Method and system for evaluating influence of power failure event of low-voltage power distribution network

Abstract

The invention discloses a low-voltage power distribution network power failure event influence assessment method and system, and relates to the technical field of low-voltage power distribution network monitoring. Multi-dimensional monitoring data of a low-voltage power distribution network are acquired, the multi-dimensional monitoring data are preprocessed to obtain corresponding target monitoring data, and the target monitoring data are acquired based on a preset feature number; the method comprises the steps of constructing a plurality of feature subgroups according to target monitoring data, performing hierarchical co-evolution on each feature subgroup according to a preset fitness function to obtain corresponding target feature data, and performing influence evaluation on the target feature data by adopting a pre-trained power failure event influence evaluation model to obtain a corresponding influence evaluation result. The technical problems that traditional low-voltage power distribution network power failure event influence evaluation mainly depends on artificial experience, but a large amount of time needs to be consumed for evaluation, misjudgment is prone to occurring, and the reliability of low-voltage power distribution network operation is reduced are solved.

Description

Technical Field

[0001] This invention relates to the field of low-voltage distribution network monitoring technology, and in particular to a method and system for assessing the impact of low-voltage distribution network outage events. Background Technology

[0002] As the "last mile" directly connecting the power system and users, the reliability of low-voltage distribution networks directly affects the continuity of residents' daily lives, normal business operations, and public services. In recent years, with the continuous growth of electricity load (such as the popularization of new energy vehicle charging piles and smart home devices), the extended service life of power grid equipment (the aging rate of equipment in some areas exceeds 30%), and the frequent occurrence of extreme weather (rainstorms, typhoons, and low-temperature freezing), the frequency and scope of low-voltage distribution network outages have shown a significant upward trend. This not only causes direct economic losses (such as production stoppages for commercial users and equipment damage for industrial users), but may also trigger indirect social risks such as traffic signal interruptions and medical equipment shutdowns. Therefore, "impact assessment" of low-voltage distribution network outages has become one of the core requirements of power grid operation and maintenance management.

[0003] Currently, the traditional impact assessment of power outages in low-voltage distribution networks mainly relies on human experience, which requires a lot of time to conduct and is prone to misjudgment, thus reducing the reliability of low-voltage distribution network operation. Summary of the Invention

[0004] This invention provides a method and system for assessing the impact of power outage events in low-voltage distribution networks. It solves the technical problem that traditional assessments of the impact of power outage events in low-voltage distribution networks mainly rely on human experience, which requires a lot of time to conduct and is prone to misjudgment, thus reducing the reliability of low-voltage distribution network operation.

[0005] The first aspect of this invention provides a method for assessing the impact of power outage events in low-voltage distribution networks, comprising:

[0006] Acquire multi-dimensional monitoring data of the low-voltage distribution network, preprocess the multi-dimensional monitoring data, and obtain the corresponding target monitoring data;

[0007] Based on a preset number of features, multiple feature subgroups are constructed according to the target monitoring data;

[0008] Based on a preset adaptation function, each of the feature subgroups is subjected to hierarchical collaborative evolution to obtain the corresponding target feature data;

[0009] The target feature data is assessed using a pre-trained power outage event impact assessment model to obtain the corresponding impact assessment results.

[0010] Optionally, each individual in the feature subgroup corresponds to a feature combination, and the step of performing hierarchical collaborative evolution on each feature subgroup according to a preset adaptation function to obtain the corresponding target feature data includes:

[0011] Each of the aforementioned features is combined and input into a preset adaptation function to obtain multiple adaptation values;

[0012] Based on a preset update function, the feature subgroups are updated according to each of the fitness values ​​to obtain multiple updated feature subgroups;

[0013] Determine whether the number of updates for each of the updated feature subgroups is greater than or equal to a preset iteration threshold;

[0014] If the number of updates of the updated feature subgroup is less than the iteration threshold, the updated feature subgroup is determined as a new feature subgroup, and the process jumps to the step of inputting each of the feature combinations into a preset fitness function to obtain multiple fitness values.

[0015] If the number of updates of the updated feature subgroup is greater than or equal to the iteration threshold, then the fitness value corresponding to the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination.

[0016] Based on a preset information entropy weighting function, feature fusion is performed on each of the target feature combinations to obtain the corresponding target feature data.

[0017] Optionally, the step of updating the feature subgroups according to each fitness value based on a preset update function to obtain multiple updated feature subgroups includes:

[0018] The minimum value among the fitness values ​​corresponding to each feature subgroup is selected as the updated fitness value, and the feature combination associated with each updated fitness value is taken as the optimal feature combination.

[0019] Based on a preset update function, individuals in each feature subgroup are updated according to each optimal feature combination and preset update factor data, resulting in multiple updated feature subgroups.

[0020] Optionally, the step of performing feature fusion on each of the target feature combinations based on a preset information entropy weighting function to obtain the corresponding target feature data includes:

[0021] Each of the target features is combined and input into a preset information entropy weighting function to obtain multiple dynamic entropy weights;

[0022] Based on the dynamic entropy weights, the target feature combinations are weighted and fused to obtain the corresponding target feature data.

