Power distribution network line weight and overload identification method based on artificial intelligence and related device

Abstract

The invention provides a power distribution network line weight and overload identification method based on artificial intelligence and a related device. The method comprises the following steps: collecting multi-source data of a power distribution network line and extracting trend characteristics, fluctuation characteristics and state characteristics of the data; establishing a primary prediction model by adopting a support vector machine, and predicting the baseline load rate of the power distribution network line; a Kalman filter is introduced, and a prediction result is optimized and corrected in combination with a real-time observation value; identifying a load peak value by using an attention mechanism-based long short-term memory network model; dividing the operation state of the power distribution network line into different grades according to the predicted load rate, and adjusting the load grade of the power distribution network line by combining the load peak value, the state characteristic and the fluctuation characteristic; and giving an alarm or early warning according to the current load grade of the power distribution network line. The method can achieve the dynamic evaluation and grading of the operation state, achieves the precise warning and early warning according to different load grades, effectively guarantees the stable operation of a power grid, and reduces the fault loss.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network, and in particular relates to a distribution network line heavy overload identification method and related devices based on artificial intelligence. Background Art

[0002] With the global energy transformation and the growth of social electricity demand, the scale and complexity of the power system have increased dramatically. Smart grid has become the core direction of power development, aiming to use advanced technology to achieve intelligent and efficient operation and meet the diverse power needs of modern society.

[0003] As a key link facing users, the operating status of the distribution network is related to the reliability of power supply. The distribution network line power supply risk assessment technology based on artificial intelligence, which uses machine learning and other means to analyze operating data, is of great significance to improving the management level of the distribution network and is an important part of the construction of smart grids.

[0004] In the actual operation of the distribution network, there are often emergency situations that require real-time decision-making, such as sudden severe weather causing sudden changes in power load. Although some algorithms have the ability to process large amounts of data, the speed of data processing and analysis cannot meet the real-time requirements in the face of these emergency situations. For example, when a line is suddenly overloaded, due to the lag in data processing, an early warning signal cannot be issued in time, and the operation and maintenance personnel cannot take timely measures, which may cause power outages, causing economic losses and inconvenience to users. Summary of the invention

[0005] In view of this, the present invention aims to provide a distribution network line heavy overload identification method and related devices based on artificial intelligence to solve the above-mentioned problems existing in the existing distribution network line power supply risk assessment technology.

[0006] In order to achieve the above object, the technical solution provided by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for identifying heavy overload of distribution network lines based on artificial intelligence, comprising the following steps:

[0008] Collect and preprocess multi-source data of distribution network lines, and extract trend characteristics, fluctuation characteristics and state characteristics of data from the pre-processed multi-source data;

[0009] A primary prediction model is constructed using support vector machine, which uses the extracted trend features, fluctuation features, and state features as input to predict the baseline load rate of distribution network lines;

[0010] The Kalman filter is introduced to optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations;

[0011] Using the long short-term memory network model based on the attention mechanism, the load data processed by feature engineering is input into the model to capture the temporal dependency of the load data and identify the load peak.

[0012] The distribution network line operating status is divided into different levels according to the predicted load rate, and the load level of the distribution network line is adjusted in combination with the load peak, state characteristics and fluctuation characteristics;

[0013] Issue an alarm or warning based on the current load level of the distribution network line.

[0014] Furthermore, the multi-source data includes:

[0015] Meteorological data, user electricity consumption behavior data, historical load data, distribution network equipment data and geographic information data.

[0016] Further, the preprocessing includes the following steps:

[0017] The collected multi-source data are cleaned, and the improved box plot method combined with wavelet transform is used to identify and remove outliers; the outliers are determined according to the following formula:

[0018]

[0019] In the formula, is the outlier identification result, is a single data point in multi-source data, is the lower quartile, is the adjustment coefficient, is the interquartile range after multi-scale wavelet transform, , is the upper quartile;

[0020] Missing values ​​are filled using linear interpolation, mean filling, or time series-based forecasting.

[0021] Normalize the cleaned and filled data.

[0022] Furthermore, when using support vector machine to construct a primary prediction model, it includes:

[0023] The radial basis kernel function is selected as the kernel function;

[0024] The optimal penalty factor and kernel function parameters are selected from the preset parameter range through cross-validation method;

[0025] The extracted trend features, fluctuation features and state features are weighted, and different weights are assigned according to the importance of the features to the baseline load rate prediction;

[0026] The support vector machine model is trained using the trend features, fluctuation features and state features after feature weighting as input and the corresponding historical data of the baseline load rate of the distribution network lines as output to obtain a primary prediction model.

