A power fault early warning method and system based on AI big model
Through the power failure warning method based on AI large model, the early warning model is constructed and evaluated using historical data, and data is collected and processed in real time, the problem of low power failure warning efficiency in the existing technology is solved, and more efficient fault warning and positioning is achieved. Improves the stability of the power system.
Patent Information
- Application Number
- CN202410392929.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-02
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-04-02
AI Technical Summary
The existing power fault warning system is inefficient, the algorithm code program is large and the structure is complex, resulting in low fault warning efficiency.
The power failure warning method based on AI large model is adopted, and the historical power system data is extracted and modeled, and the data is collected and preprocessed in real time, combined with the early warning model is used to perform fault warning, and the fault location is located and visualized.
It improves the efficiency of power failure warning, achieves more accurate fault warning capability assessment and fault degree assessment, and improves the stable operation of the power system.
Smart Images

Figure CN118300082B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power failure early warning, and in particular to a power failure early warning method and system based on an AI large model. Background Art
[0002] With the continuous advancement of computer hardware, especially the widespread use of graphics processing units (GPUs), computing power has been greatly improved, making it possible to handle the training and reasoning of large-scale models. Secondly, with the continuous accumulation of data and the continuous improvement of algorithms, researchers have found that using larger models can achieve better performance and effects, which has promoted the development of large models. Artificial Intelligence (AI) large models are models with huge parameter scales. These models learn and understand huge amounts of data through machine learning and deep learning techniques. The development of AI large models has promoted the rapid development of the field of artificial intelligence and has shown amazing application potential in various fields.
[0003] The existing power fault early warning system is a system used to monitor, detect and warn of power system faults. It uses sensors, monitoring equipment, data acquisition, processing systems and early warning mechanisms to achieve real-time monitoring and analysis of the power system status, so as to promptly detect abnormal conditions that may lead to faults and issue early warnings so that appropriate measures can be taken to avoid or mitigate the impact of possible faults.
[0004] For example, the electric power equipment fault early warning method disclosed in the application with publication number: CN117214626A includes obtaining tension values at multiple predetermined time points within a predetermined time period collected by a tension detection belt that is closely attached to the central axis of the surface of the electric power equipment; performing multi-scale feature extraction on the tension values at the multiple predetermined time points to obtain a first-scale tension value time series feature vector and a second-scale tension value time series feature vector; and determining whether to generate an early warning prompt based on the first-scale tension value time series feature vector and the second-scale tension value time series feature vector.
[0005] For example, the remote power line fault early warning monitoring system and method announced with announcement number: CN116660688B includes a data processing device and N traveling wave monitoring devices, wherein the traveling wave monitoring device is used to perform real-time monitoring of the target power line to collect the fault traveling wave signal of the target power line when a fault occurs, and send the fault traveling wave signal to the data processing device through a remote communication module; the data processing device is used to receive the fault traveling wave signal sent by the traveling wave monitoring device, determine multiple position solution groups according to the fault occurrence time of the N fault traveling wave signals, determine the fault occurrence position of the target power line according to the two fault occurrence times in the position solution group to obtain multiple fault occurrence positions, determine the ratio of the two voltage information in each position solution group as a correction amount, and use the correction amount to correct the fault occurrence position to obtain multiple corrected fault occurrence positions; determine the real-time position of the fault occurrence of the target power line according to the multiple corrected fault occurrence positions.
[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:
[0007] In the prior art, power equipment fault warning mainly involves setting upper and lower limits at corresponding measuring points to resolve equipment alarm situations. The technical means of setting upper and lower limits at corresponding measuring points to resolve equipment alarm situations is often implemented through algorithms. However, the algorithm code program volume used in the existing power equipment fault warning is large, and the application structure is relatively complex, resulting in low efficiency of power fault warning. Summary of the invention
[0008] The embodiment of the present application solves the problem of low efficiency of power fault warning in the prior art by providing a power fault warning method and system based on an AI big model, thereby improving the efficiency of power fault warning.
[0009] The embodiment of the present application provides a power fault warning method based on an AI large model, comprising the following steps: S1, extracting features from the historical power system data collected from the preset power sensors of the power system to be warned to obtain the corresponding historical power system feature data, and constructing the power fault warning model to be evaluated based on the extracted historical power system feature data; S2, performing model evaluation on the constructed power fault warning model to be evaluated to obtain a first evaluation result, and comparing the first evaluation result with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power fault warning model, otherwise the model is reconstructed, the first evaluation result is used to evaluate the model accuracy and warning capability of the power fault warning model to be evaluated, and the first power fault warning model is used to perform power fault warning on the power system to be warned; S3, collecting the initial power system data of the power system to be warned within a preset time interval in real time, and preprocessing the initial power system data to obtain power system data, the preprocessing is used to remove data missing and data abnormalities of the initial power system data; S4, performing power fault warning on the acquired power system data in combination with the first power fault warning model to obtain a warning result, and locating the fault position of the power system to be warned according to the warning result and visualizing the positioning result.