[0023] Optionally, the power outage event impact assessment model includes a first network, a fully connected layer, and a second network. The step of using the pre-trained power outage event impact assessment model to assess the impact on the target feature data and obtain the corresponding impact assessment result includes:

[0024] Temporal feature extraction is performed on the target feature data through the first network to obtain the corresponding temporal feature vector;

[0025] The time-series feature vector is nonlinearly mapped through the fully connected layer to obtain the corresponding target time-series feature vector;

[0026] The second network is used to predict the impact of the target time-series feature vector, and the corresponding impact assessment result is obtained.

[0027] Optionally, the training process for the power outage event impact assessment model is as follows:

[0028] Acquire multiple historical target feature data, and generate a target feature set using each of the historical target feature data;

[0029] The target feature set is used as input to train the preset initial power outage event impact assessment model to obtain the corresponding training impact assessment data;

[0030] Calculate the training loss function value of the target feature set based on the training impact assessment data;

[0031] When the training loss function value is greater than or equal to the preset standard loss value, the network parameters of the initial power outage event impact assessment model are adjusted, and the process jumps to the step of training the preset initial power outage event impact assessment model using the target feature set as input to obtain the corresponding training impact assessment data, until the training loss function value is less than the standard loss value.

[0032] When the training loss function value is less than the standard loss value, a power outage event impact assessment model is generated.

[0033] A second aspect of the present invention provides a low-voltage distribution network power outage event impact assessment system, comprising:

[0034] The acquisition module is used to acquire multi-dimensional monitoring data of the low-voltage distribution network, and to preprocess the multi-dimensional monitoring data to obtain the corresponding target monitoring data.

[0035] The construction module is used to construct multiple feature subgroups based on the target monitoring data, according to a preset number of features.

[0036] The co-evolution module is used to perform hierarchical co-evolution of each feature subgroup according to a preset adaptation function to obtain the corresponding target feature data;

[0037] The impact assessment module is used to assess the impact of the target feature data using a pre-trained power outage event impact assessment model, and obtain the corresponding impact assessment results.

[0038] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the low-voltage distribution network power outage event impact assessment method as described in any of the preceding claims.

[0039] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the low-voltage distribution network power outage event impact assessment method as described in any of the preceding claims.

[0040] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein, when the program instructions are executed by a computer, the computer performs the low-voltage distribution network outage event impact assessment method as described in any of the preceding claims.

[0041] As can be seen from the above technical solutions, the present invention has the following advantages:

[0042] This invention acquires multi-dimensional monitoring data from low-voltage distribution networks, preprocesses the data to obtain corresponding target monitoring data, constructs multiple feature subgroups based on a preset number of features, and performs hierarchical co-evolution of each feature subgroup according to a preset fitness function to obtain corresponding target feature data. Finally, a pre-trained power outage event impact assessment model is used to assess the impact of the target feature data, yielding corresponding impact assessment results. This overcomes the technical problem of traditional low-voltage distribution network power outage event impact assessment relying primarily on human experience, which is time-consuming and prone to misjudgment, thus reducing the reliability of low-voltage distribution network operation. Compared with traditional low-voltage distribution network power outage event impact assessment methods, this invention, by constructing multiple feature subgroups based on a preset number of features and performing hierarchical co-evolution of each feature subgroup according to a preset fitness function, quickly selects highly relevant target feature data. Finally, a pre-trained power outage event impact assessment model is used to assess the impact of the target feature data, obtaining corresponding impact assessment results. This improves the accuracy and efficiency of low-voltage distribution network power outage event impact assessment and enhances the reliability of low-voltage distribution network operation. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of the steps in a method for assessing the impact of a power outage event in a low-voltage distribution network, provided in Embodiment 1 of the present invention.

[0045] Figure 2 This is a flowchart of the steps in a method for assessing the impact of a power outage event in a low-voltage distribution network, provided in Embodiment 2 of the present invention.

[0046] Figure 3 This is a structural block diagram of a low-voltage distribution network power outage event impact assessment system provided in Embodiment 3 of the present invention;

[0047] Figure 4 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0048] This invention provides a method and system for assessing the impact of power outage events in low-voltage distribution networks. It addresses the technical problem that traditional assessments of the impact of power outage events in low-voltage distribution networks rely heavily on human experience, which is time-consuming, prone to misjudgments, and reduces the reliability of low-voltage distribution network operation.

[0049] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0050] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a method for assessing the impact of a power outage event in a low-voltage distribution network, as provided in Embodiment 1 of the present invention.

[0051] This invention provides a method for assessing the impact of power outage events in low-voltage distribution networks, comprising:

[0052] Step 101: Obtain multi-dimensional monitoring data of the low-voltage distribution network, preprocess the multi-dimensional monitoring data to obtain the corresponding target monitoring data.

