Risk level prediction method and device for power distribution network, and electronic equipment

Abstract

The invention discloses a risk level prediction method and device for a power distribution network, and electronic equipment, and relates to the technical field of power distribution network operation and maintenance or other related fields, the method comprises the following steps: obtaining current operation data of a target power distribution network, the current operation data being acquired by a sensor network in real time, comprising line equipment operation data of the target power distribution network, geographical environment data of the position where the target power distribution network is located and disaster weather data; all the current operation data are input to a target model, the risk prediction level and early warning information are output, the target model is a decision tree model pre-established based on a random forest algorithm and is used for extracting key features from the line equipment operation data, the geographical environment data and the disaster weather data, and therefore risk level study and judgment are carried out; and early warning information is generated according to different risk prediction levels. According to the invention, the technical problem of low power distribution network risk prediction accuracy in extreme weather in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network operation and maintenance, and in particular to a risk level prediction method and device, and electronic equipment for distribution networks. Background Art

[0002] The distribution network is an important part of the power system, including power generation, transmission, transformation, distribution and consumption, covering the entire business chain of power generation, transmission, transformation, distribution and consumption. The frequent occurrence of failures or risks and the high probability of clustered failures have a huge impact on the safe and stable operation of the power grid. Distribution network risk prediction refers to the use of historical data for analysis to predict the probability of failures or risks in the distribution network in the future and to provide early warning of failures or risks.

[0003] With the frequent occurrence of extreme weather, the distribution network infrastructure has been greatly damaged. Its operational safety and stability directly affect the overall performance of the power system. The traditional distribution network risk prediction method only uses traditional risk assessment methods and statistical analysis methods. It is unable to combine specific extreme weather conditions to fully reflect the large number of complex and nonlinear relationships in the distribution network, resulting in low accuracy and low efficiency in safety assessment and analysis results. In general, the risk prediction level of the distribution network under extreme weather conditions is low, and it is impossible to make differentiated emergency preparations for the distribution network according to local conditions. The level of disaster prevention and mitigation of the distribution network is low, which can easily cause huge power outage losses.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method and device for predicting the risk level of a distribution network, and an electronic device, so as to at least solve the technical problem of low accuracy in risk prediction of a distribution network under extreme weather conditions in the related art.

[0006] According to one aspect of an embodiment of the present invention, there is provided a risk level prediction method for a distribution network, comprising: obtaining current operating data of a target distribution network, wherein the current operating data is collected in real time by a sensor network, and comprises: line equipment operating data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data; inputting all the current operating data into a target model, and outputting risk prediction level and warning information, wherein the target model is a decision tree model pre-established based on a random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data, and the disaster weather data, conduct risk level analysis based on the key features, obtain risk prediction levels, and generate corresponding warning information based on different risk prediction levels.

[0007] Furthermore, the line equipment operation data includes: historical disaster loss data and equipment operation years, wherein the historical disaster loss data is used to characterize the damage suffered by the target line equipment in extreme disasters in previous years, and the equipment operation years is used to characterize the failure rate of the target line equipment.

[0008] Furthermore, the geographical environment data of the location of the target distribution network includes: slope data, soil data and ground roughness, and the disaster weather data includes: maximum wind speed level, average wind speed level, maximum wind speed direction, average wind direction and rainfall level within a preset monitoring time period, wherein the rainfall level refers to the depth level of the water layer falling to the ground within the preset monitoring time period.

[0009] Furthermore, before all the current operating data are input into the target model, it also includes: preprocessing all the current operating data, wherein the preprocessing operations include: data cleaning, data outlier processing and data standardization processing, and the data standardization processing refers to mapping all data to a unified dimensional range according to a preset normalization formula; pushing the preprocessed current operating data in full to a pre-built distribution equipment state perception model, and the model outputs the state perception result of the target distribution network; when the state perception result is obtained, inputting all the current operating data into the target model, and the step of outputting the risk prediction level and the warning information also includes: directly inputting the state perception result into the target model, and outputting the risk prediction level and the warning information.

[0010] Furthermore, the target model is obtained through the following steps: obtaining risk-related data of the target distribution network in a historical time period and constructing a risk data sample, wherein the risk-related data includes: equipment inventory data, operation hazard data, and external environment impact weight data; determining the risk-related characteristics of the target distribution network based on the risk data sample, and constructing a decision tree based on the risk-related characteristics to obtain N decision trees, wherein N is a positive integer; integrating the N decision trees based on the random forest algorithm, and assigning a unique decision vote to each decision tree to obtain the target model.

[0011] Furthermore, the risk-related characteristics of the target distribution network are determined based on the risk data sample, and a decision tree is constructed based on the risk-related characteristics to obtain the step of N decision trees, including: generating a model training data set and a model test data set based on the risk data sample, and extracting N training subsets from the model training data set using a uniform sampling strategy; for each of the training subsets, determining the data outlier rate and the total amount of risk-related characteristics, and calculating the number of features to be extracted R based on the data outlier rate and the total amount of risk-related characteristics, where R is a positive integer; extracting R risk-related characteristics from the training subset, calculating the Gini index of each risk-related characteristic for the training subset, and allocating the training data in the training subset to each decision node of the initial decision tree according to the R Gini indices and a preset node allocation strategy to obtain the decision tree; integrating the decision trees corresponding to the N training subsets to obtain N decision trees.

[0012] Furthermore, the step of assigning a unique decision vote to each decision tree includes: for each decision tree, inputting the model test data set S times into the decision tree, outputting S decision results, and comparing the S decision results with the expected results in the model test data set, and calculating the decision accuracy of the decision tree, wherein S is a specified value; obtaining the decision accuracy corresponding to each decision tree, and determining the decision vote of each decision tree based on the decision accuracy.

[0013] According to another aspect of an embodiment of the present invention, there is also provided a risk level prediction method and device for a distribution network, comprising: an acquisition unit, for acquiring current operating data of a target distribution network, wherein the current operating data is collected in real time by a sensor network, and comprises: line equipment operating data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data; an input unit, for inputting all the current operating data into a target model, and outputting risk prediction levels and warning information, wherein the target model is a decision tree model pre-established based on a random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data, and the disaster weather data, perform risk level analysis based on the key features, obtain risk prediction levels, and generate corresponding warning information based on different risk prediction levels.