[0027] Furthermore, the Kalman filter is introduced to optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations, including:

[0028] The baseline load rate of the distribution network line is set as the state vector, and the initial estimated value of the state vector is determined by the primary prediction result; at the same time, the initial covariance matrix of the state estimation error, the process noise covariance matrix and the observation noise covariance matrix are initialized;

[0029] According to the established state transfer rules, the state vector is predicted at each moment and the influence of process noise is considered. The covariance matrix of the state estimation error at the previous moment is used to predict the covariance matrix at the current moment.

[0030] Collect load data in real time to obtain observation values, calculate Kalman gain and update the state estimation error covariance matrix;

[0031] Continuously monitor load fluctuations and recalculate the noise covariance matrix if they exceed a preset threshold;

[0032] The sliding window method is used to update the state estimation and parameters to ensure the latest load changes are tracked, thereby optimizing the correction prediction results.

[0033] Furthermore, the load level classification and adjustment include:

[0034] When the load rate is between 0 and 0.8, it is classified as normal operation;

[0035] When the load rate is between 0.8 and 1.0, it is classified as heavy load operation state;

[0036] When the load rate is greater than 1.0, it is classified as an overload operation state;

[0037] If the load peak exceeds the preset proportion of the recent average peak, and the key parameters of the equipment in the status characteristics are beyond the normal range, and the fluctuation characteristics show that the load fluctuation standard deviation is greater than the set value, the current load level will be increased by one level; if only some of the conditions are met, the load level will be fine-tuned according to the preset adjustment coefficient.

[0038] Furthermore, an alarm or warning is issued according to the current load level, including:

[0039] When the load level is in heavy load operation state, a general warning is triggered and the operation and maintenance personnel are notified via SMS or email. At the same time, detailed operation information of the line is displayed on the operation and maintenance system interface, including load rate, load peak, status of related equipment, etc.;

[0040] When the load level is in an overloaded operating state, a high priority alarm is triggered. In addition to SMS and email notifications, a voice reminder is sent to the operation and maintenance personnel's mobile terminal, and an emergency work order is automatically generated.

[0041] In a second aspect, the present invention provides a distribution network line heavy overload identification device based on artificial intelligence, comprising:

[0042] A data feature extraction module is used to collect and preprocess multi-source data of distribution network lines, and extract trend features, fluctuation features and state features of the data from the pre-processed multi-source data;

[0043] A primary prediction module is used to construct a primary prediction model using a support vector machine, taking the extracted trend features, fluctuation features and state features as inputs to predict the baseline load rate of the distribution network lines;

[0044] The prediction correction module is used to introduce the Kalman filter and optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations;

[0045] The recognition module is used to use the long short-term memory network model based on the attention mechanism to input the load data processed by feature engineering into the model, capture the temporal dependency of the load data, and identify the load peak;

[0046] A level classification module is used to classify the operating status of the distribution network line into different levels according to the predicted load rate, and adjust the load level of the distribution network line in combination with the load peak, state characteristics and fluctuation characteristics;

[0047] The graded alarm and warning module is used to issue an alarm or warning according to the current load level of the distribution network line.

[0048] In a third aspect, the present invention provides a computer device, the device comprising a processor and a memory:

[0049] The memory is used to store the computer program and send the instructions of the computer program to the processor;

[0050] The processor executes a distribution network line heavy overload identification method based on artificial intelligence as described in the first aspect according to the instructions of the computer program.

[0051] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a distribution network line heavy overload identification method based on artificial intelligence as in the first aspect.

