Earthquake intensity prediction method and device, storage medium and electronic equipment

Abstract

The invention relates to a seismic intensity prediction method and device, a storage medium and electronic equipment. The method comprises the steps that seismic data of a target seismic region and state abnormal data of a communication base station are acquired, the seismic data comprise the magnitude, the influence area and the influence population density, and the state abnormal data comprise the base station power-off rate and the base station offline rate; inputting the seismic data and the state abnormal data of the target seismic region into a target prediction model, so that a plurality of target algorithms of the target prediction model perform intensity prediction on the target seismic region respectively, outputting respective prediction intensity values, and fusing all the prediction intensity values to obtain a target intensity value of the target seismic region; determining a target intensity level of the target earthquake region according to the target intensity value; and generating an earthquake intensity distribution diagram according to the target intensity value and the target intensity grade of the target earthquake region. According to the invention, the technical problem that the earthquake disaster assessment method has obvious insufficiency in timeliness and accuracy is solved.

Description

Technical Field

[0001] This application relates to the field of seismic analysis technology, and particularly to a method, device, storage medium, and electronic device for predicting seismic intensity. Background Art

[0002] Post-earthquake assessment is a key link in formulating rescue strategies and allocating resources, and its timeliness is directly related to the efficiency and effectiveness of rescue work. Currently, seismic disaster assessment methods mainly include remote sensing satellite monitoring, drone patrols, and on-site manual assessments, etc. These traditional methods have exposed a series of problems in practical applications. Especially during the "black box period" after an earthquake, that is, within a period of time after the earthquake, due to lack of information or transmission delays, it is impossible to obtain disaster situation data in a timely manner, which seriously affects the efficiency and accuracy of rescue decisions.

[0003] As a widely used disaster monitoring tool, remote sensing satellites can cover a large area, but their data acquisition is often limited by satellite orbits, imaging cycles, and time scheduling, resulting in significant delays in disaster situation information. In addition, the limitations of remote sensing images in spatial and temporal resolutions make it difficult to provide detailed information about the affected area. Although drones have high flexibility and resolution, their flight range and endurance are limited, and they are difficult to play a role in complex terrains or extreme weather conditions.

[0004] On the other hand, the on-site manual assessment method requires rescue personnel to reach the affected area to collect and count information, usually including multiple steps such as surveying, taking pictures, and recording. This method not only consumes a large amount of manpower and material resources, but is also easily restricted by factors such as geographical location, road damage, and communication interruption, and it is difficult to comprehensively and accurately reflect the specific situation of the affected area in a short time.

[0005] In summary, the existing seismic disaster assessment methods have obvious deficiencies in timeliness, accuracy, and comprehensiveness, especially in the post-disaster "black box period", they cannot provide timely and reliable disaster situation information. Therefore, there is an urgent need for a technical solution that can evaluate seismic disasters in real time and quickly to make up for the deficiencies of traditional methods. Summary of the Invention

[0006] This application provides a method, device, storage medium, and electronic device for predicting seismic intensity to solve the technical problem that the seismic disaster assessment method has obvious deficiencies in both timeliness and accuracy.

[0007] In a first aspect, the present application provides a method for predicting seismic intensity, including: obtaining seismic data of a target earthquake area and status anomaly data of communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the seismic data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate; inputting the seismic data and status anomaly data of the target earthquake area into a target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area; determining the target intensity level of the target earthquake area according to the target intensity value; and generating a seismic intensity distribution map according to the target intensity value and target intensity level of the target earthquake area.

[0008] In a second aspect, the present application provides a device for predicting seismic intensity, including: a first acquisition module, configured to obtain seismic data of a target earthquake area and status anomaly data of communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the seismic data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate; a prediction module, configured to input the seismic data and status anomaly data of the target earthquake area into a target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area; a determination module, configured to determine the target intensity level of the target earthquake area according to the target intensity value; and a generation module, configured to generate a seismic intensity distribution map according to the target intensity value and target intensity level of the target earthquake area.

[0009] As an alternative example, the above device further includes: a construction module, configured to construct an initial prediction model before inputting the seismic data and status anomaly data of the above target seismic area into the target prediction model, where the above initial prediction model includes an initial support vector machine algorithm, an initial adaptive boosting algorithm, an initial gradient boosting algorithm, and an initial single-hidden layer neural network algorithm; a second acquisition module, configured to acquire a sample data set and divide the above sample data set into a training data set and a validation data set, where the above sample data set includes seismic data of multiple historical seismic areas, status anomaly data of communication base stations, and intensity values; a training module, configured to use the above sample data set to train each algorithm of the above initial prediction model to obtain a trained prediction model; an optimization module, configured to perform hyperparameter optimization on each algorithm of the above trained prediction model based on the five-fold cross-validation method to obtain the above target prediction model, where the above target prediction model includes a target support vector machine algorithm, a target adaptive boosting algorithm, a target gradient boosting algorithm, and a target single-hidden layer neural network algorithm.

