data, including electrical features, thermodynamic features, mechanical features, and environmental features, to provide comprehensive feature information for subsequent analysis;
[0009] S4, fusion health index calculation: using multi-source data fusion technology, the extracted multi-modal features are weighted and fused to calculate the comprehensive health index of the distribution equipment, which reflects the overall operating status and health level of the equipment;
[0010] S5, health status assessment model construction: Based on the comprehensive health index, using deep learning algorithms and historical operation data, a health status assessment model for distribution equipment is established;
[0011] S6, intelligent early warning and fault diagnosis: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and generate maintenance recommendations.
[0012] Optionally, the sensor network in the S1 data collection specifically includes:
[0013] Temperature sensor: installed on the surface or inside of key components of power distribution equipment (such as transformers and circuit breakers) to monitor the temperature data of power distribution equipment;
[0014] Current sensor: installed on the current channel of the power distribution equipment to measure the current flowing through the power distribution equipment and obtain current data;
[0015] Voltage sensor: installed at the output port of the power distribution equipment to measure the working voltage of the power distribution equipment and obtain voltage data;
[0016] Vibration sensor: installed on the housing or supporting structure of the power distribution equipment to monitor the vibration of the power distribution equipment during operation and obtain vibration data;
[0017] Sound sensor: installed around the power distribution equipment to capture the sound characteristics of the equipment during operation and obtain sound data;
[0018] Environmental sensors: Installed around the equipment to monitor ambient temperature, humidity, and noise data.
[0019] Optionally, the data preprocessing specifically includes:
[0020] S21, data cleaning: processing missing values of collected sensor data;
[0021] S22, data calibration: calibrate the sensor data so that the sensor has consistent measurement results under the same conditions and ensure that the timestamps of all sensor data are consistent;
[0022] S23, abnormal data detection: using statistical methods to detect outliers;
[0023] S24, data standardization: standardize the data of all sensors so that the data of different sensors can be compared on the same scale.
[0024] Optionally, the multimodal feature extraction specifically includes:
[0025] S31, electrical feature extraction: the electrical features include current waveform features and frequency fluctuation features. All data points are extracted from the preprocessed current data, the mean and root mean square value of the current data are calculated, the maximum value in the preprocessed current data is found, the peak factor is calculated, the current waveform features are obtained, time series data is extracted from the preprocessed current data, and fast Fourier transform calculation is performed to obtain frequency fluctuation features.
[0026] S32, thermodynamic characteristics: Thermodynamic characteristics include temperature change characteristics and thermal radiation characteristics, which are obtained by extracting all data points from the preprocessed temperature data, calculating the mean of the temperature data, extracting adjacent data points from the preprocessed temperature data, calculating the temperature change rate, and obtaining the temperature change characteristics, extracting the temperature value from the preprocessed temperature data, calculating the thermal radiation intensity, and obtaining the thermal radiation characteristics;
[0027] S33, mechanical characteristics: mechanical characteristics include vibration spectrum characteristics and mechanical stress characteristics. Time series data is extracted from the preprocessed vibration data, and fast Fourier transform calculation is performed. Spectral signals are extracted from the fast Fourier transform results, and peak values are calculated to obtain vibration spectrum characteristics. Adjacent data points are extracted from the preprocessed vibration data, and stress change rates are calculated to obtain mechanical stress characteristics.
[0028] S34, environmental feature extraction: Environmental features include humidity change features and environmental noise features. All data points are extracted from the preprocessed humidity data, the mean of the humidity data is calculated, adjacent data points are extracted from the preprocessed humidity data, the humidity change rate is calculated, and the humidity change features are obtained. Noise power is extracted from the preprocessed noise data, the noise intensity is calculated, and the environmental noise features are obtained.
[0029] Optionally, the fusion health index calculation specifically includes:
[0030] S41, feature normalization: normalize the features of different modalities;
[0031] S42, feature weighting: different weights are given according to the degree of influence of the features on the health status;
[0032] S43, weighted fusion: the normalized features are weighted and summed according to the weights to obtain a comprehensive health index;
[0033] S44, comprehensive health index interpretation: the calculated comprehensive health index is interpreted to evaluate the health status of the power distribution equipment.