[0052] In summary, the present invention provides a method and related device for identifying heavy overload of distribution network lines based on artificial intelligence, by collecting multi-source data of distribution network lines and preprocessing, extracting trend characteristics, fluctuation characteristics and state characteristics of data from the preprocessed multi-source data; using support vector machine to build a primary prediction model, using the extracted trend characteristics, fluctuation characteristics and state characteristics as input, predicting the baseline load rate of distribution network lines; introducing Kalman filter, optimizing and correcting the prediction results obtained by the primary prediction model in combination with real-time observations; using a long short-term memory network model based on attention mechanism, inputting the load data processed by feature engineering into the model, capturing the temporal dependency of load data, and identifying load peak; dividing the operating status of distribution network lines into different levels according to the predicted load rate, adjusting the load level of distribution network lines in combination with load peak, state characteristics and fluctuation characteristics; and giving an alarm or warning according to the current load level of distribution network lines. The present invention can realize dynamic evaluation and classification of operating status, and then accurately give an alarm and warning according to different load levels, comprehensively improve the comprehensiveness and accuracy of heavy overload identification of distribution network lines, effectively ensure stable operation of power grid and reduce failure losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in 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 described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 A flowchart of a method for identifying heavy overload of distribution network lines based on artificial intelligence provided by an embodiment of the present invention;

[0055] Figure 2 A block diagram of a distribution network line heavy overload identification device based on artificial intelligence provided by an embodiment of the present invention;

[0056] Figure 3 A block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] See also Figure 1 The embodiment of the present invention provides a method for identifying heavy overload of distribution network lines based on artificial intelligence, comprising the following steps:

[0059] S1: Collect and preprocess the multi-source data of distribution network lines, and extract the trend characteristics, fluctuation characteristics and state characteristics of the data from the preprocessed multi-source data.

[0060] It should be noted that multi-source data refers to data from multiple different sources, such as voltage, current, power and other data of distribution network lines, as well as environmental data (such as temperature, humidity and other factors that may affect the load). Trend characteristics can reflect the long-term trend of data changes over time, such as the trend of load gradually increasing or decreasing over time. Fluctuation characteristics can reflect the fluctuations of data within a certain time range, such as periodic fluctuations of load. State characteristics can describe the relevant characteristics of the current operating state of the distribution network line, such as the operating parameters of the equipment, the switch status, etc.

[0061] This step collects various types of data on the distribution network lines through a variety of data acquisition devices, and then uses data preprocessing technology to process the data, and uses statistical analysis, signal processing and other methods to extract trends, fluctuations and state characteristics from the processed data.

[0062] S2: A primary prediction model is constructed using support vector machine, taking the extracted trend features, fluctuation features, and state features as input to predict the baseline load rate of the distribution network lines.

[0063] It should be noted that the support vector machine (SVM) is a supervised machine learning algorithm that separates different categories of data by finding an optimal hyperplane and is used to predict continuous values ​​in regression problems. The baseline load rate refers to a benchmark value of the load level of the distribution network line under normal operation.

[0064] In this step, the extracted trend, fluctuation and state features are used as the input feature vectors of the SVM, and the relationship between these features and the baseline load rate is learned by training the SVM model to establish a prediction model.

[0065] S3: Introduce the Kalman filter and optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations.

[0066] It should be noted that the Kalman filter is an algorithm that uses the linear system state equation to optimally estimate the system state through system input and output observation data, and can effectively handle noise and uncertainty.

[0067] The Kalman filter in this step continuously adjusts the prediction results to make them closer to the true value through state prediction and update steps based on the prediction values ​​of the primary prediction model and the real-time observed data.

[0068] S4: Using the long short-term memory network model based on the attention mechanism, the load data processed by feature engineering is input into the model to capture the temporal dependency of the load data and identify the load peak.

[0069] It should be noted that the attention mechanism is a technology that allows the model to automatically focus on important information when processing sequence data, thereby improving the model's ability to process key information. Long short-term memory network (LSTM) is a special recurrent neural network (RNN) that can effectively handle long-term dependency problems in long sequence data and avoid gradient vanishing and gradient explosion. Feature engineering is the extraction, transformation, and selection of features on raw data to improve the performance and generalization ability of the model. Temporal dependency is the chronological dependency of data in a time series, that is, the relationship between current data and previous data.

[0070] In this step, the load data processed by feature engineering is input into the LSTM model based on the attention mechanism in time series. The model learns the temporal dependencies of the data and uses the attention mechanism to focus on important time steps, thereby identifying the load peak.

[0071] S5: The operating status of the distribution network lines is divided into different levels according to the predicted load rate, and the load level of the distribution network lines is adjusted in combination with the load peak, state characteristics and fluctuation characteristics.