[0014] Furthermore, the line equipment operation data includes: historical disaster loss data and equipment operation years, wherein the historical disaster loss data is used to characterize the damage suffered by the target line equipment in extreme disasters in previous years, and the equipment operation years is used to characterize the failure rate of the target line equipment.

[0015] Furthermore, the geographical environment data of the location of the target distribution network includes: slope data, soil data and ground roughness, and the disaster weather data includes: maximum wind speed level, average wind speed level, maximum wind speed direction, average wind direction and rainfall level within a preset monitoring time period, wherein the rainfall level refers to the depth level of the water layer falling to the ground within the preset monitoring time period.

[0016] Furthermore, the risk level prediction method device for the distribution network also includes: a preprocessing module, used to preprocess all the current operating data, wherein the preprocessing operations include: data cleaning, data outlier processing and data standardization processing, and the data standardization processing refers to mapping all data to a unified dimension range according to a preset normalization formula; a full push module, used to push the preprocessed current operating data in full to a pre-built distribution equipment state perception model, and the model outputs the state perception result of the target distribution network; an input module, used to directly input the state perception result into the target model when the state perception result is obtained, and output the risk prediction level and the warning information.

[0017] Furthermore, the risk level prediction method device for distribution networks also includes: an acquisition module, used to obtain risk-related data of the target distribution network in a historical time period and construct a risk data sample, wherein the risk-related data includes: equipment inventory data, operation hazard data and external environment impact weight data; a determination module, used to determine the risk-related characteristics of the target distribution network based on the risk data sample, and construct a decision tree based on the risk-related characteristics to obtain N decision trees, wherein N is a positive integer; an integration module, used to integrate the N decision trees based on the random forest algorithm, and assign a unique decision vote to each decision tree to obtain the target model.

[0018] Furthermore, the determination module includes: a generation submodule, which is used to generate a model training data set and a model test data set based on the risk data sample, and extract N training subsets from the model training data set using a uniform sampling strategy; a first calculation submodule, which is used to determine the data outlier rate and the total amount of risk-related features for each of the training subsets, and calculate the number of features to be extracted R based on the data outlier rate and the total amount of risk-related features, where R is a positive integer; an allocation submodule, which is used to extract R risk-related features from the training subset, calculate the Gini index of each risk-related feature for the training subset, and allocate the training data in the training subset to each decision node of the initial decision tree based on the R Gini indices and a preset node allocation strategy to obtain the decision tree; an integration submodule, which is used to integrate the decision trees corresponding to the N training subsets to obtain N decision trees.

[0019] Furthermore, the integration module includes: a second calculation submodule, which is used to input the model test data set S times into each decision tree, output S decision results, and compare the S decision results with the expected results in the model test data set to calculate the decision accuracy of the decision tree, where S is a specified value; a determination submodule, which is used to obtain the decision accuracy corresponding to each decision tree, and determine the decision voting rights of each decision tree based on the decision accuracy.

[0020] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned risk level prediction methods for distribution network.

[0021] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned risk level prediction methods for distribution networks.

[0022] In the present invention, a risk level prediction method for a distribution network is proposed, which first obtains the current operating data of the target distribution network, wherein the current operating data is collected in real time by a sensor network, including: line equipment operating data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data, and then all the current operating data are input into a target model, and the risk prediction level and warning information are output, wherein the target model is a decision tree model pre-established based on a random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data, and the disaster weather data, and the risk level is judged based on the key features to obtain the risk prediction level, and the corresponding warning information is generated according to different risk prediction levels.

[0023] In the present invention, by collecting the line equipment operation data, geographical environment data and disaster weather data of the target distribution network in real time, the data input into the model is ensured to be timely and comprehensive, so as to capture the immediate impact of extreme weather on the distribution network. The target model trained by the random forest algorithm can effectively process high-dimensional data and automatically select the most important features for risk prediction, avoiding the reliance on feature selection in the existing technology methods, thereby improving the adaptability and accuracy of the model under extreme weather conditions. In addition, the decision tree analysis based on key features can reveal the complex relationship between extreme weather and distribution network risks. The model can reflect the potential risk changes in real time and generate corresponding early warning information by comprehensively analyzing multidimensional data, providing scientific decision-making basis for operation and maintenance personnel, achieving the technical effect of improving the risk prediction ability of the distribution network under extreme weather, and thus solving the technical problem of low accuracy of distribution network risk prediction under extreme weather in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0025] Figure 1 is a flow chart of an optional risk level prediction method for a distribution network according to an embodiment of the present invention;

[0026] Figure 2 is a flow chart of an optional decision tree generation method according to an embodiment of the present invention;

[0027] Figure 3 is a flow chart of an optional decision tree voting rights determination method according to an embodiment of the present invention;

[0028] Figure 4 is a schematic diagram of an optional risk level prediction method and device for a distribution network according to an embodiment of the present invention;

[0029] Figure 5 The present invention is a hardware structure block diagram of an electronic device (or mobile device) for a method for predicting risk levels of a distribution network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] The following embodiments of the present invention can be applied to various systems / applications / equipment that need to monitor and warn of distribution network risks under extreme weather conditions, and can implement operation risk detection and warning of distribution equipment based on random forest algorithms and decision tree models. The present invention collects distribution network operation data, geographic environment data, and disaster weather data in real time to capture the immediate impact of extreme weather on the distribution network, uses random forest algorithms to train target models to process high-dimensional data and automatically select high-risk related features, and then uses decision trees to reveal the complex relationship between extreme weather and distribution network risks, reflecting potential risk changes in real time and generating corresponding warning information.

[0033] The present invention is described in detail below in conjunction with various embodiments.

