Electric power emergency material prediction and material guarantee risk assessment method and system

Abstract

The embodiment of the invention provides an electric power emergency material prediction and material guarantee risk assessment method and system, and belongs to the technical field of emergency material prediction. The evaluation method comprises the following steps: acquiring emergency management data information; performing feature processing on the data information; constructing an emergency material supply risk quantitative evaluation model by using a multi-layer perceptron neural network; inputting the data information into an emergency material supply risk quantitative evaluation model to obtain safety inventory data; and comparing the safety inventory data with data of material data indexes in emergency management data information to obtain an emergency material supply risk level. According to the method, the emergency management data information is arranged, the emergency material prediction model is established, a practical tool is provided for the power grid operation to cope with natural disasters and emergencies, and the method has good generalizability and usability.

Description

Technical Field

[0001] The present invention relates to the technical field of emergency rescue material prediction, and in particular to a method and system for power emergency material prediction and material security risk assessment. Background Art

[0002] The power system occupies a pivotal position in the national economy, but the power system faces the challenges of various meteorological disasters. With global warming, extreme weather increases, such as extreme temperatures, wind damage, floods and other meteorological disasters. In the case of insufficient power emergency material inventory, it is difficult to ensure the normal operation of the power system. Therefore, ensuring the adequate supply of power materials has a very important role and practical significance. This puts forward higher requirements for emergency material support to strengthen risk assessment, reasonable resource allocation, and rapid response to emergency disposal. Power grid emergencies are more harmful to the social economy. Emergency decision-makers need to formulate emergency rescue plans with the highest efficiency according to the situation of the incident, reasonably dispatch emergency personnel and materials, and meet the needs of the scene as much as possible. The accuracy of emergency material demand forecast is directly related to the determination of power grid repair plans, the effective allocation of emergency resources and the economic efficiency of investment. At present, the power data is too subjective to carry out emergency power material demand forecasting, and the risk assessment and quantification of emergency material support are inaccurate. Summary of the invention

[0003] The purpose of the embodiments of the present invention is to provide a method and system for predicting power emergency materials and assessing material security risks, which solves the problems of overly subjective forecasting of power material demand for emergency events based on power data and inaccurate quantitative assessment of material security risks for emergency events.

[0004] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a method for predicting power emergency materials and assessing material security risks, the assessment method comprising:

[0005] Obtain emergency management data and information;

[0006] Performing feature processing on the data information;

[0007] Use multi-layer perceptron neural network to build a quantitative assessment model for emergency material supply risk;

[0008] Inputting the data information into the emergency material supply risk quantitative assessment model to obtain safety inventory data;

[0009] The safety inventory data is compared with the data of the material data indicators in the emergency management data information to obtain the emergency material supply risk level.

[0010] Optionally, obtain emergency management data information, including:

[0011] Obtain data information on emergency material allocation ledger data, internal data on equipment in operation, and external data on various meteorological disasters;

[0012] Constructing an emergency material supply risk assessment index system based on the data information to obtain indicators;

[0013] Using principal component analysis, the indicators are reduced in dimension to obtain emergency event indicators, meteorological data indicators and material data indicators;

[0014] The data information was collated using exploratory analysis methods.

[0015] Optionally, obtain emergency management data information, including:

[0016] Using one-hot encoding to convert the categorical features of the data of the emergency event indicator into numerical form to obtain new feature values, the categorical features including event category, warning level and event type;

[0017] Normalizing the digital features of the meteorological data indicators and the safety inventory data indicators;

[0018] Two or more features of the emergency event indicator data, the meteorological data indicator data and the safety stock data are combined to construct a new cross feature.

[0019] Optionally, the emergency material supply risk quantitative assessment model includes:

[0020] The input layer is used to receive the input data of emergency event indicators and meteorological data indicators;

[0021] A hidden layer, used for performing nonlinear transformation on the data information;

[0022] The output layer is used to output safety stock data.

[0023] Optionally, a multi-layer perceptron neural network is used to construct a quantitative assessment model for emergency material supply risk, including:

[0024] Determine the number of nodes in the input layer and the output layer, the number of nodes in the input layer is equal to the dimension of the input feature;

[0025] Selecting the activation function of the hidden layer according to formula (1);

[0026] ReLU(x)=max(0,x) , (1)

[0027] Among them, x is the input data of emergency event indicators and meteorological data indicators, and ReLU (x) is the activation function;

[0028] No activation function or a linear function is used for the output layer;

[0029] Determine the loss function according to formula (2) or (3):

[0030]

[0031] Among them, Y i is the output of the i-th evaluation model, is the target output of the i-th evaluation model, n is the number of iterative training epochs, MSE is the mean square error, and MAE is the mean absolute error;

[0032] Initialize weights and biases.