[0010] Furthermore, the specific construction process of the power fault warning model to be evaluated is as follows: Step one, construct a feature data set, which is a collection of historical power system feature data, and divide the feature data set into a training feature data set and a verification feature data set according to a preset ratio; Step two, obtain a training machine model, which is a converged model obtained by training a preset machine learning model using the training feature data set, and the preset machine learning model is a model selected according to the characteristics of the historical power system feature data; Step three, obtain a verification machine model, which is a model obtained by model verification of the training machine model using the verification feature data set.
[0011] Furthermore, a specific evaluation method for the model accuracy of the power fault warning model to be evaluated is as follows: obtaining performance evaluation parameters of the power fault warning model to be evaluated, the performance evaluation parameters including accuracy, precision and recall rate; obtaining the accuracy evaluation index of the power fault warning model to be evaluated based on the obtained performance evaluation parameters and combined with the influence of corresponding data on the accuracy evaluation index, the accuracy evaluation index is used to describe the model accuracy of the power fault warning model to be evaluated.
[0012] Furthermore, the specific evaluation method of the early warning capability is as follows: obtaining the essential data and data distribution score of the power fault early warning model to be evaluated, the essential data including data quality data, model complexity data and training data volume, and the data distribution score being the score used by preset professionals to evaluate the early warning capability of the power fault early warning model to be evaluated; obtaining the early warning capability evaluation index of the power fault early warning model to be evaluated based on the acquired essential data and data distribution score, and the early warning capability evaluation index is used to describe the early warning capability of the power fault early warning model to be evaluated.
[0013] Furthermore, the specific process of the power fault warning is as follows: the obtained power system data of the power system to be warned is compared and analyzed with the reference power system data and combined with the warning situation score to obtain the power fault warning index, the power fault warning index is used to describe the fault degree of the power system to be warned, and the warning situation score is the score evaluated by preset professionals on the warning distribution of the power system to be warned; the obtained power fault warning index is compared with the preset threshold to obtain the warning result, if the corresponding preset threshold is met, it indicates that the corresponding power system to be warned has a power fault, otherwise it indicates that the corresponding power system to be warned does not have a power fault.
[0014] Furthermore, a specific method for obtaining the power fault warning index of the power system to be warned is as follows: obtain power system data and a warning situation score of the power system to be warned, the power system data including power system current, power system voltage, power system frequency, power system impedance and power system load; obtain the power fault warning index of the power system to be warned based on the obtained power system data and warning situation score and in combination with the warning weight, the warning weight including the power system weight and the warning situation weight, the power system weight indicating the degree of influence of the power system data on the power fault warning index, the warning situation weight indicating the degree of influence of the warning situation score on the power fault warning index, and the power system weight including a first power system weight, a second power system weight, a third power system weight, a fourth power system weight and a fifth power system weight.
[0015] Furthermore, the specific steps for locating the fault position are as follows: according to the early warning results and in combination with the abnormal detection method, abnormal data points are detected on the power system to be warned to obtain abnormal data points, and the abnormal detection method is used to detect abnormal data points of the power system to be warned; the abnormal data points obtained by analyzing the power fault early warning model are whether they are fault locations. If they are fault locations, a real-time monitoring interface is created, otherwise monitoring is continued, and the real-time monitoring interface is used to display the power system data of the power system to be warned; a fault location heat map is drawn based on the power system data of the power system to be warned monitored in real time, and the fault location heat map is used to describe the positioning results of the fault location of the power system to be warned.
[0016] Furthermore, the early warning capability evaluation index is calculated using the following formula: Where i represents the number of the power failure warning model to be evaluated, i = 1, 2, ..., I, I is the total number of power failure warning models to be evaluated, k represents the number of the preset professional, k = 1, 2, ..., K, K is the total number of preset professionals, YJ i is the early warning capability evaluation index of the i-th power fault early warning model to be evaluated, L i is the data quality data of the i-th power fault warning model to be evaluated, L0 is the model reference data quality data, Z i is the model complexity data of the ith power fault warning model to be evaluated, Z0 is the reference model complexity data, Si is the amount of training data of the ith power fault warning model to be evaluated, S0 is the amount of model reference training data, is the data distribution score of the i-th power fault warning model to be evaluated obtained by the k-th preset professional, b1 is the correction factor of the sum of absolute deviations of essential data, and b2 is the correction factor of the average data distribution score; the sum of absolute deviations of essential data is the sum of absolute deviations of data quality data, model complexity data and training data volume of the power fault warning model to be evaluated.