[0072] It should be noted that this step first sets different thresholds according to the predicted load rate to divide the distribution network line operation status into different levels. Then, combined with the load peak, state characteristics and fluctuation characteristics, the load level is comprehensively evaluated and adjusted to more accurately reflect the actual operation situation.

[0073] S6: Issue an alarm or warning based on the current load level of the distribution network line.

[0074] It should be noted that this step compares the load level determined in step S5 with the preset alarm and warning thresholds, and triggers the alarm or warning mechanism when the corresponding threshold is reached.

[0075] This embodiment provides a method for identifying heavy overloads of distribution network lines based on artificial intelligence. It comprehensively uses a variety of artificial intelligence technologies and data analysis methods, starts with multi-source data, and after preprocessing and feature extraction, uses a support vector machine to predict the baseline load rate, and a Kalman filter optimizes the prediction results. The LSTM model based on the attention mechanism identifies the load peak, and then combines multiple features to divide and adjust the load level. Finally, an alarm or warning is issued according to the load level to achieve accurate identification of heavy overloads of distribution network lines. This method organically combines multiple technologies such as support vector machines, Kalman filters, and LSTM based on attention mechanisms, giving full play to their respective advantages and improving the accuracy and reliability of heavy overload identification. At the same time, not only is the level divided according to the load rate, but it is also dynamically adjusted in combination with the load peak, state characteristics, and fluctuation characteristics, so that the load level is more in line with the actual operating status.

[0076] In one embodiment, the multi-source data includes:

[0077] Meteorological data, user electricity consumption behavior data, historical load data, distribution network equipment data and geographic information data.

[0078] In this embodiment, the multi-source data includes meteorological data, user electricity consumption behavior data, historical load data, distribution network equipment data and geographic information data, which are of great significance for the identification of heavy overload of distribution network lines. Factors such as temperature and humidity in meteorological data will significantly affect users' electricity demand. For example, changes in electricity load caused by heating and cooling in winter and summer will also affect the power generation of distributed power sources, providing external environmental information for predicting load fluctuations. User electricity behavior data records the user's electricity usage time, equipment type, etc. Different user types have different electricity usage patterns. Analyzing their patterns can accurately predict electricity demand and discover peak hours and high-load areas. Historical load data reflects past load changes. Through statistical analysis, periodic, seasonal and long-term trends can be mined to assist in predicting future loads and identifying anomalies. Distribution network equipment data includes equipment operating parameters and status, which directly reflects the health status and operating capacity of the equipment and is related to the line load bearing capacity. Geographic information data such as terrain and regional distribution affect the layout of the distribution network and the power supply radius, and then affect the line load. Combining these multi-source data can comprehensively and accurately analyze the load conditions of the distribution network lines, improve the accuracy and reliability of heavy overload identification, and provide strong support for the scientific management and decision-making of the distribution network.

[0079] In one embodiment, preprocessing includes the following steps:

[0080] S11: Clean the collected multi-source data, and use the improved box plot method combined with wavelet transform to identify and remove outliers; the outliers are determined according to the following formula:

[0081]

[0082] In the formula, is the outlier identification result, is a single data point in multi-source data, is the lower quartile, is the adjustment coefficient, is the interquartile range after multi-scale wavelet transform, , The upper quartile.

[0083] In this step, we first conduct a preliminary box plot analysis on the collected multi-source data to obtain the lower quartile of the data. and upper quartile , calculate the conventional interquartile range . Then, the data is processed by multi-scale wavelet transform to remove noise and extract data features, and then the interquartile range of the transformed data is calculated. Next, each data point is compared using the above-mentioned outlier judgment formula to determine whether it is an outlier. If the outlier condition is met, the data point is removed. This combined method can more effectively identify outliers in a noisy environment and improve the quality of the data.

[0084] S12: Missing values ​​are filled using linear interpolation, mean filling, or time series-based prediction filling.

[0085] S13: normalize the cleaned and filled data.

[0086] This step transforms the cleaned and filled data accordingly according to the selected normalization method, thereby scaling the data so that it falls into a specific interval.