[0033] Optionally, a multi-layer perceptron neural network is used to construct a quantitative assessment model for emergency material supply risk, including:

[0034] Dividing the data of the emergency event indicators and the meteorological data indicators into a training set, a validation set and a test set;

[0035] Build a data pipeline to load the training set, validation set, and test set;

[0036] Performing iterative training on the training set for multiple epochs;

[0037] Calculate the performance index of the evaluation model on the training set and the validation set;

[0038] Determining whether the evaluation model is overfitting according to the performance indicator;

[0039] When it is determined that the evaluation model is overfitted, optimizing the evaluation model;

[0040] Calculating the performance index of the evaluation model on the test set;

[0041] The evaluation model is saved.

[0042] Optionally, the training set is iterated for multiple epochs, including:

[0043] Passing the training set into the evaluation model to obtain the output and loss of the evaluation model;

[0044] According to the loss function, calculating the parameter gradient;

[0045] The weights and biases of the evaluation model are updated using an optimizer according to the parameter gradients.

[0046] Optionally, judging whether the evaluation model is overfitting according to the performance indicator includes:

[0047] Record the performance indicators of each epoch on the training set and validation set;

[0048] Plotting the performance index into a curve according to epoch;

[0049] Whether the evaluation model is overfitting is determined according to the curve.

[0050] Optionally, when it is determined that the evaluation model is overfitted, optimizing the evaluation model includes:

[0051] When the performance on the validation set begins to decline, stop training and save the current best evaluation model;

[0052] Adding a regularization term to the loss function to reduce the complexity of the evaluation model;

[0053] Reduce the number of hidden layers or nodes to reduce the complexity of the evaluation model.

[0054] On the other hand, the present invention provides a power emergency material forecasting and material security risk assessment system, the assessment system comprising a processor for executing any of the assessment methods described above.

[0055] Through the above technical solution, the embodiment of the present invention provides a method and system for predicting power emergency materials and assessing material security risks. By establishing a quantitative assessment model for emergency material supply risks, the power grid's demand for emergency materials under different disaster situations can be accurately predicted, thereby improving the pertinence and efficiency of material allocation. Through the quantitative assessment model for emergency material supply risks combined with the data of material data indicators, the material security risk level of the disaster-stricken area can be determined, making material allocation more scientific and reasonable, ensuring that key areas and equipment can get the necessary material support first, and having positive significance for comprehensively improving the power system's ability to respond to natural disasters and emergencies.

[0056] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0058] Figure 1 is a flow chart of an evaluation method according to one embodiment of the present invention;

[0059] Figure 2 is a flow chart of obtaining emergency management data information according to an embodiment of the present invention;

[0060] Figure 3 is a flow chart of a feature process according to one embodiment of the present invention;

[0061] Figure 4 is a schematic diagram of an evaluation model according to one embodiment of the present invention;

[0062] Figure 5 is a flowchart of constructing an evaluation model according to one embodiment of the present invention;

[0063] Figure 6 is a flow chart of a training evaluation model according to one embodiment of the present invention;

[0064] Figure 7 is a flowchart of iterative training according to one embodiment of the present invention;

[0065] Figure 8 is a flowchart for determining whether an evaluation model is overfitting according to an embodiment of the present invention;

[0066] Fig. 9 is a flow chart of an optimization evaluation model according to one embodiment of the present invention. DETAILED DESCRIPTION

[0067] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0068] Figure 1 is a flow chart of an evaluation method according to one embodiment of the present invention. Figure 1 The assessment method may include:

[0069] In step S1, emergency management data information is obtained.

[0070] In step S2, feature processing is performed on the data information.

[0071] In step S3, a multi-layer perceptron neural network is used to construct a quantitative assessment model for emergency material supply risk.

[0072] In step S4, the data information is input into the emergency material supply risk quantitative assessment model to obtain safety inventory data.

[0073] In step S5, the safety inventory data is compared with the data of the material data indicators in the emergency management data information to obtain the emergency material supply risk level.