[0017] Furthermore, the power failure warning index is calculated using the following formula: In the formula, e represents a natural constant, y represents the number of the power system to be warned, y = 1, 2, ..., Y, Y is the total number of power systems to be warned, k represents the number of the preset professional personnel, k = 1, 2, ..., K, K is the total number of preset professional personnel, GZ y is the power failure warning index of the yth power system to be warned, is the warning score of the yth power system to be warned obtained by the kth preset professional, I y is the power system current of the yth power system to be warned, I0 is the reference power system current, U yis the power system voltage of the yth power system to be warned, U0 is the reference power system voltage, V y is the power system frequency of the yth power system to be warned, V0 is the reference power system frequency, W y is the power system impedance of the yth power system to be warned, W0 is the reference power system impedance, F y is the power system load of the yth power system to be warned, F0 is the reference power system load, c1 is the warning situation weight, c2 is the first power system weight, c3 is the second power system weight, c4 is the third power system weight, c5 is the fourth power system weight, and c6 is the fifth power system weight.
[0018] The embodiment of the present application provides an AI large model-based power fault warning system, comprising: a model construction module, a model evaluation module, a data acquisition module and a fault warning module; wherein the model construction module is used to extract features from the historical power system data collected on the preset power sensor of the power system to be warned to obtain corresponding historical power system feature data, and to construct a power fault warning model to be evaluated based on the extracted historical power system feature data; the model evaluation module is used to perform model evaluation on the constructed power fault warning model to be evaluated to obtain a first evaluation result, and compare the first evaluation result with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power fault warning model. Otherwise, the model is reconstructed, and the first evaluation result is used to evaluate the model accuracy and warning capability of the power fault warning model to be evaluated, and the first power fault warning model is used to provide power fault warning for the power system to be warned; the data acquisition module is used to collect the initial power system data of the power system to be warned within a preset time interval in real time, and preprocess the initial power system data to obtain the power system data, and the preprocessing is used to remove data missing and data anomalies in the initial power system data; the fault warning module is used to perform power fault warning on the acquired power system data in combination with the first power fault warning model to obtain a warning result, and locate the fault position of the power system to be warned according to the warning result and visualize the positioning result.
[0019] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:
[0020] 1. By extracting features from the collected historical power system data of the power system to be warned, and constructing a power fault warning model to be evaluated based on the extracted historical power system feature data, and then performing model evaluation on the constructed power fault warning model to be evaluated to obtain a first power fault warning model, then collecting power system data of the power system to be warned within a preset time interval in real time, and performing power fault warning on the power system data to obtain a warning result, and finally locating the fault position of the power system to be warned based on the warning result and visualizing the positioning result, thereby achieving stable operation of the power system, and then achieving improved power fault warning efficiency, effectively solving the problem of low power fault warning efficiency in the prior art.
[0021] 2. By obtaining the essential data and data distribution scores of the power fault warning model to be evaluated, and then calculating the warning capability evaluation index of the power fault warning model to be evaluated based on the acquired essential data and data distribution scores, and finally evaluating the warning capability of the power fault warning model to be evaluated through the calculated warning capability evaluation index, the digitization of the warning capability of the power fault warning model to be evaluated is realized, and then a more accurate evaluation of the warning capability of the power fault warning model to be evaluated is realized.
[0022] 3. By obtaining the power system data and warning scores of the power system to be warned, and then calculating the power fault warning index of the power system to be warned based on the obtained power system data and warning scores and combining the warning weight, finally evaluating the fault degree of the power system to be warned by the calculated power fault warning index, the fault degree of the power system to be warned is quantified, thereby achieving a more accurate assessment of the fault degree of the power system to be warned. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 A flow chart of a power failure early warning method based on an AI large model provided in an embodiment of the present application;
[0024] Figure 2 A flowchart for constructing a power failure warning model to be evaluated provided in an embodiment of the present application;
[0025] Figure 3 A schematic diagram of the structure of an electric power fault warning system based on an AI large model provided in an embodiment of the present application. DETAILED DESCRIPTION
[0026] The embodiment of the present application solves the problem of low efficiency of power fault warning in the prior art by providing a power fault warning method and system based on an AI large model. The method extracts features of historical power system data collected on preset power sensors of the power system to be warned to obtain corresponding historical power system feature data, and constructs a power fault warning model to be evaluated based on the extracted historical power system feature data. Then, the constructed power fault warning model to be evaluated is evaluated to obtain a first evaluation result, and the first evaluation result is compared with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power fault warning model. Then, the initial power system data of the power system to be warned within a preset time interval is collected in real time, and the initial power system data is preprocessed to obtain power system data. Finally, the acquired power system data is used in combination with the first power fault warning model to obtain a warning result. The fault position of the power system to be warned is located according to the warning result, and the positioning result is visualized, thereby improving the efficiency of power fault warning.
[0027] The technical solution in the embodiment of the present application is to solve the problem of low efficiency of power failure warning mentioned above, and the overall idea is as follows:
[0028] The method extracts features from the collected historical power system data of the power system to be warned, constructs a power fault warning model to be evaluated based on the extracted historical power system feature data, and then performs model evaluation on the constructed power fault warning model to be evaluated to obtain a first power fault warning model. Subsequently, the power system data of the power system to be warned within a preset time interval is collected in real time, and a power fault warning is performed on the power system data to obtain a warning result. Finally, the fault position of the power system to be warned is located according to the warning result, and the positioning result is visualized, thereby achieving the effect of improving the efficiency of power fault warning.