[0053] Multi-dimensional monitoring data refers to the collection of raw data covering different dimensions of power grid operation, obtained through various monitoring devices in low-voltage distribution networks. Specifically, it includes voltage (an electrical parameter reflecting the power grid supply voltage level), current (an electrical parameter reflecting the magnitude of line current), power (including active power and reactive power, parameters reflecting power transmission and consumption), frequency change (reflecting the fluctuation value of the power grid frequency deviating from the rated frequency), power factor (a parameter reflecting the efficiency of power utilization), load data (real-time power data reflecting user electricity demand), fault current information (abnormal current signals generated when the power grid experiences faults such as short circuits), and equipment status data (parameters reflecting the operating status of key equipment such as distribution transformers and circuit breakers, such as temperature, operating time, and insulation performance indicators).

[0054] Target monitoring data refers to multi-dimensional monitoring data obtained after preprocessing steps such as data cleaning, outlier and noise removal, time-series alignment, standardization and normalization.

[0055] It should be noted that preprocessing refers to data cleaning, outlier and noise removal, time-series alignment, standardization and normalization of multi-dimensional monitoring data.

[0056] In this embodiment of the invention, multi-dimensional monitoring data of a low-voltage distribution network is acquired, and the multi-dimensional monitoring data is sequentially processed by data cleaning, outlier and noise removal, time alignment, standardization and normalization to obtain the corresponding target monitoring data.

[0057] Step 102: Based on the preset number of features, construct multiple feature subgroups according to the target monitoring data.

[0058] The number of features refers to the total number of features that need to be extracted from the target monitoring data in accordance with the actual needs of low-voltage distribution network outage monitoring and prediction, based on expert experience and historical data statistics.

[0059] A feature subgroup refers to an independent set of features formed by dividing a preset number of features according to the principle of "grouping similar attributes together" based on the correlation between the physical attributes and technical functions of each feature in the target monitoring data. The features in each subgroup have the same or highly related physical measurement dimensions (such as voltage, current, fault, etc.).

[0060] In this embodiment of the invention, the target monitoring data is classified into multiple feature sets according to the correlation between the physical attributes and technical functions of each feature in the target monitoring data. Based on the preset number of features, multiple feature subgroups are constructed according to each feature set (for example, when the number of features is 5, 5 features are randomly selected from the feature set multiple times as the feature combination corresponding to the individual, and feature subgroups are generated using each individual).

[0061] Step 103: Perform hierarchical collaborative evolution on each feature subgroup according to the preset fitness function to obtain the corresponding target feature data.

[0062] Target feature data refers to the core information related to power outage events that can accurately reflect the current operating status of the low-voltage distribution network.

[0063] In this embodiment of the invention, each feature combination is input into a preset fitness function to obtain multiple fitness values. Based on a preset update function, the feature subgroups are updated according to each fitness value to obtain multiple updated feature subgroups. It is determined whether the update count of each updated feature subgroup is greater than or equal to a preset iteration threshold. If the update count of the updated feature subgroup is less than the iteration threshold, the updated feature subgroup is determined as a new feature subgroup, and the process jumps to the step of inputting each feature combination into the preset fitness function to obtain multiple fitness values. If the update count of the updated feature subgroup is greater than or equal to the iteration threshold, the fitness value corresponding to the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination. Feature fusion is performed on each target feature combination based on a preset information entropy weighting function to obtain the corresponding target feature data.

[0064] Step 104: Use a pre-trained power outage event impact assessment model to assess the impact on the target feature data and obtain the corresponding impact assessment results.

[0065] Impact assessment results refer to the final output of a pre-trained power outage event impact assessment model after processing target feature data. Specifically, it includes three core information categories: probability of power outage occurrence (quantifying the possibility of power outage), scope of impact (clarifying the affected spatial range or user scale), and severity (classifying the level of impact caused by power outage). It serves as a direct reference for power grid operation and maintenance decisions.

[0066] In this embodiment of the invention, a pre-trained power outage event impact assessment model is used to assess the impact on target feature data, obtaining the corresponding impact assessment results. The power outage event impact assessment model includes a first network, a fully connected layer, and a second network.

[0067] In this embodiment of the invention, multi-dimensional monitoring data of a low-voltage distribution network is acquired, preprocessed to obtain corresponding target monitoring data, and multiple feature subgroups are constructed based on a preset number of features. Each feature subgroup undergoes hierarchical collaborative evolution according to a preset adaptation function to obtain corresponding target feature data. A pre-trained power outage event impact assessment model is then used to assess the impact of the target feature data, yielding the corresponding impact assessment results. This overcomes the technical problem that traditional low-voltage distribution network power outage event impact assessment relies heavily on human experience, which is time-consuming and prone to misjudgment, thus reducing the reliability of low-voltage distribution network operation. Compared with traditional methods for assessing the impact of power outages in low-voltage distribution networks, this invention constructs multiple feature subgroups based on target monitoring data using a preset number of features. It then performs hierarchical collaborative evolution of each feature subgroup according to a preset adaptation function, thereby quickly selecting highly relevant target feature data. Finally, a pre-trained power outage impact assessment model is used to assess the impact of the target feature data, yielding corresponding impact assessment results. This improves the accuracy and efficiency of assessing the impact of power outages in low-voltage distribution networks and enhances the reliability of low-voltage distribution network operation.

[0068] Please see Figure 2 , Figure 2 This is a flowchart illustrating the steps of a low-voltage distribution network power outage event impact assessment method provided in Embodiment 2 of the present invention.