[0010] As an alternative example, the above training module includes: an input unit, configured to input the above training data set into the first algorithm of the above initial prediction model to obtain a result data set output by the above first algorithm, where the above first algorithm is each algorithm of the above initial prediction model; a first calculation unit, configured to calculate the difference between the above result data set and the above validation data set using a loss function to obtain a loss value; an adjustment unit, configured to adjust the algorithm parameters of the above first algorithm according to the above loss value and recalculate the above loss value until the recalculated above loss value is less than a first threshold to obtain the above trained prediction model.

[0011] As an alternative example, the above optimization module includes: a first determination unit, configured to determine the hyperparameters that need to be adjusted for the second algorithm of the above trained prediction model, where the above second algorithm is each algorithm of the above trained prediction model; a setting unit, configured to set multiple preset values for the above hyperparameters to obtain multiple hyperparameter combinations; an evaluation unit, configured to evaluate the above multiple hyperparameter combinations based on the above five-fold cross-validation method to determine an optimal hyperparameter combination; an update unit, configured to update the above second algorithm using the above optimal hyperparameter combination to obtain the above target prediction model.

[0012] As an optional example, the above prediction module includes: a first prediction unit configured to input the seismic data and status anomaly data of the above target seismic area into the target support vector machine algorithm of the above target prediction model, so that the target support vector machine algorithm predicts the intensity of the above target seismic area and outputs a first predicted intensity value; a second prediction unit configured to input the seismic data and status anomaly data of the above target seismic area into the target adaptive boosting algorithm of the above target prediction model, so that the target adaptive boosting algorithm predicts the intensity of the above target seismic area and outputs a second predicted intensity value; a third prediction unit configured to input the seismic data and status anomaly data of the above target seismic area into the target gradient boosting algorithm of the above target prediction model, so that the target gradient boosting algorithm predicts the intensity of the above target seismic area and outputs a third predicted intensity value; a fourth prediction unit configured to input the seismic data and status anomaly data of the above target seismic area into the target single-hidden layer neural network algorithm of the above target prediction model, so that the target single-hidden layer neural network algorithm predicts the intensity of the above target seismic area and outputs a fourth predicted intensity value.

[0013] As an optional example, the above prediction module includes: a second calculation unit configured to perform weighted summation on the first predicted intensity value, the second predicted intensity value, the third predicted intensity value, and the fourth predicted intensity value to obtain the above target intensity value.

[0014] As an optional example, the above determination module includes: a second determination unit configured to determine that the target intensity level of the above target seismic area is a high intensity level when the above target intensity value is greater than or equal to a second threshold; a third determination unit configured to determine that the target intensity level of the above target seismic area is a low intensity level when the above target intensity value is less than the second threshold.

[0015] In a third aspect, the present application provides a storage medium storing a computer program, wherein the computer program, when run by a processor, executes the above method for predicting seismic intensity.

[0016] In a fourth aspect, the present application further provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above method for predicting seismic intensity through the computer program.

[0017] The above technical solutions provided by the embodiments of the present application have the following advantages compared with the prior art:

[0018] This application adopts the method of obtaining seismic data of the target earthquake area and status anomaly data of communication base stations. Among them, the above-mentioned target earthquake area is each earthquake area among all earthquake areas. The above-mentioned seismic data includes magnitude, affected area, and affected population density. The above-mentioned status anomaly data includes base station power failure rate and base station offline rate. Input the seismic data and status anomaly data of the above-mentioned target earthquake area into the target prediction model, so that multiple target algorithms of the above-mentioned target prediction model respectively perform intensity prediction on the above-mentioned target earthquake area, output their respective predicted intensity values, and fuse all the above-mentioned predicted intensity values to obtain the target intensity value of the above-mentioned target earthquake area. Determine the target intensity level of the above-mentioned target earthquake area according to the above-mentioned target intensity value. According to the target intensity value and target intensity level of the above-mentioned target earthquake area, a method for generating a seismic intensity distribution map is provided. Since in the above method, by obtaining seismic data such as magnitude, affected area, and population density of the target earthquake area, as well as status anomaly data such as power failure rate and offline rate of communication base stations, and inputting the above data into a target prediction model composed of multiple target algorithms, the multiple target algorithms respectively perform intensity prediction on the target area, fuse the prediction results to obtain the final target intensity value, and finally determine the corresponding intensity level according to the target intensity value, and automatically generate a seismic intensity distribution map based on the intensity value and level. Thus, the purpose of being able to quickly and accurately evaluate the seismic intensity is achieved, and furthermore, the technical problem that the seismic disaster assessment method has obvious deficiencies in both timeliness and accuracy is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present application and, together with the specification, are used to explain the principles of the present application.

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] One or more embodiments are illustrated by way of example in the accompanying drawings, and these exemplary illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements, unless otherwise stated, and the drawings in the figures do not constitute a proportional limitation.