[0034] Embodiment 1

[0035] According to an embodiment of the present invention, an embodiment of a risk level prediction method for a distribution network is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0036] The implementing body of the embodiment of the present invention may be a distribution network monitoring system, which is used to implement distribution network risk prediction under extreme weather conditions. By processing and screening data on disaster-causing variables such as disastrous weather (heavy rain), distribution lines and equipment, and geographical locations, accurate prediction of distribution network risks under extreme weather conditions can be achieved, providing strong guarantees for the safe operation of the distribution network.

[0037] Specifically, through the analysis of the risk factors of the distribution network, heavy rainfall will accelerate the sliding of the soil around the equipment to low-lying areas or cause cracks in the mountain, leading to uneven settlement of the foundation. Equipment and its foundations in ditches, across rivers and on landslides are easily affected by erosion, long-term immersion and overall translation of heavy rainfall, causing equipment to topple over and stop operating. Substations located in low-lying areas or urban waterlogged areas are vulnerable to heavy rainfall, floods and mudslides.

[0038] The embodiments of the present invention are described in detail below in conjunction with various method steps.

[0039] Figure 1 is a flow chart of an optional method for predicting risk level of a distribution network according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0040] Step S101, obtaining current operating data of the target distribution network, wherein the current operating data is collected in real time by a sensor network, including: line equipment operating data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data.

[0041] Specifically, the sensors in the sensor network can include voltage sensors, current sensors, temperature sensors, and humidity sensors, etc., which are used to monitor various operating parameters of the equipment. The collected data is sent to the data processing center through the data transmission network to provide raw data for subsequent data preprocessing and risk assessment.

[0042] It should be noted that, considering the disaster entity factors and disaster-bearing carrier factors, the disaster recovery variables of the target distribution network can be divided into the following three dimensions: distribution line and equipment factors, geographical environment factors and disaster weather factors. Correspondingly, predicting the risk level under extreme weather requires collecting real-time data in these three dimensions: line equipment operation data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data.

[0043] Optionally, the line equipment operation data includes: historical disaster loss data and equipment operation years, wherein the historical disaster loss data is used to characterize the damage suffered by the target line equipment in extreme disasters in previous years, and the equipment operation years is used to characterize the failure rate of the target line equipment.

[0044] It should be noted that the distribution lines and equipment are mainly considered from the historical disaster experience. The equipment that has been hit by the disaster is more likely to be hit by the disaster again than the equipment that has not been hit by the disaster due to the different environment and equipment characteristics. The historical disaster loss variables are used to mark the damage of the equipment in typhoons and rainstorms in previous years. The damage is recorded as 1 if there is a disaster and 0 if there is no disaster loss, thus obtaining the historical disaster loss data.

[0045] In addition to external factors, the failure rate of the equipment itself is also an important factor in determining its ability to resist disasters. It is determined by characteristics such as equipment design parameters, aging and corrosion, and whether it has been windproof reinforced. The failure rate of most equipment conforms to a time function (i.e., the bathtub curve). In order to reduce the complexity of the model and increase the prediction speed, the equipment's operating life can be used to characterize the failure rate of the target line equipment.

[0046] Optionally, the geographical environment data of the location of the target distribution network includes: slope data, soil data and ground roughness, and the disaster weather data includes: maximum wind speed level, average wind speed level, maximum wind speed direction, average wind direction and rainfall level within a preset monitoring time period, wherein the rainfall level refers to the depth level of the water layer falling to the ground within the preset monitoring time period.

[0047] It should be noted that the disaster reliability of distribution lines and equipment is different under different geographical conditions. For example, there will be certain differences in wind speed in different terrains in the same area. The narrow tube effect of the terrain will cause the wind speed to increase in a short period of time. The differences in the density of surface buildings and vegetation will affect the movement and disappearance of typhoons, and thus affect the actual wind speed that the equipment can withstand. Therefore, the geographical environment factors mainly consider the slope, soil quality and ground roughness of the distribution lines and equipment.

[0048] Taking into account that strong winds and heavy rainfall are the main factors causing damage to distribution lines and equipment during typhoons and rainstorms, the disaster weather factors include the maximum wind speed level, average wind speed level, maximum wind speed and direction, average wind direction and rainfall level within the specified monitoring period. The rainfall level refers to the depth of the water layer that falls to the ground within a certain period of time. The levels of wind speed and rainfall are divided according to unified industry standards.

[0049] Step S102, input all current operating data into the target model, output risk prediction level and warning information, wherein the target model is a decision tree model pre-established based on the random forest algorithm, and the decision tree model is used to extract key features from line equipment operating data, geographic environment data and disaster weather data, conduct risk level analysis based on the key features, obtain risk prediction level, and generate corresponding warning information based on different risk prediction levels.

[0050] Optionally, before executing step S102, it also includes: preprocessing all current operating data, wherein the preprocessing operations include: data cleaning, data outlier processing and data standardization processing, and data standardization processing refers to mapping all data to a unified dimensional range according to a preset normalization formula; pushing the full amount of preprocessed current operating data to a pre-built distribution equipment state perception model, and the model outputs the state perception results of the target distribution network.

[0051] Specifically, data cleaning is the first step in data preprocessing, which aims to identify and correct errors, inconsistencies and missing values ​​in the data to improve data quality, including: deleting duplicate records, filling missing values, correcting erroneous data (such as format errors, spelling errors, etc.), and ensuring that the data format is consistent.

[0052] Data outlier processing focuses on identifying and processing data points that significantly deviate from the normal range. These outliers may be caused by data collection errors, sensor failures, or real extreme situations. Methods for processing outliers include deleting outliers, replacing them with more appropriate values ​​(such as mean or median), or using model methods (such as isolation forest) for more sophisticated processing.

[0053] Data standardization is the process of converting data of different dimensions and ranges to the same scale for comparison and analysis. Common methods include Z-score standardization and Min-Max standardization. The former converts the data to a form with a mean of 0 and a standard deviation of 1, while the latter scales the data to a specified interval (usually 0 to 1). Standardization allows different features to have equal weights in model training, avoiding disproportionate effects of certain features on the results. A dimension conversion formula is provided in the embodiment of the present invention: x i ′ It refers to the standardized value of the i-th data in a certain type of data field, X i refers to the true value of the i-th data, |X| max Refers to the maximum absolute value of all data in this type of data field.