[0069] This invention provides a method for assessing the impact of power outage events in low-voltage distribution networks, comprising:

[0070] Step 201: Obtain multi-dimensional monitoring data of the low-voltage distribution network, preprocess the multi-dimensional monitoring data to obtain the corresponding target monitoring data.

[0071] In this embodiment of the invention, multi-dimensional monitoring data of the low-voltage distribution network is acquired through various monitoring devices of the low-voltage distribution network (i.e., multi-dimensional monitoring data covering electrical parameters, load status, equipment operating status and fault signals are collected in real time through hardware facilities such as sensors, fault indicators and intelligent monitoring terminals installed on distribution boxes, branch line nodes, user-side smart meters and key power equipment (such as distribution transformers and circuit breakers) in the distribution substation area), and the multi-dimensional monitoring data is preprocessed to obtain the corresponding target monitoring data.

[0072] Step 202: Based on the preset number of features, construct multiple feature subgroups according to the target monitoring data. Each individual in the feature subgroup corresponds to a feature combination.

[0073] In this embodiment of the invention, the target monitoring data is classified into multiple feature sets according to the correlation between the physical attributes and technical functions of each feature in the target monitoring data. Based on a preset number of features, multiple feature subgroups are constructed according to each feature set. Each individual in a feature subgroup corresponds to a feature combination.

[0074] Step 203: Input each feature combination into the preset fitness function to obtain multiple fitness values.

[0075] In this embodiment of the invention, the fitness value of each feature combination is calculated using a preset fitness function.

[0076] It should be noted that the fitness function is specifically as follows:

[0077]

[0078] in, Let be the fitness value of the k-th individual in the l-th feature subgroup at the t-th iteration. For the k-th individual in the l-th feature subgroup, in the t-th iteration... The weights of each feature, For the l-th characteristic subgroup, the th One characteristic, This is the control coefficient for the regularization term. It is the first constant. Let l be the number of features in the l-th feature subgroup. For the l-th characteristic subgroup, the th Information entropy of each feature t is the index of the feature, l is the index of the iteration number, and k is the index of the individual.

[0079] Step 204: Based on the preset update function, update the feature subgroups according to each fitness value to obtain multiple updated feature subgroups.

[0080] Further, step 204 includes the following sub-steps:

[0081] S11. Select the minimum value among the fitness values ​​corresponding to each feature subgroup as the updated fitness value, and take the feature combination associated with each updated fitness value as the optimal feature combination.

[0082] The updated fitness value refers to the fitness value that is selected from all the individual fitness values ​​of a single feature subgroup.

[0083] In this embodiment of the invention, the minimum value among the fitness values ​​corresponding to each feature subgroup is selected as the updated fitness value, and the feature combination associated with each updated fitness value is selected as the optimal feature combination. For example, for a feature subgroup, the minimum value among the fitness values ​​corresponding to the feature subgroup is selected as the updated fitness value, and the feature associated with the updated fitness value is selected as the optimal feature combination.

[0084] S12. Based on the preset update function, update the individuals of each feature subgroup according to each optimal feature combination and the preset update factor data to obtain multiple updated feature subgroups.

[0085] It should be noted that the preset update factor data includes oscillation frequency (the period and frequency of the oscillation are adjusted using expert experience based on the complexity and convergence speed of different problems, with a reference value range of [0.1,1]) and random perturbation intensity adjustment factor, which are used to enhance the globality of the search and avoid individuals from getting trapped in local optima. They are determined according to expert experience, with a reference value range of (0,1).

[0086] In this embodiment of the invention, based on a preset update function, individuals in each feature subgroup are updated according to each optimal feature combination and preset update factor data, resulting in multiple updated feature subgroups. For example, for a feature subgroup, individuals in the feature subgroup are updated according to the preset update function using the optimal feature combination corresponding to the feature subgroup and preset update factor data, resulting in an updated feature subgroup.

[0087] It should be noted that the update function is as follows:

[0088]

[0089] in, For the k-th individual in the l-th feature subgroup in the (t+1)-th iteration, For the k-th individual in the l-th feature subgroup in the t-th iteration, For the collaboratively updated learning rate, The oscillation frequency is... To find the optimal feature combination in the l-th feature subgroup in the t-th iteration, This is a random disturbance intensity adjustment factor. For random disturbance terms, Let be the oscillation factor for the t-th iteration.

[0090] It is worth mentioning that the random perturbation term, used to increase the randomness in the individual search process, enables the individual to escape local optima in the search space and enhances the global search capability, and is generated from a uniform distribution or a normal distribution.

[0091] It is worth mentioning that the updated individuals can be adjusted by applying a local perturbation using a balance correction function, thereby obtaining the balanced corrected individuals.

[0092] The balance correction function is as follows:

[0093]

[0094] Among them, the left side of the equal sign For the balanced individuals, the right side of the equals sign For individuals before balance correction, This is a factor for adjusting the disturbance intensity. Let be the mean vector of all individuals in the l-th feature subgroup during the t-th iteration.

[0095] Step 205: Determine whether the number of updates for each updated feature subgroup is greater than or equal to the preset iteration threshold.