[0022] Figure 1 is a flowchart of an optional method for predicting seismic intensity according to an embodiment of the present application;

[0023] Figure 2 is a schematic structural diagram of an optional device for predicting seismic intensity according to an embodiment of the present application;

[0024] Figure 3 It is a schematic diagram of an optional electronic device according to an embodiment of the present application. Detailed implementation manners

[0025] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0026] The following disclosure provides many different embodiments or examples for implementing different structures of the present application. To simplify the disclosure of the present application, the components and settings of specific examples are described below. Of course, they are only examples and are not intended to limit the present application. In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplification and clarity and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0027] According to the first aspect of the embodiments of the present application, a method for predicting earthquake intensity is provided. Optionally, as Figure 1 shown, the above method includes:

[0028] S102, obtaining earthquake data of a target earthquake area and status anomaly data of communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the earthquake data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate;

[0029] S104, inputting the earthquake data and status anomaly data of the target earthquake area into a target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area;

[0030] S106, determining the target intensity level of the target earthquake area according to the target intensity value;

[0031] S108, generating an earthquake intensity distribution map according to the target intensity value and the target intensity level of the target earthquake area.

[0032] Optionally, in this embodiment, earthquake data and status anomaly data of communication base stations are collected from the target earthquake areas (i.e., each area among all earthquake-affected areas, with the township as the smallest regional unit). The earthquake data includes, but is not limited to: magnitude, representing the size of earthquake energy (such as the Richter magnitude); affected area, the geographical area range affected by the earthquake; affected population density: the density of residents in this area, which helps to evaluate the number of affected people; earthquake occurrence time, longitude and latitude, and other data. The status anomaly data includes, but is not limited to: base station power-off rate, the unavailability rate of communication base stations in the area due to power-off; base station offline rate, the proportion of communication base stations in the area going offline for various reasons (such as power interruption, equipment damage, etc.). Among them, the operating status data of communication base stations in the target earthquake area are collected in real time through sensors and wireless technologies, including, but not limited to: the total number of base stations, the number of base stations with power outages, and the number of offline base stations, so as to infer status anomaly data such as the base station power-off rate and base station offline rate of communication base stations in the target earthquake area, and important post-disaster information can be provided within seconds to minutes after the earthquake occurs.

[0033] The collected earthquake data and communication base station status anomaly data are input into the target prediction model. The target prediction model contains multiple target algorithms, such as SVM (Support Vector Machine), AdaBoost (Adaptive Boosting Algorithm), GDBT (Gradient Boosting Decision Tree), and ELM (Extreme Learning Machine, a single-hidden-layer neural network algorithm). Each independently performs intensity prediction. Each algorithm will output a predicted intensity value based on the input data, that is, an estimated value of the earthquake intensity in the current area. The prediction results of all algorithms are fused (such as through methods like averaging and weighting) to obtain a more accurate target intensity value.

[0034] According to the calculated target intensity value, the target intensity level of this area is determined. The intensity level is a classification of the earthquake impact degree (such as low level and high level), usually obtained by comparing the predicted intensity value with the preset level division standard. Finally, based on the target intensity value and intensity level of each earthquake area, an earthquake intensity distribution map is generated. The distribution map visually shows the earthquake impact degrees of different areas and can be used for the reasonable allocation of rescue resources and disaster situation assessment.

[0035] Optionally, in this embodiment, through the rapid combination of base station status data and seismic information, a second-level intensity assessment is achieved, significantly improving the response speed compared with traditional methods. By fusing the prediction results of multiple machine learning algorithms (SVM, AdaBoost, GDBT, ELM), the influence of the error of a single algorithm is reduced, and the prediction accuracy is improved. Utilizing the wide distribution of communication base stations, which is not restricted by geographical conditions, efficient monitoring of a large area can be carried out. Generating an intensity distribution map facilitates decision-making analysis and resource allocation by the rescue command system, improving the rescue efficiency. By continuously optimizing the algorithms and models, the data characteristics in different seismic environments can be adapted, and the generalization ability can be improved. This application overcomes the deficiencies of traditional assessment methods in terms of time delay, information loss, etc., and can provide more accurate and rapid earthquake intensity assessment results.

[0036] As an optional example, before inputting the seismic data and status anomaly data of the target earthquake area into the target prediction model, the above method further includes:

[0037] Construct an initial prediction model, where the initial prediction model includes an initial support vector machine algorithm, an initial adaptive boosting algorithm, an initial gradient boosting algorithm, and an initial single-hidden layer neural network algorithm;

[0038] Obtain a sample data set, and divide the sample data set into a training data set and a validation data set, where the sample data set includes seismic data of multiple historical earthquake areas, status anomaly data of communication base stations, and intensity values;

[0039] Use the sample data set to train each algorithm of the initial prediction model respectively to obtain a trained prediction model;

[0040] Based on the five-fold cross-validation method, optimize the hyperparameters of each algorithm of the trained prediction model to obtain a target prediction model, where the target prediction model includes a target support vector machine algorithm, a target adaptive boosting algorithm, a target gradient boosting algorithm, and a target single-hidden layer neural network algorithm.