[0074] The present invention can accurately predict the demand for emergency supplies by the power grid under different disaster situations by establishing a quantitative assessment model for the risk of emergency supplies, thereby improving the pertinence and efficiency of material allocation. Through the prediction model, combined with the data of material data indicators, the material security risk level of the disaster-stricken area can be determined, making the material allocation more scientific and reasonable, ensuring that key areas and equipment can get the necessary material support first, which has positive significance for comprehensively improving the power system's ability to respond to natural disasters and emergencies.

[0075] When the emergency material supply risk quantitative assessment model is needed to make a prediction, the emergency management data information is first acquired. The steps for acquiring the emergency management data information may be various steps known to those skilled in the art. In one example of the present invention, for example Figure 2 As shown, it may include:

[0076] In step S11, data information of emergency material allocation ledger data, internal data of equipment in operation, and external data of various meteorological disasters are obtained.

[0077] In step S12, an emergency material supply risk assessment index system is constructed based on the data information to obtain indicators. Among them, the emergency material supply risk quantitative assessment model is constructed based on different materials and different weather conditions. The following is an example of "disaster prediction of distribution transformers under typhoon weather conditions":

[0078]

[0079]

[0080] In step S13, principal component analysis is used to reduce the dimension of the indicators to obtain emergency event indicators, meteorological data indicators and material data indicators. Principal component analysis can reduce the number of variables in the data set while retaining the variability of the original data as much as possible.

[0081] In step S14, the data information is sorted using an exploratory analysis method. The data is cleaned and sorted using an exploratory analysis method to remove outliers and missing values, thereby ensuring the quality and integrity of the data.

[0082] In this implementation, in order to effectively use the acquired data information for the evaluation model, it is necessary to perform feature processing on the data information. The specific steps of feature processing may be various steps known to those skilled in the art. In one example of the present invention, for example Figure 3 As shown, it may include:

[0083] In step S21, the categorical features of the emergency event indicator data are converted into numerical form using one-hot encoding to obtain new feature values. The categorical features include event category, warning level and event type. For example, there are four value cases for the warning level: red, yellow, blue and orange. Assuming that there is a sample in the original data with a warning level of "red", after one-hot encoding, the sample has four new feature values ​​[1, 0, 0, 0].

[0084] In step S22, the digital features of the data of the meteorological data indicator and the safety stock data indicator are normalized. Normalization or minimum and maximum scaling is usually adopted. Normalization is to convert the feature value to a distribution with a mean of 0 and a standard deviation of 1. Minimum and maximum scaling is to scale the feature value to a given range, usually the interval [0,1].

[0085] In step S23, two or more features of the data of emergency event indicators, the data of meteorological data indicators and the safety stock data are combined to construct new cross features. Cross features are created by combining different features, which can help the evaluation model capture the interaction and complex relationship between features. For example, the product of the impact range and the duration can reflect the overall impact scale of the event, because the larger the impact range and the longer the duration, the greater the impact of the event; the product of the frequency of disaster events and the impact range can reflect the overall degree of disaster events suffered by a region. The higher the frequency of occurrence and the larger the impact range, the more serious the disaster situation in the region; the ratio of the scale of active equipment to the number of safety stocks can reflect the current equipment reserves. The larger the ratio, the more sufficient the number of active equipment and the stronger the ability to respond to emergencies; the interaction term of time features and other features: consider constructing interaction terms with time features (such as month, season, etc.) and other features (such as event type, meteorological indicators, etc.) to capture some time-related nonlinear patterns.

[0086] In this implementation, the structure of the emergency material supply risk quantitative assessment model can be any of the various structures known to those skilled in the art. In one example of the present invention, Figure 4 As shown, the evaluation model includes an input layer, a hidden layer and an output layer. The input layer is used to receive the input data of emergency event indicators and meteorological data indicators, the hidden layer is used to perform nonlinear transformation on the data information, and the output layer is used to output safety inventory data.

[0087] In this implementation, in order to further construct a quantitative assessment model for emergency material supply risk, Figure 5 As shown, it may include:

[0088] In step S31, the number of nodes in the input layer and the output layer is determined, and the number of nodes in the input layer is equal to the dimension of the input feature. The number of nodes in the input layer is equal to the dimension of the input feature, which ensures that the evaluation model can receive all feature information. The number of nodes in the output layer depends on the type of problem. The emergency material supply risk level assessment is equivalent to a regression problem, which is 1 node.

[0089] In step S32, the activation function of the hidden layer is selected according to formula (1):

[0090] ReLU(x)=max(0,x) , (1)

[0091] Among them, x is the input data of emergency event indicators and meteorological data indicators, and ReLU (x) is the activation function.

[0092] In step S33, no activation function is used for the output layer or a linear function is used.