[0096] The iteration threshold refers to the maximum number of iterations preset by expert experience, taking into account the complexity of low-voltage distribution network monitoring data (such as feature dimensions and data fluctuation) and the model accuracy requirements. The value ranges from 50 to 200.

[0097] In this embodiment of the invention, it is determined whether the number of updates for each updated feature subgroup has reached a preset iteration threshold.

[0098] Step 206: If the number of updates of the updated feature subgroup is less than the iteration threshold, the updated feature subgroup is determined as a new feature subgroup, and the process jumps to the step of inputting each feature combination into the preset fitness function to obtain multiple fitness values.

[0099] In this embodiment of the invention, if the number of updates of the updated feature subgroup is less than the iteration threshold, it indicates that the updated feature subgroup has not been optimized. The updated feature subgroup is then determined as a new feature subgroup, and the process jumps to steps 203-206.

[0100] Step 207: If the number of updates of the updated feature subgroup is greater than or equal to the iteration threshold, then the fitness value corresponding to the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination.

[0101] In this embodiment of the invention, if the number of updates of the updated feature subgroup is greater than or equal to the iteration threshold, it indicates that the updated feature subgroup has been optimized. The fitness associated with the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination.

[0102] Step 208: Based on the preset information entropy weighting function, perform feature fusion on each target feature combination to obtain the corresponding target feature data.

[0103] Furthermore, step 208 includes the following sub-steps:

[0104] S21. Input each target feature combination into a preset information entropy weighting function to obtain multiple dynamic entropy weights.

[0105] In this embodiment of the invention, a preset information entropy weighting function is used to calculate the dynamic entropy weights corresponding to each combination of target features.

[0106] It should be noted that the information entropy weighting function is as follows:

[0107]

[0108] in, The dynamic entropy weight of the i-th normalized feature. Let be the relative weight ratio of the i-th normalized feature in sample j. For the total number of features, It is the second constant. The adjustment coefficient for penalizing the characteristic fluctuation amplitude is determined based on expert experience, with a reference value range of [0.5, 5]. For the i-th normalized feature, Let be the variance of the i-th normalized feature in the sample set. Let be the relative weight of the i-th normalized feature in sample j, where i is the first index of the feature, j is the first index of the sample, and m is the total number of samples.

[0109] S22. Based on the dynamic entropy weights, the various target feature combinations are weighted and fused to obtain the corresponding target feature data.

[0110] In this embodiment of the invention, the dynamic entropy weights and the target features are combined and input into a preset weighted fusion function to obtain the corresponding target feature data.

[0111] It should be noted that the weighted fusion function is as follows:

[0112]

[0113] in, For target feature data, For the l-th optimized feature, The dynamic entropy weights corresponding to the l-th feature subgroup are: The optimal feature combination is defined as the l-th feature subgroup in the T-th iteration, where T is the iteration threshold.

[0114] Step 209: Use a pre-trained power outage event impact assessment model to assess the impact on the target feature data and obtain the corresponding impact assessment results.

[0115] Furthermore, the power outage event impact assessment model includes a first network, a fully connected layer, and a second network. Step 209 includes the following sub-steps:

[0116] S31. Temporal features are extracted from the target feature data through the first network to obtain the corresponding temporal feature vector.

[0117] The first network refers to the Long Short-Term Memory (LSTM) network.

[0118] In this embodiment of the invention, the dynamic change trend of the target feature data is captured by a long short-term memory network to obtain the corresponding temporal feature vector.

[0119] It should be noted that when using Long Short-Term Memory (LSTM) networks to capture the dynamic trends of target feature data to obtain time-series feature vectors, the target feature data must first be segmented according to the time dimension of power grid operation. Based on the acquisition frequency of low-voltage distribution network monitoring data (e.g., 1 minute / time), the target feature data is divided into subsequences with fixed time steps (e.g., every 10 time steps is a subsequence). Each subsequence contains complete change information of the target feature data within that time period (e.g., the continuous values ​​of features such as voltage fluctuation amplitude and load change rate within 10 minutes), ensuring that the subsequence can fully reflect the short-term dynamic trend of power grid operation. These time step sequences are then input into a pre-constructed LSTM network structure, which includes an input layer, hidden layers (containing memory units, input gates, forget gates, and output gates), and a temporary output layer. The memory units can selectively discard historical irrelevant information (such as redundant feature data during stable operation periods) through the forget gate, receive key feature information of the current subsequence (such as feature changes during periods of sudden load increases) through the input gate, and output the feature state at the current moment through the output gate, thereby capturing long-term dependencies in the target feature data (such as the correlation between load increases in the first 30 minutes and subsequent voltage drops). During the forward propagation of the LSTM network, the hidden layer state at each time step is updated by combining the current subsequence features with the hidden layer state of the previous time step. Finally, after all time step subsequences have been processed, the temporary output layer of the LSTM network integrates the hidden layer states of each time step to generate a vector with fixed dimensions that can represent the dynamic change trend of the target feature data, which is the temporal feature vector. This vector not only contains the numerical change information of the target feature data, but also incorporates the correlation features in the time dimension, providing a temporal foundation for subsequent deep neural networks (DNNs) to mine nonlinear correlations between features.