[0041] Optionally, in this embodiment, before using the target prediction model to predict the intensity of the target earthquake area, it is necessary to first construct and train and optimize the prediction model. Specifically, first, construct an initial prediction model, including an initial support vector machine algorithm (SVM), an initial adaptive boosting algorithm (Adaboost), an initial gradient boosting algorithm (GDBT), and an initial extreme learning machine algorithm (ELM). These algorithms serve as the basic algorithms in the multi-model framework. The support vector machine algorithm (SVM) is an algorithm for classification and regression, especially suitable for processing high-dimensional data, and performs well in dealing with such small-sample high-dimensional data. It can construct the best boundary line to accurately identify the potential connection between base station anomalies and earthquake intensity. The initial adaptive boosting algorithm (Adaboost) is a boosting algorithm that can combine multiple weak classifiers into a strong classifier to handle the problem of unbalanced sample data, especially sample characteristics such as fewer high-intensity events and a dominant proportion of non-powered-off data, significantly improving the overall performance and effectively alleviating the problem of data bias. The gradient boosting algorithm (GDBT) is an ensemble learning algorithm that continuously trains new decision trees to compensate for the errors of the previous model. Each tree is optimized based on the previous tree. Therefore, when the relationship between the strength of the earthquake and the signal anomaly of the communication base station in the sample is very complex, it can deeply analyze the relationship between earthquake intensity and base station anomalies and process complex multi-dimensional base station monitoring data. The initial extreme learning machine algorithm (ELM) is a special neural network with only one hidden layer. It has a very fast training speed. The parameters of the hidden layer are randomly generated and do not require iterative optimization. It directly solves the output weights and can handle larger sample sizes and diverse requirements, and can quickly train and accurately predict.

[0042] Then, obtain the earthquake data of multiple historical earthquake areas (including but not limited to magnitude, affected area, population density, etc.), the abnormal data of the communication base station status (including but not limited to power-off rate, offline rate, etc.), and the corresponding intensity values, construct a sample data set, and divide it into a training data set and a validation data set.

[0043] Then, use the training data set to train each initial algorithm respectively to obtain the trained prediction model. During the training process, each algorithm learns based on the features and labels of the input data to construct the model parameters and structure.

[0044] Finally, use the five-fold cross-validation method to optimize the hyperparameters of the trained prediction model to ensure that the generalization ability of each algorithm on different data sets is improved. After optimization, obtain the target prediction model, including the target support vector machine algorithm (SVM), the target adaptive boosting algorithm (Adaboost), the target gradient boosting algorithm (GDBT), and the target extreme learning machine algorithm (ELM).

[0045] Optionally, in this embodiment, a target prediction model integrating multiple algorithms is constructed and optimized through five-fold cross-validation, effectively improving the accuracy and generalization ability of the model. During the earthquake intensity assessment process, the target prediction model can adaptively adjust the parameters of each algorithm, enhancing the prediction accuracy and stability. In addition, the multi-algorithm fusion strategy significantly reduces the errors that may occur in a single algorithm, improving the reliability of the overall assessment effect.

[0046] As an optional example, the sample data set is used to train each algorithm of the initial prediction model respectively, and the trained prediction model obtained includes:

[0047] The training data set is input into the first algorithm of the initial prediction model, and the result data set output by the first algorithm is obtained, where the first algorithm is each algorithm of the initial prediction model;

[0048] The loss function is used to calculate the difference between the result data set and the validation data set, obtaining a loss value;

[0049] According to the loss value, the algorithm parameters of the first algorithm are adjusted, and the loss value is recalculated until the recalculated loss value is less than the first threshold, obtaining the trained prediction model.

[0050] Optionally, in this embodiment, during the above training process, each algorithm of the initial prediction model will be trained separately to optimize its parameters and improve the prediction accuracy. Taking any algorithm as an example, it is called the first algorithm:

[0051] The training data set (including earthquake data in historical earthquake areas, abnormal data of communication base stations, and their corresponding true intensity values) is input into the first algorithm. The first algorithm processes the input data according to its internal mechanism (such as: support vector classification, decision tree boosting, neural network, etc.) and outputs a predicted result data set;

[0052] An appropriate loss function (such as mean squared error MSE, logarithmic loss, cross-entropy loss, etc.) is used to calculate the difference between the output result data set of the first algorithm and the true intensity value in the validation data set, obtaining a loss value. The loss value is used to measure the accuracy of the algorithm's prediction result. The smaller the loss value, the higher the prediction accuracy of the model;

[0053] According to the calculated loss value, adjust the parameters of the first algorithm through optimization algorithms (such as gradient descent, grid search, random search, etc.). The adjusted parameters include but are not limited to: the kernel function parameter of SVM, the number and weight of weak classifiers of Adaboost, the tree depth and learning rate of GDBT, the input weight and bias of ELM, etc. Repeat the training and optimization until the loss value is reduced to a set first threshold (i.e., the accuracy of the model meets the expected requirements). When the loss value is lower than the first threshold, it is considered that the first algorithm has been trained and achieves a satisfactory prediction effect.

[0054] Perform the above training process for each algorithm of the initial prediction model, and finally obtain the trained prediction model.

[0055] As an optional example, based on the five-fold cross-validation method, optimize the hyperparameters of each algorithm of the trained prediction model, and the obtained target prediction model includes:

[0056] Determine the hyperparameters that need to be adjusted for the second algorithm of the trained prediction model, where the second algorithm is each algorithm of the trained prediction model;

[0057] Set multiple preset values for the hyperparameters to obtain multiple hyperparameter combinations;

[0058] Evaluate multiple hyperparameter combinations based on the five-fold cross-validation method to determine the optimal hyperparameter combination;

[0059] Update the second algorithm with the optimal hyperparameter combination to obtain the target prediction model.