[0034] Optionally, the health status assessment model construction specifically includes:
[0035] S51, historical operation data collection: collect historical operation data of power distribution equipment, and extract characteristic data, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics, and calculate the corresponding comprehensive health index based on the extracted characteristic data;
[0036] S52, feature labeling: labeling the extracted feature data with health status labels, such as normal, abnormal, and faulty;
[0037] Construct feature vectors: Each feature vector includes the feature data extracted in S51 and the comprehensive health index, and each feature vector corresponds to a health status label (normal, abnormal, faulty);
[0038] S53, establish a health status assessment model: construct a health status assessment model through a long short-term memory network algorithm;
[0039] S54, model training: taking the feature vector (including the comprehensive health index) as input and the health status label as output, and inputting it into the model for training;
[0040] S55, Model evaluation: Use accuracy, precision, recall, and F1 score to evaluate the performance of the health status assessment model;
[0041] S56, model deployment: deploy the trained health status assessment model to the power distribution equipment monitoring system, input the real-time operation data of the power distribution equipment, and predict the health status.
[0042] Optionally, generating intelligent warning information in S6 includes:
[0043] S61, real-time data input: input the real-time operation data of the power distribution equipment into the trained health status assessment model to perform health status prediction;
[0044] S62, health status judgment: the model outputs the health status of the current device, including normal, abnormal and faulty. If the health status is predicted to be abnormal or faulty, the early warning mechanism is triggered;
[0045] S63, generating warning information: when an abnormality or a fault state is detected, generating warning information.
[0046] Optionally, the intelligent fault diagnosis in S6 includes:
[0047] S64, extracting characteristic data: extracting relevant characteristics from the real-time data and historical data under the current abnormal or fault state, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics;
[0048] S65, matching historical data: matching the currently extracted features with historical fault data;
[0049] S66, determine the fault type, location and cause: determine the type of the current fault based on the matched historical data, infer the fault location based on the fault location information in the historical data and the current sensor data, and analyze the fault cause based on the fault cause in the historical data and the current environmental characteristics.
[0050] Optionally, generating a maintenance suggestion in S6 includes:
[0051] S67, comprehensive analysis: Combine the fault type, location and cause to conduct a comprehensive analysis and generate maintenance recommendations;
[0052] S68, Maintenance recommendations: Maintenance recommendations include fault type (such as overheating, short circuit), fault location (such as transformer, circuit breaker), fault cause (such as cooling system failure, line aging), emergency treatment measures (such as immediate shutdown inspection, load reduction) and long-term maintenance recommendations (such as replacement of aging parts, improvement of cooling system);
[0053] S69, Recommendation format: Generated maintenance recommendations are presented in a standard format to facilitate maintenance personnel to quickly understand and implement them.
[0054] A power distribution equipment health status assessment system, used to implement the above-mentioned power distribution equipment health status assessment method, includes the following modules:
[0055] Data acquisition module: collects multi-dimensional operating data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data;
[0056] Data preprocessing module: preprocesses the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and uses statistical methods to eliminate abnormal data;
[0057] Feature extraction module: extracts multimodal feature parameters from preprocessed data, including electrical features, thermodynamic features, mechanical features, and environmental features;
[0058] Comprehensive health index calculation module: uses multi-source data fusion technology to perform weighted fusion on the extracted multi-modal features and calculate the comprehensive health index of the distribution equipment;
[0059] Health status assessment model building module: Based on the comprehensive health index, the health status assessment model of the distribution equipment is established using the LSTM (Long Short-Term Memory Network) deep learning algorithm and historical operation data;
[0060] Intelligent early warning and fault diagnosis module: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and provide detailed maintenance suggestions
[0061] Beneficial effects of the present invention:
[0062] The present invention collects multi-dimensional operating data such as voltage, current, temperature, humidity, vibration, sound and environmental data through a sensor network to establish a comprehensive distribution equipment operating data set. Using multi-source data fusion technology, feature data from different sources are weighted and fused to calculate the comprehensive health index, making the evaluation results more comprehensive and accurate. Combined with the long short-term memory network deep learning algorithm, the equipment operation data is dynamically monitored and intelligently analyzed, which can capture subtle changes in the equipment operation status and improve the accuracy and reliability of the evaluation.