[0054] In addition, the distribution equipment state perception model is a pre-built data format compilation model. By inputting a single unordered operating data into the model, a data structure compiled according to the standard data format can be obtained, that is, the state perception result, in which the specific content recorded is still the original input data.

[0055] Furthermore, when the state perception result is obtained, the execution method of step S102 also includes: directly inputting the state perception result into the target model, and outputting the risk prediction level and warning information.

[0056] Optionally, the target model is obtained through the following steps: obtaining risk-related data of the target distribution network in a historical time period, and constructing a risk data sample, wherein the risk-related data includes: equipment inventory data, operation hazard data, and external environment impact weight data; determining the risk-related characteristics of the target distribution network based on the risk data sample, and constructing a decision tree based on the risk-related characteristics to obtain N decision trees, wherein N is a positive integer; integrating the N decision trees based on the random forest algorithm, and assigning a unique decision vote to each decision tree to obtain the target model.

[0057] It should be noted that the equipment inventory data contains basic information about the equipment, such as model, installation date, maintenance records, etc. The operation hazard data is used to record the equipment's failure history, maintenance records and inspection reports. The external environment impact weight data takes into account the impact of external factors (such as weather and load fluctuations) on the equipment.

[0058] When constructing a risk assessment model for a target distribution network, the embodiment of the present invention first obtains and integrates risk-related data, including equipment records, operating hazards, and external environmental impact weight data, to provide a data basis for subsequent analysis, and then identifies features closely related to equipment failures by analyzing historical data samples. In order to establish a decision tree, a uniform sampling strategy is used to generate multiple subsets from a training data set (for example, Bootstrap sampling is used, and n samples are extracted from each training subset with replacement, generating a total of m training subsets), and the outlier rate λ and the total amount k of risk-related features of the data are calculated for each subset, thereby determining the number of features to be extracted L (the exemplary calculation formula provided is: ).

[0059] Then, L features are extracted from each training subset, and the Gini index of each feature on the training subset is calculated to evaluate its influence on the decision node. Through the splitting process of these features, N decision trees can be constructed to form a preliminary model of random forest.

[0060] Optionally, the steps of determining risk-related characteristics of the target distribution network based on the risk data sample, and constructing a decision tree based on the risk-related characteristics to obtain N decision trees include: generating a model training data set and a model test data set based on the risk data sample, and extracting N training subsets from the model training data set using a uniform sampling strategy; for each training subset, determining the data outlier rate and the total amount of risk-related characteristics, and calculating the number of features to be extracted R based on the data outlier rate and the total amount of risk-related characteristics, where R is a positive integer; extracting R risk-related characteristics from the training subset, calculating the Gini index of each risk-related characteristic for the training subset, and allocating the training data in the training subset to each decision node of the initial decision tree according to the R Gini indices and a preset node allocation strategy to obtain a decision tree; and integrating the decision trees corresponding to the N training subsets to obtain N decision trees.

[0061] In constructing the risk assessment model of the distribution network, the embodiment of the present invention first generates a model training data set and a model test data set based on the risk data sample, ensuring the diversity and representativeness of the data and providing a basis for the learning and verification of the model. Then, a uniform sampling strategy is adopted to extract N training subsets from the training data set, which can avoid the model inaccuracy caused by sample selection bias. For each training subset, it is key to determine the outlier rate of the data and the total amount of risk-related features. The former helps to identify abnormal data points, and the latter provides a basis for feature selection.

[0062] Calculating the number of features to be extracted R based on the outlier rate and the total number of features can make feature selection more flexible and targeted. Extracting R risk-related features from the training subset and calculating their Gini index can effectively measure the contribution of each feature to the classification accuracy, thereby optimizing feature selection. According to the R Gini indices and the preset node allocation strategy, the training data are allocated to different nodes of the initial decision tree, which can ensure that the construction of the tree makes full use of important features.

[0063] Finally, integrating the decision trees corresponding to N training subsets not only improves the robustness of the model, but also enhances the comprehensive identification of risk factors. The technical effect of this process is that the model can more accurately predict potential risks and provide more scientific data support, thereby helping operation and maintenance personnel make more effective decisions and reduce safety hazards in the distribution network.

[0064] Optionally, the step of assigning a unique decision vote to each decision tree includes: for each decision tree, inputting a model test data set into the decision tree S times, outputting S decision results, and comparing the S decision results with expected results in the model test data set, and calculating the decision accuracy of the decision tree, where S is a specified value; obtaining the decision accuracy corresponding to each decision tree, and determining the decision vote of each decision tree based on the decision accuracy.

[0065] When assigning a unique decision vote to each decision tree, the embodiment of the present invention first needs to input the model test data set into each decision tree for evaluation. The specific steps are: for each decision tree, input the test data set S times, where S is a pre-specified value, and a larger value is usually selected to ensure the stability and reliability of the result. Each input can generate a decision result, so S decision results will be obtained in the end.

[0066] Next, compare these S decision results with the expected results in the model test data set. The key to this step is to evaluate the accuracy of the decision tree. By calculating the ratio of the number of correct predictions to the total number of predictions, the decision accuracy of the decision tree can be obtained, which is used to quantify the performance of each decision tree and facilitate subsequent decision-making.

[0067] After obtaining the decision accuracy of each decision tree, the decision votes of each decision tree can be determined based on these values. Decision votes can be regarded as the influence of each decision tree in the final decision, which is usually proportional to the decision accuracy. Decision trees with good performance will receive higher weights, while decision trees with poor performance will receive lower weights. This distribution method can effectively improve the accuracy of the overall model.

[0068] It is important to note that the process of multiple inputs and result comparisons of the test data set can reduce the impact of chance on model evaluation and ensure the credibility of the evaluation results. In addition, through the allocation of voting rights, the model can adaptively adjust the influence of each decision tree when making integrated decisions, thereby improving the robustness of predictions in complex situations.