[0060] Optionally, in this embodiment, hyperparameter optimization is a key step to improve the model performance. Optimize each trained algorithm through the five-fold cross-validation method to ensure that the model can maintain good performance under different data partitions. For each algorithm (referred to as the second algorithm) in the trained prediction model, determine the hyperparameters that need to be optimized. Examples of hyperparameters for different algorithms are as follows:

[0061] Support Vector Machine algorithm (SVM): kernel function type (such as linear, polynomial, Gaussian, etc.), regularization parameter C, kernel function parameter γ, etc.

[0062] Adaptive Boosting algorithm (Adaboost): the number of weak classifiers, learning rate, type of base classifier, etc.

[0063] Gradient Boosting Decision Tree algorithm (GDBT): the number of trees, tree depth, learning rate, minimum sample split number, etc.

[0064] Single-hidden-layer Neural Network algorithm (ELM): the number of hidden layer nodes, activation function type, etc.

[0065] Set multiple different possible values for the hyperparameters of each algorithm to form multiple hyperparameter combinations. For example: for GDBT, the number of trees can be set to: 50, 100, 200; the learning rate can be set to: 0.01, 0.1, 0.2; the maximum depth can be set to: 3, 5, 7. The number of combinations formed is: 3×3×3 = 27. Perform the following process for each hyperparameter combination: Randomly divide the training data set into five equal parts; Each time, select one of the folds as the validation set, and the remaining four folds as the training set, and repeat five times; Average the loss values obtained from the five trainings as the evaluation metric for this combination. This process ensures that the algorithm can maintain stable performance under different data partitions, avoiding overfitting or underfitting. Finally, based on the average loss value obtained from the five-fold cross-validation, select the best-performing group (the lowest loss value) from all hyperparameter combinations. Apply the optimal hyperparameter combination to the second algorithm, retrain the model, and obtain the final target prediction model. Perform this optimization process for each algorithm.

[0066] As an optional example, input the seismic data and status anomaly data of the target earthquake area into the target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, and the output respective predicted intensity values include:

[0067] Input the seismic data and status anomaly data of the target earthquake area into the target support vector machine algorithm of the target prediction model, so that the target support vector machine algorithm performs intensity prediction on the target earthquake area and outputs the first predicted intensity value;

[0068] Input the seismic data and status anomaly data of the target earthquake area into the target adaptive boosting algorithm of the target prediction model, so that the target adaptive boosting algorithm performs intensity prediction on the target earthquake area and outputs the second predicted intensity value;

[0069] Input the seismic data and status anomaly data of the target earthquake area into the target gradient boosting algorithm of the target prediction model, so that the target gradient boosting algorithm performs intensity prediction on the target earthquake area and outputs the third predicted intensity value;

[0070] Input the seismic data and status anomaly data of the target earthquake area into the target single-hidden layer neural network algorithm of the target prediction model, so that the target single-hidden layer neural network algorithm performs intensity prediction on the target earthquake area and outputs the fourth predicted intensity value.

[0071] Optionally, in this embodiment, by inputting the seismic data of the target earthquake area and the status anomaly data of the communication base station into the target prediction model, multiple machine learning algorithms are used to predict the seismic intensity of the area, and the prediction results of each algorithm are obtained. Specifically, the trained target support vector machine algorithm (SVM) is used to predict the input data. By finding the optimal decision boundary, the level of seismic intensity is classified, and finally the first predicted intensity value is output. The target adaptive boosting algorithm (AdaBoost) will train multiple weak classifiers in sequence. Each time, the weights of the samples misclassified previously are increased. Through the weighted integration method, a strong classifier is finally obtained for intensity prediction, and finally the second predicted intensity value is output. The target gradient boosting algorithm (GDBT) constructs multiple decision tree models. Each new model will make up for the deficiencies of the previous model. The output of each new model will be weighted to gradually improve the prediction performance, and finally the third predicted intensity value is output. The target extreme learning machine algorithm (ELM) makes predictions through a neural network structure with a single hidden layer. The input data is non-linearly transformed through the hidden layer to obtain the final prediction output, and finally the fourth predicted intensity value is output. By inputting the seismic data and status anomaly data of the target earthquake area into the target prediction model, four different machine learning algorithms (SVM, AdaBoost, GDBT, ELM) perform independent intensity predictions respectively. Each algorithm outputs a predicted intensity value according to its unique calculation method. The prediction results of the four algorithms can be fused in subsequent steps to obtain a more accurate seismic intensity prediction.

[0072] As an optional example, fusing all the predicted intensity values to obtain the target intensity value of the target earthquake area includes:

[0073] Performing weighted summation on the first predicted intensity value, the second predicted intensity value, the third predicted intensity value, and the fourth predicted intensity value to obtain the target intensity value.