[0120] S32. The temporal feature vector is nonlinearly mapped through a fully connected layer to obtain the corresponding target temporal feature vector.

[0121] In this embodiment of the invention, a fully connected layer is used to perform nonlinear mapping on the temporal feature vector to output the corresponding target temporal feature vector.

[0122] S33. The impact prediction of the target time series feature vector is performed through the second network to obtain the corresponding impact assessment results.

[0123] The second network refers to the deep neural network (DNN).

[0124] In this embodiment of the invention, the impact prediction of the target time-series feature vector is performed by a deep neural network to obtain the corresponding impact assessment result.

[0125] It should be noted that when using a deep neural network (DNN) to predict the impact of a target temporal feature vector to obtain an impact assessment result, the input format of the target temporal feature vector must first be clarified. This vector is the output of the Long Short-Term Memory (LSTM) network after capturing the dynamic change trend of the target feature data. It needs to be adjusted to a dimension that matches the number of neurons in the DNN input layer (e.g., mapping a 128-dimensional temporal feature vector to 64 dimensions to match the setting of 64 neurons in the DNN input layer) to ensure that the feature vector can be successfully input into the DNN for subsequent processing. The adapted target temporal feature vector is then input into a pre-constructed DNN structure, which contains multiple fully connected layers (e.g., 3 layers with 64, 32, and 16 neurons respectively) and one output layer. Each fully connected layer uses the ReLU activation function for nonlinear transformation, which can transform the temporal correlation information (such as load fluctuation cycle and voltage anomaly trend) contained in the target temporal feature vector into deeper nonlinear features (such as the coupling relationship between "load surge + voltage drop" and power outage risk). At the same time, the Dropout technique (with a dropout rate of 0.2) is used to reduce the model's dependence on local redundant features and avoid overfitting. During the forward propagation of the DNN, the target temporal feature vector undergoes progressive computation and feature refinement through multiple fully connected layers before being passed to the output layer. The output layer sets the number of neurons and activation functions according to the needs of the impact prediction task: for predicting the probability of power outage events (classification task), the output layer has 1 neuron and uses the sigmoid activation function (i.e., the S-type activation function) to map the result to the [0,1] interval; for predicting the scope of impact (number of affected users) and severity (power outage duration level) (regression task), the output layer has 2 neurons and uses the linear activation function (i.e., the linear activation function) to output continuous numerical results. These output results together constitute the corresponding impact assessment results, which can directly provide power grid operation and maintenance personnel with decision-making basis for power outage risk warning and emergency repair priority ranking.

[0126] It should be noted that the training process for the power outage impact assessment model is as follows:

[0127] A1. Obtain multiple historical target feature data and use each historical target feature data to generate a target feature set.

[0128] Historical target feature data refers to the target feature data obtained after processing historical multi-dimensional monitoring data through steps 201-208.

[0129] In this embodiment of the invention, multiple historical target feature data are acquired, and a target feature set is generated using each historical target feature data, wherein the target feature set includes multiple historical target feature data.

[0130] A2. The target feature set is used as input to train the preset initial power outage event impact assessment model to obtain the corresponding training impact assessment data.

[0131] Training impact assessment data refers to the prediction results of the target feature set output by the initial power outage event impact assessment model after training on the target feature set.

[0132] In this embodiment of the invention, the target feature set is input into a preset initial power outage event impact assessment model for training, and the corresponding training impact assessment data is obtained.

[0133] A3. Calculate the training loss function value of the target feature set based on the training impact assessment data.

[0134] In this embodiment of the invention, the training loss function value of the target feature set is calculated based on a preset loss function and the training impact assessment data. The loss function includes, but is not limited to, the cross-entropy loss function and the mean squared error loss function.

[0135] A4. When the training loss function value is greater than or equal to the preset standard loss value, the network parameters of the initial power outage event impact assessment model are adjusted, and the process jumps to the step of training the initial power outage event impact assessment model with the target feature set as input to obtain the corresponding training impact assessment data, until the training loss function value is less than the standard loss value.

[0136] In this embodiment of the invention, when the training loss function value is greater than or equal to the preset standard loss value, the network parameters of the initial power outage event impact assessment model are adjusted, and steps A2-A4 are executed until the training loss function value is less than the standard loss value.

[0137] A5. When the training loss function value is less than the standard loss value, a power outage event impact assessment model is generated.

[0138] In this embodiment of the invention, if the training loss function value is less than the standard loss value, a power outage event impact assessment model is generated.