[0087] The preprocessing of this embodiment uses an improved box plot method combined with wavelet transform to identify and eliminate outliers, which can effectively eliminate the interference of abnormal data on analysis in a noisy environment and improve data reliability; linear interpolation, mean filling or time series-based prediction filling methods are used to fill missing values, which can restore data integrity and provide more comprehensive data for subsequent analysis and modeling; the cleaned and filled data is normalized to eliminate the impact of dimensions, make different feature data comparable, accelerate model convergence, and improve model accuracy and stability. Overall, these preprocessing steps comprehensively improve the quality and availability of multi-source data, provide a solid and reliable data foundation for accurately identifying heavy overloads in distribution network lines, and thus improve the accuracy and reliability of identifying heavy overloads in distribution network lines.

[0088] In one embodiment, when a primary prediction model is constructed using a support vector machine, the following steps are included:

[0089] S21: The radial basis kernel function (RBF) is selected as the kernel function.

[0090] It should be noted that when constructing an SVM model, after selecting the RBF kernel function, the SVM will use the kernel function to map the input feature data to a high-dimensional space, and then find the optimal hyperplane in this high-dimensional space (for regression problems, it is to find the optimal fitting function). Since the RBF kernel function can flexibly handle nonlinear relationships, it is suitable for complex distribution of data in many practical problems.

[0091] S22: Select the optimal penalty factor and kernel function parameters from a preset parameter range through a cross-validation method.

[0092] It should be noted that the value range of the penalty factor and kernel function parameters is preset first. Then, in the cross-validation process, for each possible parameter combination, the data set is divided, and different subsets are used for training and testing, and the performance indicators of the model on the test set (such as mean square error, etc.) are calculated. By comparing the performance indicators under different parameter combinations, the parameter combination with the best performance indicator is selected as the final penalty factor and kernel function parameter.

[0093] S23: performing feature weighting on the extracted trend features, fluctuation features, and state features, and assigning different weights according to the importance of the features to the baseline load rate prediction.

[0094] It should be noted that there are many ways to determine the importance of features and assign weights. For example, the correlation between each feature and the baseline load rate can be calculated, and features with higher correlations are assigned higher weights; or some feature selection algorithms (such as recursive feature elimination) can be used to evaluate the importance of features. Then, according to the evaluation results, a corresponding weight is assigned to each feature, and the original feature is multiplied by the corresponding weight to obtain the weighted feature.

[0095] S24: Using the trend features, fluctuation features and state features after feature weighting processing as input and the corresponding historical data of the baseline load rate of the distribution network line as output, the support vector machine model is trained to obtain a primary prediction model.

[0096] It should be noted that the trend features, fluctuation features and state features after feature weighting are used to form the input vector, and the historical data of the baseline load rate of the distribution network line corresponding to these features are used as output labels. These input and output data are provided to the support vector machine model. The model adjusts the parameters of the model according to the selected kernel function (RBF kernel function) and the optimal parameters (penalty factor and kernel function parameters) through the optimization algorithm (such as the sequence minimum optimization algorithm, etc.), minimizes the training error, and thus learns the relationship between the input features and the baseline load rate, and obtains a primary prediction model that can be used for prediction.

[0097] When this embodiment uses support vector machine to construct the primary prediction model, radial basis kernel function (RBF) is selected to enable it to handle the nonlinear relationship between the distribution network line characteristics and the baseline load rate, thereby improving the model prediction performance; the cross-validation method is used to select the optimal penalty factor and kernel function parameters from the preset range to avoid blind parameter selection and enhance the model's fit and generalization ability on the training data; the extracted trend, fluctuation and state features are weighted, and weights are assigned according to the importance of the features to highlight key features and suppress interference, thereby helping the model to accurately capture information related to the baseline load rate; the model is trained with weighted features as input and baseline load rate historical data as output, so that the model can learn the inherent laws of the two and accurately predict the baseline load rate. Overall, these steps construct a primary prediction model with excellent performance, laying a solid and reliable foundation for the accurate identification of heavy overloads in the distribution network lines.

[0098] In one embodiment, a Kalman filter is introduced to optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations, including:

[0099] S31: Set the distribution network line baseline load rate as the state vector, and determine the initial estimate of the state vector based on the primary prediction result; at the same time, initialize and set the initial covariance matrix of the state estimation error, the process noise covariance matrix, and the observation noise covariance matrix.

[0100] After the state vector is determined in this step, the output of the primary prediction model is used as the initial estimate of the state vector to provide a starting reference for the subsequent filtering process. At the same time, based on experience or prior knowledge, the covariance matrix of the state estimation error, the process noise covariance matrix, and the observation noise covariance matrix are reasonably initialized. These matrices will be used to adjust the state estimation and error calculation in subsequent filtering iterations.