[0074] Optionally, in this embodiment, after obtaining the predicted intensity values output by each prediction algorithm (SVM, AdaBoost, GDBT, ELM), the next step is to fuse these prediction results to obtain the target intensity value of the target earthquake area. Specifically, the predicted intensity values of each algorithm are weighted and summed. The purpose of weighted summation is to give different weights according to the performance of different algorithms in training and prediction. A common method is to set weights according to the accuracy of the algorithm or other evaluation metrics. The selection of weights can be determined through experiments or tuned through some optimization methods (such as cross-validation) to obtain the optimal fusion effect. The final target intensity value after weighted summation is obtained, and this value represents the comprehensive intensity prediction result of the target earthquake area. By weighted summing the predicted intensity values of the four algorithms, the advantages of each algorithm can be fully utilized to obtain a more accurate and stable target intensity value. This fusion method can effectively reduce the errors and biases that may be brought by single-algorithm prediction, thereby improving the reliability and accuracy of the overall prediction. The fusion method can also be selected as the simple average method, voting method, stacking method, feature-level fusion, etc. They have different applicability in different scenarios, and which method to choose specifically needs to be determined according to the characteristics of the task, the type of model, and the requirements of the target result.

[0075] As an optional example, determining the target intensity level of the target earthquake area according to the target intensity value includes:

[0076] When the target intensity value is greater than or equal to the second threshold, determining that the target intensity level of the target earthquake area is high intensity;

[0077] When the target intensity value is less than the second threshold, determining that the target intensity level of the target earthquake area is low intensity.

[0078] Optionally, in this embodiment, first, a second threshold is set to distinguish the earthquake intensity level. This threshold is usually obtained through experience or data training based on historical data or experimental analysis. If the target intensity value of the target earthquake area is greater than or equal to the second threshold, the intensity level of this area is determined as "high intensity", which means that the earthquake impact in this area is relatively serious and emergency rescue and resource allocation may be required. If the target intensity value of the target earthquake area is less than the second threshold, the intensity level of this area is determined as "low intensity". This means that the earthquake impact in this area is relatively light, the disaster situation is relatively small, and large-scale emergency responses may not be required immediately. Through such a classification method, the appropriate intensity level can be quickly and simply assigned to each earthquake area, thus helping with rescue decision-making and resource allocation.

[0079] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0080] According to another aspect of the embodiments of the present application, there is also provided a device for predicting earthquake intensity, as Figure 2 shown, including:

[0081] A first acquisition module 202, configured to acquire earthquake data of a target earthquake area and status anomaly data of communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the earthquake data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate;

[0082] A prediction module 204, configured to input the earthquake data and status anomaly data of the target earthquake area into a target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain a target intensity value of the target earthquake area;

[0083] A determination module 206, configured to determine a target intensity level of the target earthquake area according to the target intensity value;

[0084] A generation module 208, configured to generate an earthquake intensity distribution map according to the target intensity value and the target intensity level of the target earthquake area.

[0085] It should be noted that the first acquisition module 202 in this embodiment can be used to execute step S102 in the embodiments of the present application, the prediction module 204 in this embodiment can be used to execute step S104 in the embodiments of the present application, the determination module 206 in this embodiment can be used to execute step S106 in the embodiments of the present application, and the generation module 208 in this embodiment can be used to execute step S108 in the embodiments of the present application.

[0086] As an optional example, the above device further includes:

[0087] A construction module, configured to construct an initial prediction model before inputting the earthquake data and status anomaly data of the target earthquake area into the target prediction model, where the initial prediction model includes an initial support vector machine algorithm, an initial adaptive boosting algorithm, an initial gradient boosting algorithm, and an initial single-hidden layer neural network algorithm;

[0088] A second acquisition module, configured to acquire a sample data set and divide the sample data set into a training data set and a validation data set, where the sample data set includes seismic data of multiple historical earthquake regions, status anomaly data of communication base stations, and intensity values;

[0089] A training module, configured to train each algorithm of an initial prediction model using the sample data set to obtain a trained prediction model;

[0090] An optimization module, configured to perform hyperparameter optimization on each algorithm of the trained prediction model based on a five-fold cross-validation method to obtain a target prediction model, where the target prediction model includes a target support vector machine algorithm, a target adaptive boosting algorithm, a target gradient boosting algorithm, and a target single-hidden layer neural network algorithm.

[0091] As an optional example, the training module includes:

[0092] An input unit, configured to input the training data set into a first algorithm of the initial prediction model to obtain a result data set output by the first algorithm, where the first algorithm is each algorithm of the initial prediction model;

[0093] A first calculation unit, configured to calculate the difference between the result data set and the validation data set using a loss function to obtain a loss value;

[0094] An adjustment unit, configured to adjust the algorithm parameters of the first algorithm according to the loss value and recalculate the loss value until the recalculated loss value is less than a first threshold to obtain a trained prediction model.

[0095] As an optional example, the optimization module includes:

[0096] A first determination unit, configured to determine hyperparameters to be adjusted for a second algorithm of the trained prediction model, where the second algorithm is each algorithm of the trained prediction model;

[0097] A setting unit, configured to set multiple preset values for the hyperparameters to obtain multiple hyperparameter combinations;

[0098] An evaluation unit, configured to evaluate the multiple hyperparameter combinations based on a five-fold cross-validation method to determine an optimal hyperparameter combination;

[0099] An update unit, configured to update the second algorithm using the optimal hyperparameter combination to obtain a target prediction model.