[0139] In this embodiment of the invention, multi-dimensional monitoring data of a low-voltage distribution network is acquired, preprocessed to obtain corresponding target monitoring data, and multiple feature subgroups are constructed based on a preset number of features. Each feature subgroup undergoes hierarchical collaborative evolution according to a preset adaptation function to obtain corresponding target feature data. A pre-trained power outage event impact assessment model is then used to assess the impact of the target feature data, yielding the corresponding impact assessment results. This overcomes the technical problem that traditional low-voltage distribution network power outage event impact assessment relies heavily on human experience, which is time-consuming and prone to misjudgment, thus reducing the reliability of low-voltage distribution network operation. Compared with traditional methods for assessing the impact of power outages in low-voltage distribution networks, this invention constructs multiple feature subgroups based on target monitoring data using a preset number of features. It then performs hierarchical collaborative evolution of each feature subgroup according to a preset adaptation function, thereby quickly selecting highly relevant target feature data. Finally, a pre-trained power outage impact assessment model is used to assess the impact of the target feature data, yielding corresponding impact assessment results. This improves the accuracy and efficiency of assessing the impact of power outages in low-voltage distribution networks and enhances the reliability of low-voltage distribution network operation.

[0140] Please see Figure 3 , Figure 3 This is a structural block diagram of a low-voltage distribution network power outage event impact assessment system provided in Embodiment 3 of the present invention.

[0141] This invention provides a low-voltage distribution network power outage event impact assessment system, comprising:

[0142] The acquisition module 301 is used to acquire multi-dimensional monitoring data of the low-voltage distribution network, preprocess the multi-dimensional monitoring data, and obtain the corresponding target monitoring data.

[0143] Module 302 is used to construct multiple feature subgroups based on a preset number of features and target monitoring data;

[0144] The co-evolution module 303 is used to perform hierarchical co-evolution of each feature subgroup according to a preset adaptation function to obtain the corresponding target feature data;

[0145] The impact assessment module 304 is used to assess the impact of a pre-trained power outage event impact assessment model on the target feature data and obtain the corresponding impact assessment results.

[0146] Furthermore, the co-evolution module 303 includes:

[0147] The adaptation submodule is used to combine various features and input them into a preset adaptation function to obtain multiple adaptation values;

[0148] The update submodule is used to update the feature subgroups according to each fitness value based on a preset update function, so as to obtain multiple updated feature subgroups.

[0149] The analysis submodule is used to determine whether the number of updates for each updated feature subgroup is greater than or equal to a preset iteration threshold.

[0150] If the number of updates to the updated feature subgroup is less than the iteration threshold, the updated feature subgroup is determined as a new feature subgroup, and the process jumps to execute the step of inputting each feature combination into the preset fitness function to obtain multiple fitness values.

[0151] If the number of updates to the updated feature subgroup is greater than or equal to the iteration threshold, then the fitness value corresponding to the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination.

[0152] The feature fusion submodule is used to perform feature fusion on various target feature combinations based on a preset information entropy weighting function to obtain the corresponding target feature data.

[0153] Furthermore, update the submodules, including:

[0154] The selection unit is used to select the minimum value among the fitness values ​​corresponding to each feature subgroup as the updated fitness value, and to take the feature combination associated with each updated fitness value as the optimal feature combination.

[0155] The update unit is used to update the individuals of each feature subgroup based on a preset update function, according to each optimal feature combination and preset update factor data, to obtain multiple updated feature subgroups.

[0156] Furthermore, the feature fusion submodule includes:

[0157] The information entropy weighting unit is used to combine each target feature and input it into a preset information entropy weighting function to obtain multiple dynamic entropy weights.

[0158] The weighted fusion unit is used to perform weighted fusion of various target feature combinations based on each dynamic entropy weight to obtain the corresponding target feature data.

[0159] Furthermore, the power outage event impact assessment model includes a first network, a fully connected layer, and a second network. The impact assessment module 304 includes:

[0160] The temporal feature extraction unit is used to extract temporal features from the target feature data through the first network to obtain the corresponding temporal feature vector.

[0161] The mapping unit is used to perform nonlinear mapping on the temporal feature vector through the fully connected layer to obtain the corresponding target temporal feature vector;

[0162] The impact prediction unit is used to predict the impact of the target time-series feature vector through the second network to obtain the corresponding impact assessment results.

[0163] Furthermore, the training process for the power outage impact assessment model is as follows:

[0164] Acquire multiple historical target feature data, and use each historical target feature data to generate a target feature set;

[0165] The target feature set is used as input to train the initial power outage event impact assessment model, and the corresponding training impact assessment data is obtained.

[0166] Calculate the training loss function value of the target feature set based on the training impact assessment data;

[0167] When the training loss function value is greater than or equal to the preset standard loss value, the network parameters of the initial power outage event impact assessment model are adjusted, and the process jumps to the step of training the initial power outage event impact assessment model with the target feature set as input to obtain the corresponding training impact assessment data, until the training loss function value is less than the standard loss value.

[0168] When the training loss function value is less than the standard loss value, a power outage event impact assessment model is generated.

[0169] Please see Figure 4 , Figure 4 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention.

[0170] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402. The memory 401 stores a computer program. When the computer program is executed by the processor 402, the processor 402 executes the low-voltage distribution network power outage event impact assessment method as described in any of the above embodiments.

[0171] Memory 401 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 401 has storage space 403 for program code 413 for performing any of the method steps described above. For example, storage space 403 for program code may include individual program codes 413 for implementing the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When these codes are run by a computing processing device, the device performs the various steps in the low-voltage distribution network outage event impact assessment method described above.

[0172] Embodiment 5 of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the low-voltage distribution network power outage event impact assessment method as described in any of the above embodiments.