[0029] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.
[0030] like Figure 1As shown, it is a flow chart of a power fault warning method based on an AI big model provided in an embodiment of the present application. The power fault warning method based on an AI big model provided in an embodiment of the present application includes the following steps: S1, model construction, feature extraction of historical power system data collected on preset power sensors of the power system to be warned to obtain corresponding historical power system feature data, and constructing a power fault warning model to be evaluated based on the extracted historical power system feature data; S2, model evaluation, model evaluation of the constructed power fault warning model to be evaluated to obtain a first evaluation result, and compare the first evaluation result with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power Fault warning model, otherwise the model is rebuilt, the first evaluation result is used to evaluate the model accuracy and warning ability of the power fault warning model to be evaluated, and the first power fault warning model is used to perform power fault warning on the power system to be warned; S3, data acquisition, real-time acquisition of initial power system data of the power system to be warned within a preset time interval, and preprocessing the initial power system data to obtain power system data, and the preprocessing is used to remove data missing and data anomalies of the initial power system data; S4, fault warning, combined with the first power fault warning model, the acquired power system data is used to perform power fault warning to obtain warning results, and the fault position of the power system to be warned is located according to the warning results, and the positioning results are visualized.
[0031] In this embodiment, while obtaining the early warning result, the cause of the fault can also be predicted; the causes of the fault generally include power outage, voltage anomaly, current anomaly, power equipment failure, short circuit and power system instability; power outage is generally caused by power supply system failure, equipment failure, weather influence, human factors and other reasons; voltage anomaly may be manifested as voltage too high or too low, resulting in equipment damage, equipment malfunction and power transmission interruption, and the reasons generally include power system overload, poor voltage regulation, equipment failure, etc.; current anomaly may be manifested as current overload, current fluctuation and current imbalance, etc., which may cause equipment damage, overheating and fire, etc., and the reasons generally include equipment failure, power load fluctuation, short circuit, etc.; power equipment failure includes transformer failure, switchgear failure, poor wire contact, cable breakage and insulation failure, etc., which are generally caused by aging, lack of maintenance, overload and external damage, etc.; short circuit is the current in the circuit directly flowing through two interconnected nodes, which is easy to cause equipment damage, fire, power system interruption and other consequences, and short circuit is generally caused by equipment failure, external damage and operation error, etc. ; The instability of the power system is manifested in frequency fluctuations, voltage fluctuations and unbalanced power load, which will lead to equipment damage, unstable power supply and other problems. The reasons generally include power load fluctuations, generator failures, transmission line overloads, etc.; Specific targeted analysis can be carried out in combination with actual conditions; The preset power sensors mainly include temperature sensors, current sensors and voltage sensors; The power fault warning model to be evaluated is an initial model of the association between historical power system characteristic data and power fault conditions; The historical power system data is used to describe the nature of the historical data of the power system to be warned where the preset power sensor is located; The power system data represents the parameters of the power system to be warned during operation, and the parameters of the power system to be warned during operation generally include voltage, current, power, frequency, impedance, load, generator parameters, transformer parameters and line parameters; Generator parameters generally include the rated power, voltage, speed and excitation current of the generator, transformer parameters generally include the rated capacity, transformation ratio, resistance, reactance and capacitance of the transformer, and line parameters generally include the resistance, inductance, capacitance and length of the line; The efficiency of power fault warning is improved.
[0032] Further, such as Figure 2As shown, it is a flow chart for constructing the power fault warning model to be evaluated provided in an embodiment of the present application, and the specific construction process of the power fault warning model to be evaluated is as follows: Step one, construct a feature data set, the feature data set is a collection of historical power system feature data, and divide the feature data set into a training feature data set and a verification feature data set according to a preset ratio; Step two, obtain a training machine model, the training machine model is a converged model obtained by training a preset machine learning model using the training feature data set, and the preset machine learning model is a model selected according to the characteristics of the historical power system feature data; Step three, obtain a verification machine model, the verification machine model is a model obtained by model verification of the training machine model using the verification feature data set.
[0033] In this embodiment, a training feature data set is used to train the model, and the model parameters are adjusted through an optimization algorithm to minimize the loss function; a verification feature data set is used to evaluate the model performance, and the model hyperparameters are adjusted to improve the performance; thereby achieving an improvement in the accuracy of warnings for power systems to be warned.
[0034] Furthermore, a specific evaluation method for the model accuracy of the power fault warning model to be evaluated is as follows: obtaining performance evaluation parameters of the power fault warning model to be evaluated, the performance evaluation parameters including accuracy, precision and recall rate; obtaining the accuracy evaluation index of the power fault warning model to be evaluated based on the obtained performance evaluation parameters and combined with the influence of the corresponding data on the accuracy evaluation index, the accuracy evaluation index is used to describe the model accuracy of the power fault warning model to be evaluated.