[0101] S32: According to the established state transfer rules, the state vector is predicted at each moment and the influence of process noise is considered, and the covariance matrix of the state estimation error at the previous moment is used to predict the covariance matrix at the current moment.

[0102] This step predicts the state vector at the current moment based on the state transition rule and the state estimation value at the previous moment, and considers the influence of process noise to make the prediction more consistent with the uncertainty of the actual system. At the same time, the covariance matrix of the state estimation error at the previous moment is used to calculate the covariance matrix at the current moment through a specific formula to update the understanding of the uncertainty of the state estimation error.

[0103] S33: Collect load data in real time to obtain observation values, calculate Kalman gain and update the state estimation error covariance matrix.

[0104] After collecting the load data in real time to obtain the observed values, this step calculates the Kalman gain based on the current predicted state estimation error covariance matrix, the observation matrix, and the observation noise covariance matrix. The Kalman gain determines the weight of the observed value and the predicted value when updating the state estimate. Then, the state estimate is updated using the Kalman gain and the observed value, and the state estimation error covariance matrix is ​​updated at the same time to reflect the uncertainty of the updated state estimate.

[0105] S34: Continuously monitor load fluctuations, and if the load fluctuations exceed a preset threshold, recalculate the noise covariance matrix.

[0106] This step continuously monitors the load fluctuation of the distribution network lines and compares it with the preset threshold. When the load fluctuation exceeds the preset threshold, it means that the operating environment or characteristics of the system may have changed, and the previously set noise covariance matrix may no longer be accurate. Therefore, it is necessary to recalculate the noise covariance matrix to adapt to the new situation and ensure the performance of the Kalman filter.

[0107] S35: The sliding window method is used to update the state estimation and parameters to ensure the latest load changes are tracked, thereby optimizing the correction prediction results.

[0108] This step uses the sliding window method, which only considers the latest data in the window to update the state estimate and parameters. As new load data is continuously collected, the window slides forward, old data is removed, and new data is included. Based on the data in the window, the state estimate, covariance matrix and other parameters are recalculated to reflect the latest changes in the load.

[0109] This embodiment introduces a Kalman filter and takes a series of steps to optimize and correct the results of the primary prediction model, which can effectively deal with the uncertainty and noise in the distribution network lines, improve the accuracy, robustness and real-time performance of the baseline load rate estimation, and provide more reliable basic data for the identification of heavy overloads in the distribution network lines.

[0110] In one embodiment, the load level division and adjustment include:

[0111] When the load rate is between 0 and 0.8, it is classified as normal operation;

[0112] When the load rate is between 0.8 and 1.0, it is classified as heavy load operation state;

[0113] When the load rate is greater than 1.0, it is classified as an overload operation state;

[0114] If the load peak exceeds the preset proportion of the recent average peak, and the key parameters of the equipment in the status characteristics are beyond the normal range, and the fluctuation characteristics show that the load fluctuation standard deviation is greater than the set value, the current load level will be increased by one level; if only some of the conditions are met, the load level will be fine-tuned according to the preset adjustment coefficient.

[0115] This embodiment comprehensively considers multiple factors such as load peak, state characteristics and fluctuation characteristics. When the load peak exceeds the preset proportion of the recent average peak, and the key parameters of the equipment are beyond the normal range, and the load fluctuation standard deviation is greater than the set value, it means that the operation status of the line has deteriorated, and the current load level is increased by one level to more accurately reflect the actual risk. If only some of the conditions are met, it means that the operation status of the line has changed to a certain extent, but has not reached the level of serious deterioration. At this time, the load level is fine-tuned according to the preset adjustment coefficient to make the adjustment of the load level more flexible and accurate.

[0116] In one embodiment, an alarm or warning is issued according to the current load level, including:

[0117] When the load level is in heavy load operation state, a general warning is triggered and the operation and maintenance personnel are notified via SMS or email. At the same time, detailed operation information of the line is displayed on the operation and maintenance system interface, including load rate, load peak, status of related equipment, etc.;

[0118] When the load level is in an overloaded operating state, a high priority alarm is triggered. In addition to SMS and email notifications, a voice reminder is sent to the operation and maintenance personnel's mobile terminal, and an emergency work order is automatically generated.