[0063] The present invention uses a trained health status assessment model to predict the health status of the equipment operation data input in real time. When the health status of the equipment is abnormal, the system generates intelligent early warning information. Through real-time monitoring and early warning mechanisms, it can timely discover the potential operation risks of the distribution equipment, avoid fault expansion, and ensure the safe and stable operation of the power system.
[0064] In the present invention, when the health status of the equipment is predicted to be abnormal or faulty, the system combines historical data and fault characteristics to perform intelligent fault diagnosis to determine the fault type, location and possible cause. By matching historical fault data, the similarity algorithm is used to find the most similar fault case and infer the detailed information of the current fault. The system generates detailed maintenance suggestions, including fault type, fault location, possible cause, emergency treatment measures and long-term maintenance suggestions. Through intelligent fault diagnosis and detailed maintenance suggestions, maintenance personnel are helped to quickly identify and handle problems, improve equipment maintenance efficiency, extend equipment service life and reduce maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0066] Figure 1 A schematic diagram of the steps of the evaluation method according to an embodiment of the present invention;
[0067] Figure 2 Schematic diagram of the evaluation system flow of an embodiment of the present invention. DETAILED DESCRIPTION
[0068] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0069] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0070] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0071] like Figure 1 As shown, a method for evaluating the health status of a power distribution device comprises the following steps:
[0072] S1, data collection: collect multi-dimensional operation data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data, and establish a comprehensive power distribution equipment operation data set;
[0073] S2, data preprocessing: preprocess the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and use statistical methods to eliminate abnormal data to ensure the accuracy and consistency of the data;
[0074] S3, multimodal feature extraction: extract multimodal feature parameters from the preprocessed data, including electrical features (such as current waveform, frequency fluctuation), thermodynamic features (such as temperature change, thermal radiation), mechanical features (such as vibration spectrum, mechanical stress) and environmental features (such as humidity change, environmental noise), to provide comprehensive feature information for subsequent analysis;
[0075] S4, fusion health index calculation: using multi-source data fusion technology, the extracted multi-modal features are weighted and fused to calculate the comprehensive health index of the distribution equipment, which reflects the overall operating status and health level of the equipment;
[0076] S5, health status assessment model construction: Based on the comprehensive health index, using deep learning algorithms and historical operation data, a health status assessment model for distribution equipment is established;
[0077] S6, intelligent early warning and fault diagnosis: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and generate maintenance recommendations.
[0078] The sensor network in S1 data collection specifically includes:
[0079] Temperature sensor: installed on the surface or inside of key components of power distribution equipment (such as transformers and circuit breakers) to monitor the temperature data of power distribution equipment;
[0080] Current sensor: installed on the current channel of the power distribution equipment to measure the current flowing through the power distribution equipment and obtain current data;
[0081] Voltage sensor: installed at the output port of the power distribution equipment to measure the working voltage of the power distribution equipment and obtain voltage data;
[0082] Vibration sensor: installed on the housing or supporting structure of the power distribution equipment to monitor the vibration of the power distribution equipment during operation and obtain vibration data;
[0083] Sound sensor: installed around the power distribution equipment to capture the sound characteristics of the equipment during operation and obtain sound data;
[0084] Environmental sensors: installed around the equipment to monitor ambient temperature, humidity and noise data;
[0085] Through the use of the above sensors, comprehensive collection of multi-dimensional operating data of power distribution equipment can be achieved, providing reliable data support for equipment health status assessment.