[0035] In this embodiment, the accuracy evaluation index describes the accuracy of the model. In addition to being directly obtained through the accuracy rate, the model accuracy can also be more accurately calculated through the following formula; the accuracy evaluation index is calculated using the following formula: Where, e represents a natural constant, i represents the number of the power fault warning model to be evaluated, i = 1, 2, ..., I, I is the total number of power fault warning models to be evaluated, ZQ i is the accuracy evaluation index of the i-th power fault warning model to be evaluated, Q i is the accuracy of the i-th power fault warning model to be evaluated, Q0 is the model reference accuracy, ΔQ is the model accuracy reference deviation, J i is the accuracy of the i-th power fault warning model to be evaluated, J0 is the model reference accuracy, ΔJ is the model accuracy reference deviation, H iis the recall rate of the i-th power fault warning model to be evaluated, H0 is the model reference recall rate, ΔH is the model recall rate reference deviation, a1 is the influence of the absolute deviation of accuracy on the accuracy evaluation index, a2 is the influence of the absolute deviation of precision on the accuracy evaluation index, and a3 is the influence of the absolute deviation of recall on the accuracy evaluation index; accuracy is used to describe the proportion of samples correctly predicted by the power fault warning model to be evaluated, precision is used to describe the proportion of samples that are truly positive among all samples predicted to be positive, and recall is used to describe the proportion of samples that are correctly predicted to be positive among all samples that are actually positive; the numerical evaluation of the accuracy of the power fault warning model to be evaluated is realized.
[0036] Furthermore, a specific evaluation method for early warning capability is as follows: obtaining the essential data and data distribution score of the power fault early warning model to be evaluated, the essential data including data quality data, model complexity data and training data volume, and the data distribution score being the score used by preset professionals to evaluate the early warning capability of the power fault early warning model to be evaluated; obtaining the early warning capability evaluation index of the power fault early warning model to be evaluated based on the obtained essential data and data distribution score, and the early warning capability evaluation index is used to describe the early warning capability of the power fault early warning model to be evaluated.
[0037] In this embodiment, the essential data is positively correlated with the early warning capability evaluation index; the key indicators related to the early warning capability generally include voltage and current waveforms, frequency, power factor, transformer temperature, cable temperature, load curve, load prediction error, equipment vibration, equipment noise, alarm response time and alarm accuracy; and the numerical evaluation of the early warning capability of the power fault early warning model to be evaluated is realized.
[0038] Furthermore, the specific process of power fault warning is as follows: the power system data of the power system to be warned is compared and analyzed with the reference power system data and combined with the warning situation score to obtain the power fault warning index, the power fault warning index is used to describe the fault degree of the power system to be warned, and the warning situation score is the score of the preset professionals to evaluate the warning distribution of the power system to be warned; the obtained power fault warning index is compared with the preset threshold to obtain the warning result, if the corresponding preset threshold is met, it indicates that the corresponding power system to be warned has a power fault, otherwise it indicates that the corresponding power system to be warned does not have a power fault.
[0039] In this embodiment, the power failure warning is made more precise.
[0040] Furthermore, a specific method for obtaining the power fault warning index of the power system to be warned is as follows: obtain power system data and a warning situation score of the power system to be warned, the power system data including power system current, power system voltage, power system frequency, power system impedance and power system load; obtain the power fault warning index of the power system to be warned based on the obtained power system data and warning situation score and in combination with the warning weight, the warning weight including the power system weight and the warning situation weight, the power system weight indicating the degree of influence of the power system data on the power fault warning index, the warning situation weight indicating the degree of influence of the warning situation score on the power fault warning index, and the power system weight including a first power system weight, a second power system weight, a third power system weight, a fourth power system weight and a fifth power system weight.
[0041] In this embodiment, a numerical evaluation of the fault degree of the power system to be warned is achieved.
[0042] Furthermore, the specific steps for locating the fault position are as follows: according to the early warning results and in combination with the abnormal detection method, abnormal data points of the power system to be warned are detected to obtain abnormal data points, and the abnormal detection method is used to detect abnormal data points of the power system to be warned; the abnormal data points obtained by analyzing the power fault early warning model are whether they are the fault locations. If they are the fault locations, a real-time monitoring interface is created, otherwise monitoring is continued, and the real-time monitoring interface is used to display the power system data of the power system to be warned; a fault location heat map is drawn based on the power system data of the power system to be warned monitored in real time, and the fault location heat map is used to describe the positioning results of the fault location of the power system to be warned.