[0069] Through the above steps, the embodiment of the present invention first enhances the overall performance of the integrated model, making the final prediction result more reliable; secondly, the model can better identify efficient decision trees and improve the sensitivity to risk characteristics by assigning corresponding weights; finally, through the allocation of decision-making votes, the model's adaptability to different scenarios is improved, making it more flexible and effective in practical applications.

[0070] Through the above steps S101 to S102, the current operating data of the target distribution network can be obtained first, wherein the current operating data is collected in real time by the sensor network, including: the line equipment operating data of the target distribution network, the geographical environment data of the location of the target distribution network, and the disaster weather data, and then all the current operating data are input into the target model, and the risk prediction level and warning information are output, wherein the target model is a decision tree model pre-established based on the random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data, and the disaster weather data, and the risk level is judged based on the key features to obtain the risk prediction level, and the corresponding warning information is generated according to different risk prediction levels.

[0071] In the embodiment of the present invention, by collecting the line equipment operation data, geographical environment data and disaster weather data of the target distribution network in real time, the data input into the model is ensured to be timely and comprehensive, so as to capture the immediate impact of extreme weather on the distribution network. The target model trained by the random forest algorithm can effectively process high-dimensional data and automatically select the most important features for risk prediction, avoiding the reliance on feature selection in the prior art methods, thereby improving the adaptability and accuracy of the model under extreme weather conditions. In addition, the decision tree analysis based on key features can reveal the complex relationship between extreme weather and distribution network risks. The model can reflect the potential risk changes in real time and generate corresponding early warning information by comprehensively analyzing multidimensional data, providing scientific decision-making basis for operation and maintenance personnel, achieving the technical effect of improving the risk prediction ability of the distribution network under extreme weather, and thus solving the technical problem of low accuracy of risk prediction of the distribution network under extreme weather in the related technology.

[0072] The present invention is described below in conjunction with another specific implementation manner.

[0073] The embodiment of the present invention aims to provide a method for predicting distribution network risks under extreme weather conditions. By processing and screening data on disaster-causing variables such as disastrous weather (rainstorms), distribution lines and equipment, and geographical locations, accurate prediction of distribution network risks under extreme weather conditions can be achieved, thereby providing strong guarantees for the safe operation of the distribution network.

[0074] Through the analysis of the risk factors of the distribution network, it is known that heavy rainfall will accelerate the sliding of the soil around the equipment to low-lying areas or cause cracks in the mountain, resulting in uneven settlement of the foundation. The equipment and its foundations in ditches, across rivers and on landslides are easily affected by the erosion, long-term immersion and overall translation of heavy rainfall, which may cause the equipment to topple over and shut down. Substations located in low-lying areas or urban waterlogged areas are susceptible to the impact of heavy rainfall, floods and mudslides. Therefore, the disaster entity factors and disaster-bearing carrier factors are used as the sample space of the prediction model, which is divided into three dimensions: distribution line and equipment factors, geographical environment factors and disaster weather factors.

[0075] Specifically, the distribution network risk prediction method under extreme weather conditions provided by the embodiment of the present invention includes:

[0076] S1: Obtain historical data related to distribution network risks, and standardize the historical data to obtain standardized data, wherein the relevant historical data includes equipment inventory data, operation hidden danger data, and external environment impact weights;

[0077] S2: Obtain real-time operation data related to distribution network equipment, identify and push all data, and build a distribution equipment status perception model based on the health status of the equipment;

[0078] S3: Based on the output results of the distribution equipment state perception, a distribution network risk prediction model is established in combination with a multi-data weight comprehensive evaluation method;

[0079] S4: Based on the distribution network risk prediction model, the distribution network operation risk is predicted.

[0080] In the process of predicting the operation risks of the distribution network, it is necessary to first construct data samples for model analysis, including: distribution line and equipment variables, geographical environment variables and disaster weather variables.

[0081] It should be noted that the variables of power distribution lines and equipment mainly consider historical disaster experience. The equipment that has been affected is more likely to be affected again than the equipment that has not been affected due to the different environment and equipment characteristics. The historical disaster loss variables are used to mark the damage of equipment in typhoons and rainstorms in previous years. If there is a disaster loss, it is recorded as 1, and if there is no disaster loss, it is recorded as 0. In addition to external factors, the failure rate of the equipment itself is also an important factor in determining its ability to resist disasters. It is determined by the equipment design parameters, aging and corrosion, whether it has been reinforced against wind, and other characteristics. The failure rate of most equipment is a function of time, that is, the bathtub curve. Therefore, in order to reduce the complexity of the model and improve the prediction speed and accuracy, the equipment commissioning years are used instead of the equipment failure rate.

[0082] In addition, considering that the disaster reliability of distribution lines and equipment is different under different geographical conditions, for example, the wind speed in different terrains in the same area will be different. The narrow channel effect of the terrain will cause the wind speed to increase in a short period of time. The density of surface buildings and vegetation will affect the movement and disappearance of typhoons, and thus affect the actual wind speed that the equipment can withstand. Therefore, the geographical environment variables mainly consider the slope, soil quality and ground roughness of the distribution lines and equipment.

[0083] Finally, considering that strong winds and heavy rainfall are the main factors causing damage to distribution lines and equipment in typhoon and rainstorm weather, the impact on the distribution network is determined by the instantaneous maximum wind speed, continuous wind speed, wind direction and rainfall. Therefore, the disaster weather variables consider the maximum wind speed level, average wind speed level, maximum wind speed and wind direction, average wind direction and rainfall level during the monitoring period, among which the rainfall level is the depth level of the water layer falling to the ground during the monitoring period. The wind speed and rainfall levels are divided according to the standards specified by the China Meteorological Administration.

[0084] Furthermore, before using the model for data analysis, the data samples need to be normalized and screened, including: converting different words in the text type field into ordered discrete variables according to the degree of description, and then normalizing them: Among them, x i ′ is the normalized value of the i-th data, X iis the true value of the i-th data, |X| max The maximum absolute value in the data of this field.