[0086] Data preprocessing specifically includes:
[0087] S21, data cleaning: processing missing values of collected sensor data and deleting duplicate data;
[0088] Use the mean method to fill in missing values, expressed as: ;
[0089] in, is a missing value, is the number of valid data points;
[0090] S22, data calibration: calibrate the sensor data so that the sensor has consistent measurement results under the same conditions and ensure that the timestamps of all sensor data are consistent;
[0091] The sensor calibration uses linear regression, expressed as: ;
[0092] in, is the raw reading of the sensor, is the reading after calibration, and is the regression coefficient, which is determined by fitting known standard data;
[0093] The timestamp calibration uses linear interpolation, which is expressed as: ;
[0094] in, is the interpolation result, is the time point that needs to be interpolated, and are the timestamps of known data points, and is the corresponding known data point;
[0095] S23, abnormal data detection: using statistical methods to detect outliers;
[0096] The statistical method for detecting anomalies uses the Z-score calculation, which is expressed as: ;
[0097] in, is the data point, is the mean of the data, is the standard deviation of the data, is the Z-score, when When the data point is an outlier;
[0098] S24, data standardization: standardize the data of all sensors so that the data of different sensors can be compared on the same scale;
[0099] Z-score normalization formula: ;
[0100] in, is the original data, is the standardized data, is the mean of the data, is the standard deviation of the data;
[0101] Through the above steps, the collected raw data can be preprocessed effectively. Data cleaning ensures the integrity and consistency of the data, data calibration ensures that the readings of different sensors are comparable under the same conditions, abnormal data detection identifies and removes unreasonable data points, and data standardization scales the data to a uniform range. These steps provide a reliable data foundation for subsequent data analysis and health status assessment.
[0102] Multimodal feature extraction specifically includes:
[0103] S31, electrical feature extraction: the electrical features include current waveform features and frequency fluctuation features. All data points are extracted from the preprocessed current data, the mean and root mean square value of the current data are calculated, the maximum value in the preprocessed current data is found, the peak factor is calculated, the current waveform features are obtained, time series data is extracted from the preprocessed current data, and fast Fourier transform calculation is performed to obtain frequency fluctuation features.
[0104] The average value of the current data is calculated as: ,in, is the current data point, is the total number of data points;
[0105] The RMS value of the current data is calculated as: ,in, is the current data point, is the total number of data points;
[0106] The crest factor of the current signal is calculated as: ,in, is the maximum value of the current, RMS is the root mean square value;
[0107] Perform fast Fourier transform on the current signal and calculate it as: ,in, is a time domain signal, is the frequency domain signal, is the fast Fourier transform length;
[0108] S32, thermodynamic characteristics: Thermodynamic characteristics include temperature change characteristics and thermal radiation characteristics, which are obtained by extracting all data points from the preprocessed temperature data, calculating the mean of the temperature data, extracting adjacent data points from the preprocessed temperature data, calculating the temperature change rate, and obtaining the temperature change characteristics, extracting the temperature value from the preprocessed temperature data, calculating the thermal radiation intensity, and obtaining the thermal radiation characteristics;
[0109] The average value of the temperature data is calculated as: ,in, is the temperature data point, is the total number of data points;
[0110] The rate of change of adjacent temperature data points is calculated as: ;in, and are adjacent temperature data points, is the time interval;
[0111] Thermal radiation intensity is calculated as: ,in, is the radiation intensity, is the Stefan-Boltzmann constant, is the temperature;
[0112] S33, mechanical characteristics: mechanical characteristics include vibration spectrum characteristics and mechanical stress characteristics. Time series data is extracted from the preprocessed vibration data, and fast Fourier transform calculation is performed. Spectral signals are extracted from the fast Fourier transform results, and peak values are calculated to obtain vibration spectrum characteristics. Adjacent data points are extracted from the preprocessed vibration data, and stress change rates are calculated to obtain mechanical stress characteristics.