[0043] In this embodiment, the anomaly detection method determines the type, characteristics and possible impact range of the abnormal event, and then locates the fault using a model or rule based on the characteristics and impact range of the abnormal event, analyzes the possible fault location in combination with real-time monitoring data and historical records, and then narrows the range of the fault location based on the fault type and equipment status, and finally verifies it through real-time monitoring data; the abnormal data point represents the suspected fault location in the power system to be warned; the power system data to be warned mainly includes current, voltage and frequency, etc., and uses elements such as dashboards, curve charts and tables to clearly display the changes in the status of each device and power parameters; the fault location heat map displays the possible area of the fault location, the color depth indicates the possibility, and provides interactive tools so that users can freely explore the heat map to find potential fault locations; the efficiency of locating and detecting the fault location is improved.
[0044] Furthermore, the early warning capability evaluation index describes the early warning capability of the model. The early warning capability of the model can be directly obtained through the confusion matrix and ROC curve, and can also be calculated more accurately through the following formula; the early warning capability evaluation index is calculated using the following formula: Where i represents the number of the power failure warning model to be evaluated, i = 1, 2, ..., I, I is the total number of power failure warning models to be evaluated, k represents the number of the preset professional, k = 1, 2, ..., K, K is the total number of preset professionals, YJ i is the early warning capability evaluation index of the i-th power fault early warning model to be evaluated, L i is the data quality data of the i-th power fault warning model to be evaluated, L0 is the model reference data quality data, Z i is the model complexity data of the i-th power fault warning model to be evaluated, Z0 is the reference model complexity data, S i is the amount of training data for the ith power fault warning model to be evaluated, S0 is the amount of model reference training data, is the data distribution score of the i-th power fault warning model to be evaluated obtained by the k-th preset professional, b1 is the correction factor of the sum of absolute deviations of essential data, and b2 is the correction factor of the average data distribution score; the sum of absolute deviations of essential data is the sum of absolute deviations of data quality data, model complexity data and training data volume of the power fault warning model to be evaluated.
[0045] In this embodiment, data quality data is obtained by professionals comparing and analyzing the actual situation with preset data, and model complexity data is obtained by professionals based on the acquired data stream analysis; a more accurate evaluation of the warning capability of the power fault warning model to be evaluated is achieved.
[0046] Furthermore, the power failure warning index describes the degree of power system failure. In addition to being directly obtained through power system monitoring data, the degree of power system failure can also be more accurately calculated using the following formula; the power failure warning index is calculated using the following formula: In the formula, e represents a natural constant, y represents the number of the power system to be warned, y = 1, 2, ..., Y, Y is the total number of power systems to be warned, k represents the number of the preset professional personnel, k = 1, 2, ..., K, K is the total number of preset professional personnel, GZ y is the power failure warning index of the yth power system to be warned, is the warning score of the yth power system to be warned obtained by the kth preset professional, I y is the power system current of the yth power system to be warned, I0 is the reference power system current, U yis the power system voltage of the yth power system to be warned, U0 is the reference power system voltage, V y is the power system frequency of the yth power system to be warned, V0 is the reference power system frequency, W y is the power system impedance of the yth power system to be warned, W0 is the reference power system impedance, F y is the power system load of the yth power system to be warned, F0 is the reference power system load, c1 is the warning situation weight, c2 is the first power system weight, c3 is the second power system weight, c4 is the third power system weight, c5 is the fourth power system weight, and c6 is the fifth power system weight.
[0047] In this embodiment, the warning situation weight is the influence of the average warning situation score on the power fault warning index, the first power system weight is the influence of the power system current ratio on the power fault warning index, the second power system weight is the influence of the power system voltage ratio on the power fault warning index, the third power system weight is the influence of the power system frequency ratio on the power fault warning index, the fourth power system weight is the influence of the power system impedance ratio on the power fault warning index, and the fifth power system weight is the influence of the power system load ratio on the power fault warning index; a more accurate assessment of the fault degree of the power system to be warned is achieved.
[0048] like Figure 3As shown, it is a structural schematic diagram of an electric fault warning system based on an AI large model provided in an embodiment of the present application. An electric fault warning system based on an AI large model provided in an embodiment of the present application includes a model construction module, a model evaluation module, a data acquisition module and a fault warning module; wherein, the model construction module is used to extract features of historical electric system data collected on preset electric sensors of the electric system to be warned to obtain corresponding historical electric system feature data, and construct a to-be-assessed electric fault warning model based on the extracted historical electric system feature data; the model evaluation module is used to perform model evaluation on the constructed to-be-assessed electric fault warning model to obtain a first evaluation result, and compare the first evaluation result with the electric fault warning evaluation index. If the electric fault warning evaluation index is met, , the corresponding model is recorded as the first power fault warning model, otherwise the model is rebuilt, the first evaluation result is used to evaluate the model accuracy and warning ability of the power fault warning model to be evaluated, and the first power fault warning model is used to provide power fault warning for the power system to be warned; the data acquisition module is used to collect the initial power system data of the power system to be warned within a preset time interval in real time, and pre-process the initial power system data to obtain the power system data, and the pre-processing is used to remove data missing and data anomalies of the initial power system data; the fault warning module is used to perform power fault warning on the acquired power system data in combination with the first power fault warning model to obtain a warning result, and locate the fault position of the power system to be warned according to the warning result and visualize the positioning result.