[0119] In this embodiment, when the system detects that the load level of the distribution network line reaches the overload operation state, the corresponding early warning mechanism is triggered. The mechanism will automatically send a text message to the mobile phone number of the operation and maintenance personnel, send an email to their mailbox, and update and display the detailed operation information of the line on the operation and maintenance system interface, such as load rate, load peak value, and the status of related equipment and other parameters.

[0120] When the load level is detected as an overloaded operating state, the system will immediately trigger a high-priority alarm mechanism. In addition to notifying the operation and maintenance personnel via SMS and email, a voice reminder will also be sent to their mobile terminals to ensure that the operation and maintenance personnel can pay attention to the alarm information as soon as possible. At the same time, the system will automatically generate an emergency work order, record the overload status of the line, relevant equipment information, and recommended treatment measures in detail, and assign the work order to the corresponding operation and maintenance personnel.

[0121] Based on the same inventive concept, the embodiment of the present application also provides a device for identifying heavy overloads of distribution network lines based on artificial intelligence for implementing the above-mentioned method for identifying heavy overloads of distribution network lines based on artificial intelligence. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the embodiment of the device for identifying heavy overloads of distribution network lines based on artificial intelligence provided below can be referred to the limitations of the method for identifying heavy overloads of distribution network lines based on artificial intelligence above, and will not be repeated here.

[0122] See also Figure 2 The present invention provides a distribution network line heavy overload identification device based on artificial intelligence, comprising:

[0123] A data feature extraction module is used to collect and preprocess multi-source data of distribution network lines, and extract trend features, fluctuation features and state features of the data from the pre-processed multi-source data;

[0124] A primary prediction module is used to construct a primary prediction model using a support vector machine, taking the extracted trend features, fluctuation features and state features as inputs to predict the baseline load rate of the distribution network lines;

[0125] The prediction correction module is used to introduce the Kalman filter and optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observations;

[0126] The recognition module is used to use the long short-term memory network model based on the attention mechanism to input the load data processed by feature engineering into the model, capture the temporal dependency of the load data, and identify the load peak;

[0127] A level classification module is used to classify the operating status of the distribution network line into different levels according to the predicted load rate, and adjust the load level of the distribution network line in combination with the load peak, state characteristics and fluctuation characteristics;

[0128] The graded alarm and warning module is used to issue an alarm or warning according to the current load level of the distribution network line.

[0129] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0130] Reference Figure 3 An embodiment of the present invention further provides a computer device, comprising: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, an artificial intelligence-based distribution network line heavy overload identification method as described in any one of the above methods is implemented.

[0131] The computer device may be a desktop computer, a notebook, a PDA, a cloud server or other computing device. The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that Figure 3 It is only an example of a computer device and does not constitute a limitation of the computer device. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0132] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0133] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device. Further, the memory may include both an internal storage unit and an external storage device of the computer device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is to be output.

[0134] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements a distribution network line heavy overload identification method based on artificial intelligence as described in any one of the above methods.

[0135] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, USB flash drive, mobile hard disk, disk or optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0136] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0137] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0138] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0139] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying heavy overload of distribution network lines based on artificial intelligence, characterized in that: The steps include: Collecting and preprocessing multi-source data of distribution network lines, and extracting trend characteristics, fluctuation characteristics and state characteristics of the data from the preprocessed multi-source data; A primary prediction model is constructed using a support vector machine, and the extracted trend feature, fluctuation feature and state feature are used as input to predict the baseline load rate of the distribution network line; A Kalman filter is introduced to optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observation values; Using the long short-term memory network model based on the attention mechanism, the load data processed by feature engineering is input into the model to capture the temporal dependency of the load data and identify the load peak. Dividing the operating status of the distribution network line into different levels according to the predicted load rate, and adjusting the load level of the distribution network line in combination with the load peak, the state characteristics and the fluctuation characteristics; An alarm or pre-warning is issued according to the current load level of the distribution network line.

2. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 1 is characterized in that: The multi-source data includes: Meteorological data, user electricity consumption behavior data, historical load data, distribution network equipment data and geographic information data.

3. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 1 is characterized in that: The pre-processing comprises the following steps: The collected multi-source data is cleaned, and an improved box plot method combined with wavelet transform is used to identify and remove outliers; wherein the outliers are determined according to the following formula: In the formula, is the outlier identification result, is a single data point in the multi-source data, is the lower quartile, is the adjustment coefficient, is the interquartile range after multi-scale wavelet transform, , is the upper quartile; Missing values ​​are filled using linear interpolation, mean filling, or time series-based forecasting. Normalize the cleaned and filled data.

4. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 1 is characterized in that: When using support vector machines to build a primary prediction model, it includes: The radial basis kernel function is selected as the kernel function; The optimal penalty factor and kernel function parameters are selected from the preset parameter range through cross-validation method; The extracted trend feature, the fluctuation feature and the state feature are weighted, and different weights are assigned according to the importance of the feature to the baseline load rate prediction; The trend feature, the fluctuation feature and the state feature after feature weighting are used as input, and the corresponding historical data of the baseline load rate of the distribution network line is used as output to train the support vector machine model, thereby obtaining the primary prediction model.

5. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 1 is characterized in that: The Kalman filter is introduced to optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observation values, including: The distribution network line baseline load rate is set as a state vector, and the initial estimated value of the state vector is determined by the primary prediction result; at the same time, the initial covariance matrix of the state estimation error, the process noise covariance matrix and the observation noise covariance matrix are initialized; According to the established state transfer rules, the state vector is predicted at each moment and the influence of process noise is considered, and the covariance matrix of the state estimation error at the previous moment is used to predict the covariance matrix at the current moment; Collect load data in real time to obtain observation values, calculate Kalman gain and update the state estimation error covariance matrix; Continuously monitor load fluctuations and recalculate the noise covariance matrix if they exceed a preset threshold; The sliding window method is used to update the state estimation and parameters to ensure the latest load changes are tracked, thereby optimizing the correction prediction results.

6. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 1 is characterized in that: The division and adjustment of the load levels include: When the load rate is between 0 and 0.8, it is classified as normal operation; When the load rate is between 0.8 and 1.0, it is classified as heavy load operation state; When the load rate is greater than 1.0, it is classified as an overload operation state; If the load peak exceeds the preset proportion of the recent average peak, and the key parameters of the equipment in the status characteristics are beyond the normal range, and the fluctuation characteristics show that the load fluctuation standard deviation is greater than the set value, the current load level will be increased by one level; if only some of the conditions are met, the load level will be fine-tuned according to the preset adjustment coefficient.

7. The method for identifying heavy overload of distribution network lines based on artificial intelligence according to claim 6 is characterized in that: Produce an alarm or pre-warning according to the current load level, including: When the load level is in a heavy load operation state, a general warning is triggered, and the operation and maintenance personnel are notified via SMS or email. At the same time, detailed operation information of the line is displayed on the operation and maintenance system interface, including load rate, load peak, status of related equipment, etc.; When the load level is in an overloaded operating state, a high priority alarm is triggered. In addition to SMS and email notifications, a voice reminder is sent to the mobile terminal of the operation and maintenance personnel, and an emergency work order is automatically generated.

8. A distribution network line heavy overload identification device based on artificial intelligence, characterized in that: include: A data feature extraction module is used to collect and preprocess multi-source data of distribution network lines, and extract trend features, fluctuation features and state features of the data from the pre-processed multi-source data; A primary prediction module, used to construct a primary prediction model using a support vector machine, and to predict a baseline load rate of a distribution network line by taking the extracted trend feature, the fluctuation feature and the state feature as input; A prediction correction module, used to introduce a Kalman filter and optimize and correct the prediction results obtained by the primary prediction model in combination with real-time observation values; The recognition module is used to use the long short-term memory network model based on the attention mechanism to input the load data processed by feature engineering into the model, capture the temporal dependency of the load data, and identify the load peak; A level classification module, used to classify the operating status of the distribution network line into different levels according to the predicted load rate, and adjust the load level of the distribution network line in combination with the load peak, the state characteristics and the fluctuation characteristics; The graded alarm and warning module is used to issue an alarm or warning according to the current load level of the distribution network line.

9. A computer device, characterized in that: The device comprises a processor and a memory: The memory is used to store a computer program and send instructions of the computer program to the processor; The processor executes the distribution network line heavy overload identification method based on artificial intelligence as described in any one of claims 1-7 according to the instructions of the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for identifying heavy and overloaded distribution network lines based on artificial intelligence as described in any one of claims 1 to 7 is implemented.

Images