[0113] Perform fast Fourier transform on the vibration signal and calculate it as: ,in, is a time domain signal, is the frequency domain signal, is the fast Fourier transform length;
[0114] The peak value in the vibration signal spectrum, calculated as: ,in, is the spectrum signal, Indicates the maximum value;
[0115] The rate of change of adjacent stress data points is calculated as: ,in, and are adjacent stress data points, is the time interval;
[0116] S34, environmental feature extraction: environmental features include humidity change features and environmental noise features. All data points are extracted from the preprocessed humidity data, the mean of the humidity data is calculated, adjacent data points are extracted from the preprocessed humidity data, the humidity change rate is calculated, and the humidity change feature is obtained. Noise power is extracted from the preprocessed noise data, the noise intensity is calculated, and the environmental noise feature is obtained.
[0117] The mean of the humidity data is calculated as: ,in, is the humidity data point, is the total number of data points;
[0118] The rate of change of adjacent humidity data points is calculated as: ,in, and are adjacent humidity data points, is the time interval;
[0119] Noise intensity, calculated as: ,in, is the noise power, is the reference power;
[0120] Through these steps, multimodal feature parameters can be extracted from the preprocessed data, providing a rich information basis for subsequent data fusion and health status assessment.
[0121] The calculation of the Fusion Health Index specifically includes:
[0122] S41, feature normalization: normalize the features of different modalities to ensure that they are compared and fused at the same scale;
[0123] For each feature Use Min-Max normalization processing, the Min-Max normalization formula is:
[0124] ,in, and Characteristics The minimum and maximum values of
[0125] S42, feature weighting: different weights are given according to the degree of influence of the features on the health status;
[0126] Determine weights: The weight of each feature can be determined by expert experience or data-driven methods (such as principal component analysis PCA) ;
[0127] Features features, and their weights are ,satisfy: ;
[0128] S43, weighted fusion: the normalized features are weighted and summed according to the weights to obtain a comprehensive health index;
[0129] The comprehensive health index is calculated as: ;
[0130] in, is a comprehensive health index. It is After normalization, It is The weight of each feature;
[0131] S44, comprehensive health index interpretation: interpreting the calculated comprehensive health index to evaluate the health status of the power distribution equipment;
[0132] Defining the range of health status: Based on the comprehensive health index The value of defines different health status ranges;
[0133] For example:
[0134] : Good health;
[0135] : Need to pay attention;
[0136] : Poor health status, need maintenance;
[0137] Through the above steps, multi-source data fusion technology can be used to perform weighted fusion on the extracted multi-modal features, calculate the comprehensive health index of the distribution equipment, comprehensively reflect the operating status of the equipment, and provide strong support for equipment maintenance and fault warning.
[0138] The construction of health status assessment model specifically includes:
[0139] S51, historical operation data collection: collect historical operation data of power distribution equipment, and extract characteristic data, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics, and calculate the corresponding comprehensive health index based on the extracted characteristic data;
[0140] S52, feature labeling: labeling the extracted feature data with health status labels, such as normal, abnormal, and faulty;
[0141] Construct feature vectors: Each feature vector includes the feature data extracted in S51 and the comprehensive health index, and each feature vector corresponds to a health status label (normal, abnormal, faulty);
[0142] S53, establish a health status assessment model: construct a health status assessment model through a long short-term memory network algorithm. The specific steps are as follows:
[0143] (1) Define the model architecture:
[0144] Input layer: time series input ;
[0145] LSTM layer: multiple LSTM units extract time series features;
[0146] Fully connected layer: maps the output of the LSTM layer to the health state;
[0147] Output layer: health status classification (normal, abnormal, fault);
[0148] (2) Build an LSTM model;
[0149] Input dimension: the number of features;
[0150] LSTM layer: contains a certain number of LSTM units;
[0151] Fully connected layer: contains a certain number of neurons for health status classification;
[0152] Activation function: ReLU for hidden layers, Softmax for output layers;
[0153] S54, model training: taking the feature vector (including the comprehensive health index) as input and the health status label as output, and inputting it into the model for training;
[0154] S55, Model evaluation: Use accuracy, precision, recall, and F1 score to evaluate the performance of the health status assessment model;
[0155] S56, model deployment: deploy the trained health status assessment model to the power distribution equipment monitoring system, input the real-time operation data of the power distribution equipment, and predict the health status.