[0049] In this embodiment, the systematization of power failure early warning is achieved.
[0050] The technical scheme in the above-mentioned embodiment of the present application has at least the following technical effects or advantages: relative to the electric power equipment fault warning method disclosed in the publication number: CN117214626A, the embodiment of the present application obtains the essential data and data distribution score of the electric power fault warning model to be evaluated, and then calculates the warning capability evaluation index of the electric power fault warning model to be evaluated based on the acquired essential data and data distribution score, and finally evaluates the warning capability of the electric power fault warning model to be evaluated through the calculated warning capability evaluation index, thereby realizing the digitization of the warning capability of the electric power fault warning model to be evaluated, and then realizes a more accurate evaluation of the warning capability of the electric power fault warning model to be evaluated; relative to the remote power line fault warning monitoring system and method disclosed in the publication number: CN116660688B, the embodiment of the present application obtains the electric power system data and the warning situation score of the electric power system to be warned, and then calculates the electric power fault warning index of the electric power system to be warned based on the acquired electric power system data and the warning situation score, and combines the warning weight, and finally evaluates the fault degree of the electric power system to be warned through the calculated electric power fault warning index, thereby realizing the digitization of the fault degree of the electric power system to be warned, and then realizes a more accurate evaluation of the fault degree of the electric power system to be warned.
[0051] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0052] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0053] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0054] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0055] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0056] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A power failure early warning method based on AI large model, characterized in that: The following steps are involved: S1, extracting features from the historical power system data collected from the preset power sensors of the power system to be warned to obtain corresponding historical power system feature data, and constructing a power fault warning model to be evaluated based on the extracted historical power system feature data; S2, performing a model evaluation on the constructed power fault warning model to be evaluated to obtain a first evaluation result, and comparing the first evaluation result with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power fault warning model, otherwise the model is reconstructed, and the first evaluation result is used to evaluate the model accuracy and warning capability of the power fault warning model to be evaluated, and the first power fault warning model is used to perform power fault warning on the power system to be warned; S3, collecting in real time the initial power system data of the power system to be warned within a preset time interval, and preprocessing the initial power system data to obtain power system data, wherein the preprocessing is used to remove data missing and data anomalies of the initial power system data; S4, combining the first power fault warning model to perform power fault warning on the acquired power system data to obtain a warning result, and locating the fault position of the power system to be warned according to the warning result and visualizing the positioning result; The specific process of the power failure early warning is as follows: Compare and analyze the acquired power system data of the power system to be warned with the reference power system data and combine the warning situation score to obtain the power fault warning index, the power fault warning index is used to describe the fault degree of the power system to be warned, and the warning situation score is the score of the warning distribution of the power system to be warned evaluated by the preset professionals; The obtained power fault warning index is compared with a preset threshold to obtain a warning result. If the corresponding preset threshold is met, it indicates that the corresponding power system to be warned has a power fault, otherwise it indicates that the corresponding power system to be warned does not have a power fault; The specific method for obtaining the power failure warning index of the power system to be warned is as follows: Acquire power system data and a warning situation score of the power system to be warned, wherein the power system data includes power system current, power system voltage, power system frequency, power system impedance and power system load; Obtaining a power failure warning index of the power system to be warned according to the acquired power system data and the warning situation score and in combination with the warning weight, wherein the warning weight includes a power system weight and a warning situation weight, wherein the power system weight indicates the influence of the power system data on the power failure warning index, and the warning situation weight indicates the influence of the warning situation score on the power failure warning index, and the power system weight includes a first power system weight, a second power system weight, a third power system weight, a fourth power system weight, and a fifth power system weight; The power failure warning index is calculated using the following formula: In the formula, e represents a natural constant, y represents the number of the power system to be warned, y = 1, 2, ..., Y, Y is the total number of power systems to be warned, k represents the number of the preset professional personnel, k = 1, 2, ..., K, K is the total number of preset professional personnel, GZ y is the power failure warning index of the yth power system to be warned, is the warning situation score of the yth power system to be warned obtained by the kth preset professional, Iy is the power system current of the yth power system to be warned, I0 is the reference power system current, Uy is the power system voltage of the yth power system to be warned, U0 is the reference power system voltage, Vy is the power system frequency of the yth power system to be warned, V0 is the reference power system frequency, Wy is the power system impedance of the yth power system to be warned, W0 is the reference power system impedance, Fy is the power system load of the yth power system to be warned, F0 is the reference power system load, c1 is the warning situation weight, c2 is the first power system weight, c3 is the second power system weight, c4 is the third power system weight, c5 is the fourth power system weight, and c6 is the fifth power system weight.