[0085] Considering that some of the multiple disaster-causing variables have strong correlations, directly bringing them into the prediction model will increase the model calculation complexity and make the prediction results overfit and deviate from reality. Therefore, by calculating the variable mutual information correlation matrix, the strongly correlated variables are eliminated and the optimal key disaster-causing variables (hereinafter referred to as key variables) that affect the distribution network disaster losses are screened out. The variable correlation coefficient r is: In the formula, I(X,Y) is the mutual information between variables X and Y, H(X) is the information entropy of variable X, and H(Y) is the information entropy of variable Y. It is generally believed that a correlation coefficient r greater than 0.8 is a strong correlation, so the variables with stronger correlation with other variables in the strongly correlated variable pairs are screened and removed to form a key variable sample set.

[0086] In the process of building a distribution equipment state perception model, it is necessary to first generate a decision tree based on a training set consisting of data samples. Figure 2 is a flow chart of an optional method for generating a decision tree according to an embodiment of the present invention, such as Figure 2 As shown in FIG. 1 , the method includes: performing Bootstrap sampling on the training set, extracting n samples from the training set with replacement for each training subset, and generating a total of m training subsets; counting the data outlier rate λ and the number of feature quantity types k of each training subset; for each training subset, randomly extracting L feature quantities as candidate features for decision making at the current node, and generating a decision tree through the Cart algorithm, and the number of extracted feature quantities L is determined by λ and k obtained in the previous step: In the process of generating decision trees, considering the large sample data, it is necessary to limit the maximum depth of the decision tree according to the distribution of the data, that is, the adaptive decision tree level weight determines the decision tree level, and m decision trees are obtained.

[0087] Among them, the data outlier rate is the ratio of the number of data in the training subset whose proximity to all other data is less than the threshold. The higher the outlier rate, the more classifications are needed to ensure the accuracy of the prediction classification. The present invention uses this ratio as the hierarchical weight of the adaptive decision tree to determine the number of extracted features. On the one hand, it can reduce the number of unnecessary calculations of high-quality data sets, and on the other hand, it can ensure the accuracy of the decision tree. The number of feature types is the number of features available for extraction.

[0088] After generating each decision tree, it is necessary to determine the voting rights of each decision tree based on the test set composed of data samples. Figure 3 is a flowchart of an optional decision tree voting rights determination method according to an embodiment of the present invention. Figure 3As shown, the method includes: using all decision trees in the random forest algorithm to predict the test set, calculating the average prediction accuracy τ of each decision tree as the decision accuracy weight to determine the voting power of each tree.

[0089] The distribution network risk prediction method under extreme weather conditions proposed in the embodiment of the present invention predicts the risk situation of the distribution network, thereby providing early warning for key risk areas, so that the distribution network can carry out targeted pre-disaster prevention and emergency preparation work and reduce disaster losses.

[0090] Another optional method for predicting distribution network risks under extreme weather conditions can also be implemented in the following way.

[0091] Obtain historical data related to distribution network risks, and perform standardization on the historical data to obtain standardized data. The relevant historical data includes equipment inventory data, operation hidden danger data, and external environment impact weights. Among them, standardization processing includes data collection, data cleaning, data completion, data aggregation and other steps, and organizes the original data into standardized data for subsequent use and analysis.

[0092] It should be noted that in actual application scenarios, it is necessary to predict the damage of power distribution lines and equipment in the region to configure emergency repair resources in advance. Therefore, it is necessary to classify the prediction results of power distribution lines and equipment damage in the same region to provide theoretical support for relevant departments and units to carry out pre-disaster defense and post-disaster repair work.

[0093] The prediction results of damage to distribution lines and equipment in the target area are graded into four levels: blue, yellow, orange and red, according to the degree of damage, urgency and development trend that may be caused by meteorological disasters. The classification standards are as follows:

[0094] 1) No warning: Less than 15% of the distribution lines and equipment in the area are predicted to fail.

[0095] 2) Blue warning: 15% to 30% of the distribution lines and equipment in the area are predicted to fail.

[0096] 3) Yellow warning: 30% to 50% of the distribution lines and equipment in the area are predicted to fail.

[0097] 4) Orange warning: 50% to 70% of the distribution lines and equipment in the area are predicted to fail.

[0098] 5) Red warning: More than 70% of the distribution lines and equipment in the area are predicted to fail.

[0099] Furthermore, the present invention also provides a distribution network equipment operation risk assessment system, including: a data acquisition module, a data preprocessing module, a distribution network risk prediction model construction module and a distribution network risk prediction and evaluation module.

[0100] The data acquisition module is used to collect the operating data of the distribution network equipment in real time through the sensor network. These sensors can include voltage sensors, current sensors, temperature sensors, humidity sensors, etc., which are used to monitor various operating parameters of the equipment. The collected data is sent to the data processing center through the data transmission network to provide raw data for subsequent data preprocessing and risk assessment.

[0101] The data preprocessing module is used to receive the operating data from the data acquisition module and perform a series of data processing operations. First, through data cleaning operations, outliers, duplicate values ​​and other data that do not meet the requirements are removed. Then, outlier processing is performed, and statistical methods or machine learning algorithms are used to identify and correct abnormal data. Finally, through data standardization operations, the data is converted into a unified dimensional range so that the subsequent risk assessment model can better process and analyze the data.

[0102] The distribution network risk prediction model construction module uses a machine learning algorithm to construct a distribution network risk prediction model. First, historical operation data is collected as a training data set, including data under normal operating conditions and risk conditions. Then, the random forest algorithm is selected for model training to establish a mapping relationship between the distribution network operating status and risk. The trained model can be saved for subsequent real-time risk assessment.

[0103] The distribution network risk prediction and assessment module is used to generate corresponding risk warning information based on the risk level of the distribution network risk prediction model. When the distribution network operation risk level exceeds the preset threshold, the distribution network risk prediction and assessment module will trigger the warning mechanism and send warning information to the operation and maintenance personnel through SMS, email or system pop-up windows. The warning information includes risk level, location, possible causes and recommended countermeasures, etc., to help operation and maintenance personnel quickly locate and deal with potential operation risks.