[0156] The intelligent warning information generated in S6 includes:
[0157] S61, real-time data input: input the real-time operation data of the power distribution equipment into the trained health status assessment model to perform health status prediction;
[0158] S62, health status judgment: the model outputs the health status of the current device, including normal, abnormal and faulty. If the health status is predicted to be abnormal or faulty, the early warning mechanism is triggered;
[0159] S63, generate warning information: when an abnormal or faulty state is detected, generate warning information, which should include the following contents:
[0160] Device ID;
[0161] Monitoring time;
[0162] Health status (abnormal or faulty);
[0163] Alert level (e.g. low, medium, high);
[0164] Brief description (such as abnormal current fluctuation, excessive temperature, etc.).
[0165] Intelligent fault diagnosis in S6 includes:
[0166] S64, extracting characteristic data: extracting relevant characteristics from the real-time data and historical data under the current abnormal or fault state, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics;
[0167] S65, matching historical data: matching the currently extracted features with the historical fault data, and using a similarity algorithm (such as cosine similarity, Euclidean distance) to find the most similar historical fault case;
[0168] S66, determine the fault type, location and cause: determine the type of the current fault based on the matched historical data. For example, combined with the fault type label of the historical data, it can be determined that the current fault may be overheating, short circuit, mechanical failure, etc. According to the fault location information in the historical data and the current sensor data, infer the fault location. For example, by analyzing the vibration data and current fluctuation data, it can be determined that the fault occurred in the transformer, circuit breaker or other components. Combined with the fault causes in the historical data and the current environmental characteristics, analyze the fault causes. For example, the temperature increase may be due to the failure of the cooling system, and the current fluctuation may be due to line aging or loose connection.
[0169] Maintenance recommendations generated in S6 include:
[0170] S67, comprehensive analysis: Combine the fault type, location and cause to conduct a comprehensive analysis and generate maintenance recommendations;
[0171] S68, Maintenance recommendations: Maintenance recommendations include fault type (such as overheating, short circuit), fault location (such as transformer, circuit breaker), fault cause (such as cooling system failure, line aging), emergency treatment measures (such as immediate shutdown inspection, load reduction) and long-term maintenance recommendations (such as replacement of aging parts, improvement of cooling system);
[0172] S69, Suggestion format: The generated maintenance suggestions are presented in a standard format to facilitate maintenance personnel to quickly understand and implement them. The standard format template is as follows:
[0173] Device ID: XYZ123;
[0174] Monitoring time: year-month-day 00:00;
[0175] Health status: Fault;
[0176] Warning level: High;
[0177] Fault type: Overheating;
[0178] Fault location: Transformer;
[0179] Possible reasons: Cooling system failure;
[0180] Emergency measures: Please stop the machine immediately and check the cooling system;
[0181] Long-term maintenance advice: Consider replacing cooling system components and check coolant levels regularly;
[0182] Through the above steps, when the health status of the power distribution equipment is detected to be abnormal, the system can generate intelligent early warning information, and combine historical data and fault characteristics to perform intelligent fault diagnosis, determine the fault type, location and cause, and provide detailed maintenance suggestions. This process not only helps maintenance personnel quickly identify and handle problems, but also provides strong support for the long-term stable operation of the equipment.
[0183] like Figure 2 As shown, a power distribution equipment health status assessment system is used to implement the above-mentioned power distribution equipment health status assessment method, including the following modules:
[0184] Data acquisition module: collects multi-dimensional operating data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data;
[0185] Data preprocessing module: preprocesses the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and uses statistical methods to eliminate abnormal data;
[0186] Feature extraction module: extracts multimodal feature parameters from preprocessed data, including electrical features, thermodynamic features, mechanical features, and environmental features;
[0187] Comprehensive health index calculation module: uses multi-source data fusion technology to perform weighted fusion on the extracted multi-modal features and calculate the comprehensive health index of the distribution equipment;
[0188] Health status assessment model building module: Based on the comprehensive health index, the health status assessment model of the distribution equipment is established using the LSTM (Long Short-Term Memory Network) deep learning algorithm and historical operation data;
[0189] Intelligent early warning and fault diagnosis module: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and provide detailed maintenance recommendations.