[0093] In step S34, the loss function is determined according to formula (2) or (3):

[0094]

[0095] Among them, Y i is the output of the i-th evaluation model, is the target output of the i-th evaluation model, n is the number of iterative training epochs, MSE is the mean square error, and MAE is the mean absolute error.

[0096] In step S35, weights and biases are initialized. Common weight initialization methods include uniform distribution initialization, Xavier initialization, He initialization, etc. Usually, He initialization is used for ReLU activation function, Xavier initialization is used for Tanh or Sigmoid activation function, and the bias is usually initialized to 0 or a very small value.

[0097] In this embodiment, in order to obtain a complete emergency material supply risk quantitative assessment model, it is also necessary to input data information into the emergency material supply risk quantitative assessment model for training. The steps for training the assessment model can be various steps known to those skilled in the art. In one example of the present invention, the specific training steps can be as follows: Figure 6 As shown. Figure 6 In the training step, the training step may include:

[0098] In step S41, the data of emergency event indicators and meteorological data indicators are divided into a training set, a validation set and a test set.

[0099] In step S42, a data pipeline is constructed to load the training set, validation set, and test set. Using a data loading library in Python, such as Pytorch's DataLoader, to construct a data pipeline, efficiently load batch data, and set the batch size, helps to accelerate training and improve model generalization capabilities.

[0100] In step S43, the training set is iterated for multiple epochs. The training set is used for training, and the parameters of the evaluation model are updated through multiple iterations.

[0101] In step S44, the performance index of the evaluation model on the training set and the validation set is calculated. At the end of each epoch, the performance index of the evaluation model on the validation set is calculated. For regression problems, the root mean square error is generally calculated.

[0102] In step S45, it is determined whether the evaluation model is overfitting according to the performance index. If it is determined that the evaluation model is overfitting, step S46 is executed.

[0103] In step S46, the evaluation model is optimized.

[0104] In step S47, the performance index of the evaluation model on the test set is calculated.

[0105] In step S48, the assessment model is saved. After the training is completed, the parameters of the assessment model are saved, and the trained assessment model is deployed to the production environment to provide real-time risk level assessment services, and the performance of the model is regularly monitored to ensure the accuracy of the assessment results.

[0106] In this implementation, iterative training of multiple epochs can enable the evaluation model to continuously learn and optimize its weights and biases to improve the performance of the evaluation model. The specific training steps can be as follows: Figure 6 shown.

[0107] Specifically, in Figure 6 In the training step, the training step may include:

[0108] In step S421, the training set is passed into the evaluation model to obtain the output and loss of the evaluation model.

[0109] In step S422, the parameter gradient is calculated according to the loss function.

[0110] In step S423, the optimizer is used to update the weights and biases of the evaluation model according to the parameter gradients.

[0111] In this implementation, there may be various methods known to those skilled in the art for determining whether the evaluation model is overfitting. In one example of the present invention, for example, Figure 7 As shown, it may include:

[0112] In step S441, the performance indicators of each epoch on the training set and the validation set are recorded.

[0113] In step S442, the performance index is plotted as a curve according to epoch.

[0114] In step S443, it is determined whether the evaluation model is overfitting based on the curve. Ideally, the performance curves on the training set and the validation set should show a trend of gradual improvement and eventual convergence. If the performance curve on the validation set begins to decline after a certain epoch, while the performance curve on the training set continues to rise, overfitting may have occurred. Overfitting means that the model performs well on the training set, but has poor generalization ability on new data, i.e., the test set.

[0115] In this embodiment, in order to improve the prediction accuracy, when the evaluation model is judged to be overfitting, the evaluation model is optimized. The specific optimization steps can be as follows: Figure 8 Specifically, Figure 8 In the optimization step, the optimization step may include:

[0116] In step S451, when the performance on the validation set begins to decline, training is stopped and the current best evaluation model is saved.

[0117] In step S452, a regularization term is added to the loss function to reduce the complexity of the evaluation model.

[0118] In step S453, the number of hidden layers or nodes is reduced to reduce the complexity of the evaluation model.

[0119] On the other hand, the present invention provides a power emergency material forecasting and material security risk assessment system, the assessment system comprising a processor for executing any of the assessment methods described above.