[0104] The distribution network risk prediction algorithm under extreme weather conditions proposed in the present invention constructs a distribution network risk data sample system, and improves the model operation speed by screening the variable mutual information matrix; uses the random forest algorithm to improve the prediction accuracy; predicts and analyzes the regional distribution and level prediction of distribution network risks, and analyzes the interaction between distribution network risks and three types of disaster-causing factors, namely meteorological, geographical environment, and line and equipment conditions, based on the prediction results. The prediction results of the model can provide an effective reference for differentiated prevention of distribution network risks and emergency resource allocation.

[0105] The present invention is described below in conjunction with another optional embodiment.

[0106] Embodiment 2

[0107] A risk level prediction method and device for a distribution network provided in this embodiment includes multiple implementation units, each of which corresponds to each implementation step in the above-mentioned embodiment 1.

[0108] Figure 4 is a schematic diagram of an optional risk level prediction method and device for a distribution network according to an embodiment of the present invention, such as Figure 4 As shown, the device may include: an acquisition unit 41 and an input unit 42.

[0109] Among them, the acquisition unit 41 is used to obtain the current operation data of the target distribution network, wherein the current operation data is collected in real time by the sensor network, including: line equipment operation data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data.

[0110] The input unit 42 is used to input all current operating data into the target model and output the risk prediction level and warning information, wherein the target model is a decision tree model pre-established based on the random forest algorithm, and the decision tree model is used to extract key features from line equipment operation data, geographic environment data and disaster weather data, conduct risk level analysis based on the key features, obtain the risk prediction level, and generate corresponding warning information based on different risk prediction levels.

[0111] The above-mentioned risk level prediction method and device for distribution network can first obtain the current operating data of the target distribution network through the acquisition unit 41, wherein the current operating data is collected in real time by the sensor network, including: the line equipment operating data of the target distribution network, the geographical environment data of the location of the target distribution network, and the disaster weather data, and then input all the current operating data into the target model through the input unit 42, and output the risk prediction level and warning information, wherein the target model is a decision tree model pre-established based on the random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data and the disaster weather data, and conduct risk level analysis based on the key features to obtain the risk prediction level, and generate corresponding warning information based on different risk prediction levels.

[0112] In the embodiment of the present invention, by collecting the line equipment operation data, geographical environment data and disaster weather data of the target distribution network in real time, the data input into the model is ensured to be timely and comprehensive, so as to capture the immediate impact of extreme weather on the distribution network. The target model trained by the random forest algorithm can effectively process high-dimensional data and automatically select the most important features for risk prediction, avoiding the reliance on feature selection in the prior art methods, thereby improving the adaptability and accuracy of the model under extreme weather conditions. In addition, the decision tree analysis based on key features can reveal the complex relationship between extreme weather and distribution network risks. The model can reflect the potential risk changes in real time and generate corresponding early warning information by comprehensively analyzing multidimensional data, providing scientific decision-making basis for operation and maintenance personnel, achieving the technical effect of improving the risk prediction ability of the distribution network under extreme weather, and thus solving the technical problem of low accuracy of risk prediction of the distribution network under extreme weather in the related technology.

[0113] Optionally, the line equipment operation data includes: historical disaster loss data and equipment operation years, wherein the historical disaster loss data is used to characterize the damage suffered by the target line equipment in extreme disasters in previous years, and the equipment operation years is used to characterize the failure rate of the target line equipment.

[0114] Optionally, the geographical environment data of the location of the target distribution network includes: slope data, soil data and ground roughness, and the disaster weather data includes: maximum wind speed level, average wind speed level, maximum wind speed direction, average wind direction and rainfall level within a preset monitoring time period, wherein the rainfall level refers to the depth level of the water layer falling to the ground within the preset monitoring time period.

[0115] Optionally, the risk level prediction method device for the distribution network also includes: a preprocessing module, which is used to preprocess all current operating data, wherein the preprocessing operations include: data cleaning, data outlier processing and data standardization processing, and data standardization processing refers to mapping all data to a unified dimension range according to a preset normalization formula; a full push module, which is used to push the preprocessed current operating data in full to a pre-built distribution equipment state perception model, and the model outputs the state perception result of the target distribution network; an input module, which is used to directly input the state perception result into the target model when the state perception result is obtained, and output the risk prediction level and warning information.

[0116] Optionally, the risk level prediction method device for the distribution network also includes: an acquisition module, used to obtain risk-related data of the target distribution network in a historical time period, and construct a risk data sample, wherein the risk-related data includes: equipment inventory data, operation hazard data, and external environment impact weight data; a determination module, used to determine the risk-related characteristics of the target distribution network based on the risk data sample, and construct a decision tree based on the risk-related characteristics to obtain N decision trees, wherein N is a positive integer; an integration module, used to integrate the N decision trees based on the random forest algorithm, and give each decision tree a unique decision vote to obtain a target model.

[0117] Optionally, the determination module includes: a generation submodule, which is used to generate a model training data set and a model test data set based on the risk data sample, and extract N training subsets from the model training data set using a uniform sampling strategy; a first calculation submodule, which is used to determine the data outlier rate and the total amount of risk-related features for each training subset, and calculate the number of features to be extracted R based on the data outlier rate and the total amount of risk-related features, where R is a positive integer; an allocation submodule, which is used to extract R risk-related features from the training subset, calculate the Gini index of each risk-related feature for the training subset, and allocate the training data in the training subset to each decision node of the initial decision tree according to the R Gini indices and a preset node allocation strategy to obtain a decision tree; an integration submodule, which is used to integrate the decision trees corresponding to the N training subsets to obtain N decision trees.

[0118] Optionally, the integration module includes: a second calculation submodule, used to input the model test data set into each decision tree S times, output S decision results, and compare the S decision results with the expected results in the model test data set to calculate the decision accuracy of the decision tree, where S is a specified value; a determination submodule, used to obtain the decision accuracy corresponding to each decision tree, and determine the decision vote of each decision tree based on the decision accuracy.