[0190] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0191] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for evaluating the health status of power distribution equipment, characterized in that: The following steps are involved: S1, data collection: collect multi-dimensional operating data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data; S2, data preprocessing: preprocess the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and use statistical methods to eliminate abnormal data; S3, multimodal feature extraction: extracting multimodal feature parameters from the preprocessed data, including electrical features, thermodynamic features, mechanical features, and environmental features; The multimodal feature extraction specifically includes: S31, electrical feature extraction: the electrical features include current waveform features and frequency fluctuation features. All data points are extracted from the preprocessed current data, the mean and root mean square value of the current data are calculated, the maximum value in the preprocessed current data is found, the peak factor is calculated, the current waveform features are obtained, time series data is extracted from the preprocessed current data, and fast Fourier transform calculation is performed to obtain frequency fluctuation features. S32, thermodynamic characteristics: Thermodynamic characteristics include temperature change characteristics and thermal radiation characteristics, which are obtained by extracting all data points from the preprocessed temperature data, calculating the mean of the temperature data, extracting adjacent data points from the preprocessed temperature data, calculating the temperature change rate, and obtaining the temperature change characteristics, extracting the temperature value from the preprocessed temperature data, calculating the thermal radiation intensity, and obtaining the thermal radiation characteristics; S33, mechanical characteristics: mechanical characteristics include vibration spectrum characteristics and mechanical stress characteristics. Time series data is extracted from the preprocessed vibration data, and fast Fourier transform calculation is performed. Spectral signals are extracted from the fast Fourier transform results, and peak values are calculated to obtain vibration spectrum characteristics. Adjacent data points are extracted from the preprocessed vibration data, and stress change rates are calculated to obtain mechanical stress characteristics. S34, environmental feature extraction: environmental features include humidity change features and environmental noise features. All data points are extracted from the preprocessed humidity data, the mean of the humidity data is calculated, adjacent data points are extracted from the preprocessed humidity data, the humidity change rate is calculated, and the humidity change feature is obtained. Noise power is extracted from the preprocessed noise data, the noise intensity is calculated, and the environmental noise feature is obtained. S4, fusion health index calculation: using multi-source data fusion technology, the extracted multi-modal features are weighted and fused to calculate the comprehensive health index of the power distribution equipment. The fusion health index calculation includes: S41, feature normalization: normalize the features of different modalities; S42, feature weighting: different weights are given according to the degree of influence of the features on the health status; S43, weighted fusion: the normalized features are weighted and summed according to the weights to obtain a comprehensive health index; S44, comprehensive health index interpretation: interpreting the calculated comprehensive health index to evaluate the health status of the power distribution equipment; S5, health status assessment model construction: Based on the comprehensive health index, using deep learning algorithms and historical operation data, a health status assessment model for distribution equipment is established. The health status assessment model construction includes: S51, historical operation data collection: collect historical operation data of power distribution equipment, and extract characteristic data, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics, and calculate the corresponding comprehensive health index based on the extracted characteristic data; S52, feature labeling: labeling the extracted feature data with a health status label; Construct feature vectors: Each feature vector includes the feature data extracted in S51 and the comprehensive health index, and each feature vector corresponds to a health status label; S53, establish a health status assessment model: construct a health status assessment model through a long short-term memory network algorithm; S54, model training: taking the feature vector as input and the health status label as output, and inputting them into the model for training; S55, Model evaluation: Use accuracy, precision, recall, and F1 score to evaluate the performance of the health status assessment model; S56, model deployment: deploy the trained health status assessment model to the power distribution equipment monitoring system, input the real-time operation data of the power distribution equipment, and predict the health status; S6, intelligent early warning and fault diagnosis: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and generate maintenance recommendations.