[0100] As an optional example, the prediction module includes:

[0101] The first prediction unit is configured to input the seismic data and status anomaly data of the target earthquake area into the target support vector machine algorithm of the target prediction model, so that the target support vector machine algorithm predicts the intensity of the target earthquake area and outputs a first predicted intensity value;

[0102] The second prediction unit is configured to input the seismic data and status anomaly data of the target earthquake area into the target adaptive boosting algorithm of the target prediction model, so that the target adaptive boosting algorithm predicts the intensity of the target earthquake area and outputs a second predicted intensity value;

[0103] The third prediction unit is configured to input the seismic data and status anomaly data of the target earthquake area into the target gradient boosting algorithm of the target prediction model, so that the target gradient boosting algorithm predicts the intensity of the target earthquake area and outputs a third predicted intensity value;

[0104] The fourth prediction unit is configured to input the seismic data and status anomaly data of the target earthquake area into the target single-hidden layer neural network algorithm of the target prediction model, so that the target single-hidden layer neural network algorithm predicts the intensity of the target earthquake area and outputs a fourth predicted intensity value.

[0105] As an optional example, the prediction module includes:

[0106] The second calculation unit is configured to perform weighted summation on the first predicted intensity value, the second predicted intensity value, the third predicted intensity value, and the fourth predicted intensity value to obtain a target intensity value.

[0107] As an optional example, the determination module includes:

[0108] The second determination unit is configured to determine that the target intensity level of the target earthquake area is a high intensity level when the target intensity value is greater than or equal to a second threshold;

[0109] The third determination unit is configured to determine that the target intensity level of the target earthquake area is a low intensity level when the target intensity value is less than the second threshold.

[0110] For other examples of this embodiment, please refer to the above examples and will not be elaborated here.

[0111] Figure 3 is a schematic diagram of an optional electronic device according to an embodiment of the present application. As Figure 3 shown, it includes a processor 302, a communication interface 304, a memory 306, and a communication bus 308. Among them, the processor 302, the communication interface 304, and the memory 306 communicate with each other through the communication bus 308. Among them,

[0112] The memory 306 is used to store computer programs;

[0113] When the processor 302 executes the computer program stored in the memory 306, the following steps are implemented:

[0114] Obtain the seismic data of the target earthquake area and the status anomaly data of the communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the seismic data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate;

[0115] Input the seismic data and status anomaly data of the target earthquake area into the target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area;

[0116] Determine the target intensity level of the target earthquake area according to the target intensity value;

[0117] Generate a seismic intensity distribution map according to the target intensity value and the target intensity level of the target earthquake area.

[0118] Optionally, in this embodiment, the above communication bus may be a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 3 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used for communication between the above electronic device and other devices.

[0119] The memory may include a RAM, and may also include a non-volatile memory, for example, at least one disk memory. Optionally, the memory may also be at least one storage device located far from the aforementioned processor.

[0120] As an example, the above memory 306 may but is not limited to include the first acquisition module 202, the prediction module 204, the determination module 206, and the generation module 208 in the above prediction device for seismic intensity. In addition, it may also include but is not limited to other module units in the above prediction device for seismic intensity, which will not be elaborated in this example.

[0121] The above-mentioned processor can be a general-purpose processor, including but not limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; it can also be a DSP (Digital Signal Processing), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0122] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiment, and will not be elaborated herein.

[0123] Those of ordinary skill in the art can understand that Figure 3 The structure shown is only schematic. The device for implementing the above earthquake intensity prediction method can be a terminal device, which can be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a handheld computer, and a Mobile Internet Device (MID), a PAD, and other terminal devices. Figure 3 It does not limit the structure of the above electronic device. For example, the electronic device may further include more or fewer components (such as a network interface, a display device, etc.) than those shown Figure 3 in the figure, or have a different configuration from that shown Figure 3 in the figure.

[0124] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware of the terminal device through a program, and the program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a ROM, a RAM, a magnetic disk, or an optical disc, etc.

[0125] According to another aspect of the embodiments of the present application, there is also provided a computer-readable storage medium, in which a computer program is stored. When the computer program is run by a processor, it executes the steps in the above earthquake intensity prediction method.

[0126] Optionally, in this embodiment, those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware of the terminal device through a program, and this program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0127] The serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.

[0128] If the integrated unit in the above embodiments is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the above computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing one or more computer devices (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0129] In the above embodiments of the present application, the descriptions of the various embodiments each have their own focuses. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0130] In the several embodiments provided by the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. 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 displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces. The indirect coupling or communication connection of units or modules can be in an electrical or other form.