[0119] The above-mentioned risk level prediction method device for distribution network may also include a processor and a memory. The above-mentioned acquisition unit 41, input unit 42, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0120] The above processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set to obtain the current operation data of the target distribution network by adjusting the kernel parameters, input all the current operation data into the target model, and output the risk prediction level and warning information.

[0121] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0122] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining current operating data of a target distribution network, wherein the current operating data is collected in real time by a sensor network, including: line equipment operating data of the target distribution network, geographic environment data of the location of the target distribution network, and disaster weather data; inputting all current operating data into a target model, and outputting risk prediction levels and warning information, wherein the target model is a decision tree model pre-established based on a random forest algorithm, and the decision tree model is used to extract key features from line equipment operating data, geographic environment data, and disaster weather data, conduct risk level analysis based on the key features, obtain risk prediction levels, and generate corresponding warning information based on different risk prediction levels.

[0123] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the risk level prediction methods for distribution networks in the above-mentioned embodiment 1.

[0124] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement any one of the risk level prediction methods for distribution networks in the above-mentioned embodiment one.

[0125] Figure 5 1 is a hardware structure block diagram of an electronic device (or mobile device) for a method for predicting risk level of a distribution network according to an embodiment of the present invention. Figure 5 As shown, the electronic device may include one or more ( Figure 5 502a, 502b, ..., 502n are used to illustrate) processor 502 (processor 502 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), memory 504 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 5The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 5 More or fewer components as shown, or with Figure 5 Different configurations shown.

[0126] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0127] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0128] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, 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 units or modules, which can be electrical or other forms.

[0129] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0130] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0131] 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 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, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0132] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A risk level prediction method for a distribution network, characterized in that: include: Acquire current operation data of the target distribution network, wherein the current operation data is collected in real time by a sensor network, including: line equipment operation data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data; All the current operating data are input into the target model, and the risk prediction level and warning information are output, wherein the target model is a decision tree model pre-established based on the random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographical environment data and the disaster weather data, and conduct risk level analysis based on the key features to obtain the risk prediction level, and generate corresponding warning information based on different risk prediction levels.

2. The risk level prediction method for distribution network according to claim 1, characterized in that: The line equipment operation data includes: historical disaster loss data and equipment operation years, wherein the historical disaster loss data is used to characterize the damage suffered by the target line equipment in extreme disasters in previous years, and the equipment operation years is used to characterize the failure rate of the target line equipment.

3. The risk level prediction method for distribution network according to claim 1, characterized in that: The geographical environment data of the location of the target distribution network includes: slope data, soil data and ground roughness; the disaster weather data includes: maximum wind speed level, average wind speed level, maximum wind speed direction, average wind direction and rainfall level within a preset monitoring time period, wherein the rainfall level refers to the depth level of the water layer falling to the ground within the preset monitoring time period.

4. The risk level prediction method for distribution network according to claim 1, characterized in that: Before inputting all the current operating data into the target model, it also includes: Preprocessing all the current running data, wherein the preprocessing operations include: data cleaning, data outlier processing and data standardization processing, wherein the data standardization processing refers to mapping all the data to a unified dimension range according to a preset normalization formula; Pushing the preprocessed current operation data in full to a pre-built distribution equipment state perception model, and having the model output a state perception result of the target distribution network; When the state perception result is obtained, the step of inputting all the current operation data into the target model and outputting the risk prediction level and warning information also includes: The state perception result is directly input into the target model, and the risk prediction level and the warning information are output.

5. The risk level prediction method for distribution network according to claim 1, characterized in that: The target model is obtained by the following steps: Acquire risk-related data of the target distribution network in a historical period and construct a risk data sample, wherein the risk-related data includes: equipment inventory data, operation hidden danger data, and external environment impact weight data; Determine risk-related characteristics of the target distribution network based on the risk data sample, and construct a decision tree based on the risk-related characteristics to obtain N decision trees, where N is a positive integer; The target model is obtained by integrating the N decision trees based on the random forest algorithm and assigning a unique decision vote to each decision tree.

6. The risk level prediction method for distribution network according to claim 5, characterized in that: The steps of determining risk-related characteristics of the target distribution network based on the risk data sample, and constructing a decision tree based on the risk-related characteristics to obtain N decision trees include: Generate a model training data set and a model test data set based on the risk data sample, and extract N training subsets from the model training data set using a uniform sampling strategy; For each of the training subsets, determine the data outlier rate and the total amount of risk-related features, and calculate the number of features to be extracted R based on the data outlier rate and the total amount of risk-related features, where R is a positive integer; Extracting R risk-related features from the training subset, calculating the Gini index of each risk-related feature for the training subset, and allocating the training data in the training subset to each decision node of the initial decision tree according to the R Gini indexes and a preset node allocation strategy, to obtain the decision tree; The decision trees corresponding to the N training subsets are integrated to obtain N decision trees.

7. The risk level prediction method for distribution network according to claim 6, characterized in that: The step of assigning a unique decision vote to each decision tree comprises: For each decision tree, input the model test data set S times into the decision tree, output S decision results, and compare the S decision results with the expected results in the model test data set to calculate the decision accuracy of the decision tree, where S is a specified value; The decision accuracy rate corresponding to each decision tree is obtained, and the decision voting weight of each decision tree is determined based on the decision accuracy rate.

8. A method and device for predicting risk level of distribution network, characterized in that: include: An acquisition unit is used to acquire current operation data of the target distribution network, wherein the current operation data is collected in real time by a sensor network, including: line equipment operation data of the target distribution network, geographical environment data of the location of the target distribution network, and disaster weather data; An input unit is used to input all the current operating data into a target model, and output risk prediction levels and warning information, wherein the target model is a decision tree model pre-established based on a random forest algorithm, and the decision tree model is used to extract key features from the line equipment operating data, the geographic environment data, and the disaster weather data, conduct risk level analysis based on the key features, obtain risk prediction levels, and generate corresponding warning information based on different risk prediction levels.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the risk level prediction method for a distribution network as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the risk level prediction method for distribution network as described in any one of claims 1 to 7.

Images