[0120] Through the above technical solution, the embodiment of the present invention provides a method and system for predicting power emergency materials and assessing material security risks. By establishing a quantitative assessment model for emergency material supply risks, the power grid's demand for emergency materials under different disaster situations can be accurately predicted, thereby improving the pertinence and efficiency of material allocation. Through the quantitative assessment model for emergency material supply risks combined with the data of material data indicators, the material security risk level of the disaster-stricken area can be determined, making material allocation more scientific and reasonable, ensuring that key areas and equipment can get the necessary material support first, and having positive significance for comprehensively improving the power system's ability to respond to natural disasters and emergencies.

[0121] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0122] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for predicting power emergency materials and assessing material security risks, characterized in that: The evaluation methods include: Obtain emergency management data and information; Performing feature processing on the data information; Use multi-layer perceptron neural network to build a quantitative assessment model for emergency material supply risk; Inputting the data information into the emergency material supply risk quantitative assessment model to obtain safety inventory data; The safety inventory data is compared with the data of the material data indicators in the emergency management data information to obtain the emergency material supply risk level.

2. The evaluation method according to claim 1, characterized in that: Access emergency management data and information, including: Obtain data information on emergency material allocation ledger data, internal data on equipment in operation, and external data on various meteorological disasters; Constructing an emergency material supply risk assessment index system based on the data information to obtain indicators; Using principal component analysis, the indicators are reduced in dimension to obtain emergency event indicators, meteorological data indicators and material data indicators; The data information was collated using exploratory analysis methods.

3. The evaluation method according to claim 2, characterized in that: Access emergency management data and information, including: Using one-hot encoding to convert the categorical features of the data of the emergency event indicator into numerical form to obtain new feature values, the categorical features including event category, warning level and event type; Normalizing the digital features of the meteorological data indicators and the safety inventory data indicators; Two or more features of the emergency event indicator data, the meteorological data indicator data and the safety stock data are combined to construct a new cross feature.

4. The evaluation method according to claim 3, characterized in that: The emergency material supply risk quantitative assessment model includes: The input layer is used to receive the input data of emergency event indicators and meteorological data indicators; A hidden layer, used for performing nonlinear transformation on the data information; The output layer is used to output safety stock data.

5. The evaluation method according to claim 4, characterized in that: A multi-layer perceptron neural network is used to build a quantitative assessment model for emergency material supply risk, including: Determine the number of nodes in the input layer and the output layer, the number of nodes in the input layer is equal to the dimension of the input feature; Selecting the activation function of the hidden layer according to formula (1); ReLU(x)=max(0,x) , (1) Among them, x is the input data of emergency event indicators and meteorological data indicators, and ReLU (x) is the activation function; No activation function or a linear function is used for the output layer; Determine the loss function according to formula (2) or (3): Among them, Y i is the output of the i-th evaluation model, is the target output of the i-th evaluation model, n is the number of iterative training epochs, MSE is the mean square error, and MAE is the mean absolute error; Initialize weights and biases.

6. The evaluation method according to claim 5, characterized in that: A multi-layer perceptron neural network is used to build a quantitative assessment model for emergency material supply risk, including: Dividing the data of the emergency event indicators and the meteorological data indicators into a training set, a validation set and a test set; Build a data pipeline to load the training set, validation set, and test set; Performing iterative training on the training set for multiple epochs; Calculate the performance index of the evaluation model on the training set and the validation set; Determining whether the evaluation model is overfitting according to the performance indicator; When it is determined that the evaluation model is overfitted, optimizing the evaluation model; Calculating the performance index of the evaluation model on the test set; The evaluation model is saved.

7. The evaluation method according to claim 6, characterized in that: The training set is iterated for multiple epochs, including: Passing the training set into the evaluation model to obtain the output and loss of the evaluation model; According to the loss function, calculating the parameter gradient; The weights and biases of the evaluation model are updated using an optimizer according to the parameter gradients.

8. The evaluation method according to claim 7, characterized in that: Judging whether the evaluation model is overfitting according to the performance indicator includes: Record the performance indicators of each epoch on the training set and validation set; Plotting the performance index into a curve according to epoch; Whether the evaluation model is overfitting is determined according to the curve.

9. The evaluation method according to claim 8, characterized in that: When it is determined that the evaluation model is overfitted, optimizing the evaluation model includes: When the performance on the validation set begins to decline, stop training and save the current best evaluation model; Adding a regularization term to the loss function to reduce the complexity of the evaluation model; Reduce the number of hidden layers or nodes to reduce the complexity of the evaluation model.

10. A power emergency material forecasting and material security risk assessment system, characterized in that: The evaluation system comprises a processor, configured to execute the evaluation method according to any one of claims 1 to 9.

Images