2. A method for evaluating the health status of power distribution equipment according to claim 1, characterized in that: The sensor network in the S1 data collection specifically includes: Temperature sensor: used to monitor the temperature data of power distribution equipment; Current sensor: used to measure the current flowing through the power distribution equipment and obtain current data; Voltage sensor: used to measure the working voltage of power distribution equipment and obtain voltage data; Vibration sensor: used to monitor the vibration of power distribution equipment during operation and obtain vibration data; Sound sensor: captures the sound characteristics of the device when it is running and obtains sound data; Environmental sensors: used to monitor ambient temperature, humidity and noise data.
3. A method for evaluating the health status of power distribution equipment according to claim 2, characterized in that: The data preprocessing specifically includes: S21, data cleaning: processing missing values of collected sensor data; S22, data calibration: calibrate the sensor data so that the sensor has consistent measurement results under the same conditions and ensure that the timestamps of all sensor data are consistent; S23, abnormal data detection: using statistical methods to detect outliers; S24, data standardization: standardize the data of all sensors so that the data of different sensors can be compared on the same scale.
4. A method for evaluating the health status of power distribution equipment according to claim 1, characterized in that: The generating of intelligent early warning information in S6 includes: S61, real-time data input: input the real-time operation data of the power distribution equipment into the trained health status assessment model to perform health status prediction; S62, health status judgment: the model outputs the health status of the current device, including normal, abnormal and faulty. If the health status is predicted to be abnormal or faulty, the early warning mechanism is triggered; S63, generating warning information: when an abnormality or a fault state is detected, generating warning information.
5. A method for evaluating the health status of power distribution equipment according to claim 4, characterized in that: The intelligent fault diagnosis in S6 includes: S64, extracting characteristic data: extracting relevant characteristics from the real-time data and historical data under the current abnormal or fault state, including electrical characteristics, thermodynamic characteristics, mechanical characteristics and environmental characteristics; S65, matching historical data: matching the currently extracted features with historical fault data; S66, determine the fault type, location and cause: determine the type of the current fault based on the matched historical data, infer the fault location based on the fault location information in the historical data and the current sensor data, and analyze the fault cause based on the fault cause in the historical data and the current environmental characteristics.
6. A method for evaluating the health status of power distribution equipment according to claim 5, characterized in that: Generating maintenance suggestions in S6 includes: S67, comprehensive analysis: Combine the fault type, location and cause to conduct a comprehensive analysis and generate maintenance recommendations; S68, Maintenance suggestion content: Maintenance suggestions include fault type, fault location, fault cause, emergency treatment measures and long-term maintenance suggestions; S69, Recommendation format: Generated maintenance recommendations are presented in a standard format to facilitate maintenance personnel to quickly understand and implement them.
7. A power distribution equipment health status assessment system, used to implement a power distribution equipment health status assessment method as described in any one of claims 1 to 6, characterized in that: Includes the following modules: Data acquisition module: collects multi-dimensional operating data of power distribution equipment through sensor networks, including voltage, current, temperature, humidity, vibration, sound and environmental data; Data preprocessing module: preprocesses the collected multi-dimensional operation data, including data cleaning, data calibration and abnormal data detection, and uses statistical methods to eliminate abnormal data; Feature extraction module: extracts multimodal feature parameters from preprocessed data, including electrical features, thermodynamic features, mechanical features, and environmental features; Comprehensive health index calculation module: uses multi-source data fusion technology to perform weighted fusion on the extracted multi-modal features and calculate the comprehensive health index of the distribution equipment; Health status assessment model building module: Based on the comprehensive health index, the health status assessment model of the distribution equipment is established using the LSTM (Long Short-Term Memory Network) deep learning algorithm and historical operation data; Intelligent early warning and fault diagnosis module: When abnormal equipment health status is detected, intelligent early warning information is generated, and intelligent fault diagnosis is performed based on historical data and fault characteristics to determine the fault type, location and cause, and provide detailed maintenance recommendations.
Citation Information
Cited By
Multi-modal fusion and correlation degree driven prevention decision-making method and system for power equipment
CN121211347A
Preventive decision method and system based on power equipment multi-modal fusion and correlation degree driving
CN121211347B