[0173] Embodiment 6 of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the low-voltage distribution network outage event impact assessment method as described in any of the above embodiments.

[0174] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0175] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0176] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0177] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0178] If the integrated unit is implemented as 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 technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0179] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assessing the impact of power outage events in low-voltage distribution networks, characterized in that, include: Acquire multi-dimensional monitoring data of the low-voltage distribution network, preprocess the multi-dimensional monitoring data, and obtain the corresponding target monitoring data; Based on a preset number of features, multiple feature subgroups are constructed according to the target monitoring data; Based on a preset adaptation function, each of the feature subgroups is subjected to hierarchical collaborative evolution to obtain the corresponding target feature data; The target feature data is assessed using a pre-trained power outage event impact assessment model to obtain the corresponding impact assessment results.

2. The method for assessing the impact of low-voltage distribution network outage events according to claim 1, characterized in that, Each individual in the feature subgroup corresponds to a feature combination. The step of performing hierarchical collaborative evolution on each feature subgroup according to a preset adaptation function to obtain the corresponding target feature data includes: Each of the aforementioned features is combined and input into a preset adaptation function to obtain multiple adaptation values; Based on a preset update function, the feature subgroups are updated according to each of the fitness values ​​to obtain multiple updated feature subgroups; Determine whether the number of updates for each of the updated feature subgroups is greater than or equal to a preset iteration threshold; If the number of updates of the updated feature subgroup is less than the iteration threshold, the updated feature subgroup is determined as a new feature subgroup, and the process jumps to the step of inputting each of the feature combinations into a preset fitness function to obtain multiple fitness values. If the number of updates of the updated feature subgroup is greater than or equal to the iteration threshold, then the fitness value corresponding to the updated feature subgroup is determined as the target fitness value, and the feature combination corresponding to the minimum value among the target fitness values ​​is selected as the target feature combination. Based on a preset information entropy weighting function, feature fusion is performed on each of the target feature combinations to obtain the corresponding target feature data.

3. The method for assessing the impact of low-voltage distribution network outage events according to claim 2, characterized in that, The step of updating the feature subgroups according to each fitness value based on a preset update function to obtain multiple updated feature subgroups includes: The minimum value among the fitness values ​​corresponding to each feature subgroup is selected as the updated fitness value, and the feature combination associated with each updated fitness value is taken as the optimal feature combination. Based on a preset update function, individuals in each feature subgroup are updated according to each optimal feature combination and preset update factor data, resulting in multiple updated feature subgroups.

4. The method for assessing the impact of low-voltage distribution network outage events according to claim 2, characterized in that, The step of fusing features of each target feature combination based on a preset information entropy weighting function to obtain the corresponding target feature data includes: Each of the target features is combined and input into a preset information entropy weighting function to obtain multiple dynamic entropy weights; Based on the dynamic entropy weights, the target feature combinations are weighted and fused to obtain the corresponding target feature data.

5. The method for assessing the impact of low-voltage distribution network outage events according to claim 1, characterized in that, The power outage event impact assessment model includes a first network, a fully connected layer, and a second network. The step of using the pre-trained power outage event impact assessment model to assess the impact on the target feature data and obtain the corresponding impact assessment results includes: Temporal feature extraction is performed on the target feature data through the first network to obtain the corresponding temporal feature vector; The time-series feature vector is nonlinearly mapped through the fully connected layer to obtain the corresponding target time-series feature vector; The second network is used to predict the impact of the target time-series feature vector, and the corresponding impact assessment result is obtained.

6. The method for assessing the impact of low-voltage distribution network outage events according to claim 1, characterized in that, The specific training process for the power outage event impact assessment model is as follows: Acquire multiple historical target feature data, and generate a target feature set using each of the historical target feature data; The target feature set is used as input to train the preset initial power outage event impact assessment model to obtain the corresponding training impact assessment data; Calculate the training loss function value of the target feature set based on the training impact assessment data; When the training loss function value is greater than or equal to the preset standard loss value, the network parameters of the initial power outage event impact assessment model are adjusted, and the process jumps to the step of training the preset initial power outage event impact assessment model using the target feature set as input to obtain the corresponding training impact assessment data, until the training loss function value is less than the standard loss value. When the training loss function value is less than the standard loss value, a power outage event impact assessment model is generated.

7. A low-voltage distribution network outage event impact assessment system, characterized in that, include: The acquisition module is used to acquire multi-dimensional monitoring data of the low-voltage distribution network, and to preprocess the multi-dimensional monitoring data to obtain the corresponding target monitoring data. The construction module is used to construct multiple feature subgroups based on the target monitoring data, according to a preset number of features. The co-evolution module is used to perform hierarchical co-evolution of each feature subgroup according to a preset adaptation function to obtain the corresponding target feature data; The impact assessment module is used to assess the impact of the target feature data using a pre-trained power outage event impact assessment model, and obtain the corresponding impact assessment results.

8. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the low-voltage distribution network power outage event impact assessment method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the low-voltage distribution network power outage event impact assessment method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the low-voltage distribution network outage event impact assessment method as described in any one of claims 1-6.

Images