[0131] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, in each embodiment of the present application, each functional unit can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0133] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for predicting earthquake intensity, characterized in that, Including: Obtain seismic data of the target earthquake area and status abnormal data of communication base stations. Among them, the target earthquake area is each earthquake area among all earthquake areas. The seismic data includes magnitude, affected area, and affected population density. The status abnormal data includes base station power failure rate and base station offline rate. Input the seismic data and status abnormal data of the target earthquake area into a target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area. Determine the target intensity level of the target earthquake area according to the target intensity value. Generate a seismic intensity distribution map according to the target intensity value and target intensity level of the target earthquake area.

2. The method according to claim 1, wherein Before inputting the seismic data and status abnormal data of the target earthquake area into the target prediction model, the method further includes: Construct an initial prediction model, where the initial prediction model includes an initial support vector machine algorithm, an initial adaptive boosting algorithm, an initial gradient boosting algorithm, and an initial single-hidden layer neural network algorithm. Obtain a sample data set and divide the sample data set into a training data set and a validation data set. Among them, the sample data set includes seismic data of multiple historical earthquake areas, status abnormal data of communication base stations, and intensity values. Use the sample data set to train each algorithm of the initial prediction model respectively to obtain a trained prediction model. Based on the five-fold cross-validation method, perform hyperparameter optimization on each algorithm of the trained prediction model to obtain the target prediction model, where the target prediction model includes a target support vector machine algorithm, a target adaptive boosting algorithm, a target gradient boosting algorithm, and a target single-hidden layer neural network algorithm.

3. The method according to claim 2, wherein The step of using the sample data set to train each algorithm of the initial prediction model respectively to obtain a trained prediction model includes: Input the training data set into the first algorithm of the initial prediction model to obtain a result data set output by the first algorithm, where the first algorithm is each algorithm of the initial prediction model. Use a loss function to calculate the difference between the result data set and the validation data set to obtain a loss value. Adjust the algorithm parameters of the first algorithm according to the loss value, and recalculate the loss value until the recalculated loss value is less than a first threshold to obtain the trained prediction model.

4. The method according to claim 2, characterized in that The step of performing hyperparameter optimization on each algorithm of the trained prediction model based on the five-fold cross-validation method to obtain the target prediction model includes: Determine the hyperparameters that need to be adjusted for the second algorithm of the trained prediction model, where the second algorithm is each algorithm of the trained prediction model. Set multiple preset values for the hyperparameters to obtain multiple hyperparameter combinations. Evaluate the multiple hyperparameter combinations based on the five-fold cross-validation method to determine the optimal hyperparameter combination. Update the second algorithm with the optimal hyperparameter combination to obtain the target prediction model.

5. The method according to claim 1, wherein Inputting the seismic data and status anomaly data of the target earthquake area into the target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area and output their respective predicted intensity values includes: Inputting the seismic data and status anomaly data of the target earthquake area into the target support vector machine algorithm of the target prediction model, so that the target support vector machine algorithm performs intensity prediction on the target earthquake area and outputs a first predicted intensity value; Inputting the seismic data and status anomaly data of the target earthquake area into the target adaptive boosting algorithm of the target prediction model, so that the target adaptive boosting algorithm performs intensity prediction on the target earthquake area and outputs a second predicted intensity value; Inputting the seismic data and status anomaly data of the target earthquake area into the target gradient boosting algorithm of the target prediction model, so that the target gradient boosting algorithm performs intensity prediction on the target earthquake area and outputs a third predicted intensity value; Inputting the seismic data and status anomaly data of the target earthquake area into the target single-hidden layer neural network algorithm of the target prediction model, so that the target single-hidden layer neural network algorithm performs intensity prediction on the target earthquake area and outputs a fourth predicted intensity value.

6. The method according to claim 5, wherein Fusing all the predicted intensity values to obtain the target intensity value of the target earthquake area includes: Performing weighted summation on the first predicted intensity value, the second predicted intensity value, the third predicted intensity value, and the fourth predicted intensity value to obtain the target intensity value.

7. The method according to claim 1, wherein Determining the target intensity level of the target earthquake area according to the target intensity value includes: When the target intensity value is greater than or equal to the second threshold, determining that the target intensity level of the target earthquake area is high intensity; When the target intensity value is less than the second threshold, determining that the target intensity level of the target earthquake area is low intensity.

8. A device for predicting earthquake intensity, characterized in that, Includes: A first acquisition module, configured to acquire seismic data of the target earthquake area and status anomaly data of communication base stations, where the target earthquake area is each earthquake area among all earthquake areas, the seismic data includes magnitude, affected area, and affected population density, and the status anomaly data includes base station power-off rate and base station offline rate; A prediction module, configured to input the seismic data and status anomaly data of the target earthquake area into the target prediction model, so that multiple target algorithms of the target prediction model respectively perform intensity prediction on the target earthquake area, output their respective predicted intensity values, and fuse all the predicted intensity values to obtain the target intensity value of the target earthquake area; A determination module, configured to determine the target intensity level of the target earthquake area according to the target intensity value; A generation module, configured to generate a seismic intensity distribution map according to the target intensity value and the target intensity level of the target earthquake area.

9. A computer-readable storage medium storing a computer program, characterized in that, The computer program, when run by a processor, executes the method described in any one of claims 1 to 7.

10. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 7 through the computer program.

Images