2. The power failure early warning method based on AI large model as claimed in claim 1, characterized in that: The specific construction process of the power failure early warning model to be evaluated is as follows: Step 1: construct a feature data set, where the feature data set is a collection of historical power system feature data, and divide the feature data set into a training feature data set and a verification feature data set according to a preset ratio; Step 2: obtaining a training machine model, wherein the training machine model is a converged model obtained by training a preset machine learning model using a training feature data set, wherein the preset machine learning model is a model selected according to the characteristics of historical power system feature data; Step three: obtain a verification machine model, where the verification machine model is a model obtained by performing model verification on the training machine model using a verification feature data set.
3. The power failure early warning method based on AI large model as claimed in claim 1, characterized in that: The specific evaluation method of the model accuracy of the power failure early warning model to be evaluated is as follows: Obtaining performance evaluation parameters of the power fault warning model to be evaluated, wherein the performance evaluation parameters include accuracy, precision and recall rate; The accuracy evaluation index of the power fault warning model to be evaluated is obtained according to the obtained performance evaluation parameters and combined with the influence of the corresponding data on the accuracy evaluation index. The accuracy evaluation index is used to describe the model accuracy of the power fault warning model to be evaluated.
4. The power failure early warning method based on AI large model as claimed in claim 1, characterized in that: The specific evaluation method of the early warning capability is as follows: Obtaining essential data and data distribution scores of the power fault early warning model to be evaluated, wherein the essential data includes data quality data, model complexity data, and training data volume, and the data distribution score is a score for evaluating the early warning capability of the power fault early warning model to be evaluated by a preset professional; An early warning capability evaluation index of the power fault early warning model to be evaluated is obtained according to the acquired essential data and data distribution score, and the early warning capability evaluation index is used to describe the early warning capability of the power fault early warning model to be evaluated.
5. The power failure early warning method based on AI large model as claimed in claim 1, characterized in that: The specific steps for locating the fault position are as follows: According to the early warning result and in combination with an abnormality detection method, abnormal data point detection is performed on the power system to be warned to obtain abnormal data points, wherein the abnormality detection method is used to detect abnormal data points of the power system to be warned; Whether the abnormal data point obtained by analyzing the power fault early warning model is the fault location, if it is the fault location, a real-time monitoring interface is created, otherwise the monitoring is continued, and the real-time monitoring interface is used to display the power system data of the power system to be warned; A fault location heat map is drawn based on the power system data of the power system to be warned that is monitored in real time, and the fault location heat map is used to describe the location result of the fault position of the power system to be warned.
6. The power failure early warning method based on AI large model as claimed in claim 4, characterized in that: The early warning capability evaluation index is calculated using the following formula: Wherein, i represents the number of the power fault warning model to be evaluated, i=1,2,...,I, I is the total number of power fault warning models to be evaluated, k represents the number of the preset professional, k=1,2,...,K, K is the total number of preset professionals, YJi is the warning capability evaluation index of the ith power fault warning model to be evaluated, Li is the data quality data of the ith power fault warning model to be evaluated, L0 is the model reference data quality data, Zi is the model complexity data of the ith power fault warning model to be evaluated, Z0 is the reference model complexity data, Si is the training data volume of the ith power fault warning model to be evaluated, S0 is the model reference training data volume, is the data distribution score of the i-th power fault warning model to be evaluated obtained by the k-th preset professional, b1 is the correction factor of the sum of absolute deviations of essential data, and b2 is the correction factor of the average data distribution score; the sum of absolute deviations of essential data is the sum of absolute deviations of data quality data, model complexity data and training data volume of the power fault warning model to be evaluated.
7. A system using a power failure early warning method based on an AI large model as described in any one of claims 1 to 6, characterized in that: include: Model building module, model evaluation module, data acquisition module and fault warning module; The model building module is used to extract features from the historical power system data collected on the preset power sensors of the power system to be warned to obtain corresponding historical power system feature data, and to build a power fault warning model to be evaluated based on the extracted historical power system feature data; The model evaluation module is used to perform a model evaluation on the constructed power fault warning model to be evaluated to obtain a first evaluation result, and compare the first evaluation result with the power fault warning evaluation index. If the power fault warning evaluation index is met, the corresponding model is recorded as the first power fault warning model; otherwise, the model is reconstructed. The first evaluation result is used to evaluate the model accuracy and warning capability of the power fault warning model to be evaluated. The first power fault warning model is used to perform power fault warning on the power system to be warned. The data acquisition module is used to collect the initial power system data of the power system to be warned within a preset time interval in real time, and pre-process the initial power system data to obtain power system data, wherein the pre-processing is used to remove data missing and data anomalies of the initial power system data; The fault warning module is used to perform power fault warning on the acquired power system data in combination with the first power fault warning model to obtain a warning result, and locate the fault position of the power system to be warned according to the warning result and visualize the positioning result.
Citation Information
Cited By
Refrigerating system fault early warning method based on AI model
CN121117976A
A refrigeration system fault early warning method based on an AI model
CN121117976B
