Earthquake emergency material demand prediction method based on BP neural network optimized by SSA

Optimizing the BP neural network through the sparrow search algorithm solves the problem of slow convergence speed and easy to fall into local optimality in the prediction of earthquake emergency material demand, achieving more efficient and accurate prediction results, and supporting scientific decision-making.

CN119026741APending Publication Date: 2024-11-26SHENYANG INST OF COMPUTING TECH CO LTD THE CHINESE ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411110026.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Traditional BP neural networks converge slowly in the forecast of earthquake emergency material demand and are prone to fall into local optimality, making it difficult to meet the requirements of fast and accurate prediction.

Method used

The BP neural network is optimized by using Sparrow Search Algorithm (SSA) to update the weight and bias parameters by simulating sparrow foraging behavior, achieving a balance between global and local searches, and improving prediction accuracy and generalization capabilities.

Benefits of technology

It significantly improves prediction accuracy and efficiency, provides more accurate forecasting of emergency material demand, reduces human resource consumption and errors, and improves post-disaster recovery efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026741A_ABST
    Figure CN119026741A_ABST
Patent Text Reader

Abstract

The invention discloses an earthquake emergency material demand prediction method based on an SSA optimized BP neural network. According to the method, the casualty condition after an earthquake is predicted through multi-dimensional features, and the emergency material demand is indirectly predicted in combination with a safety inventory theory. In order to solve the problem that the BP neural network is slow in convergence speed and easy to fall into local optimum, global optimization is carried out on the weight and the threshold value of the BP neural network by adopting the global search capability of SSA, so that the casualty condition after the earthquake is predicted more accurately, the emergency material demand prediction result is more accurate, and the rescue resource allocation and post-disaster recovery efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of earthquake emergency management, and in particular relates to a method for predicting earthquake emergency material demand based on SSA optimized BP neural network. Background Art

[0002] Earthquakes are one of the most threatening natural disasters to human life and property safety. After an earthquake occurs, timely and accurate prediction of emergency material needs is of great significance for improving emergency response efficiency, reducing disaster losses, and protecting the lives of disaster-affected people.

[0003] In recent years, with the increasing complexity and suddenness of earthquake emergency material demand, traditional methods relying on experience and manual statistics can no longer meet the requirements of fast and accurate prediction. These methods are not only time-consuming and labor-intensive, but also prone to errors. In order to improve the accuracy and efficiency of prediction, the use of information systems and intelligent algorithms to predict the demand for earthquake emergency materials has become an urgent problem to be solved. If the neural network model is optimized through intelligent algorithms, scientific decision support can be provided to better respond to earthquake disasters.

[0004] The forecasting methods for emergency material demand can be divided into two categories: direct forecasting and indirect forecasting. The direct forecasting method relies on case-based reasoning and spatial case-based reasoning to forecast the demand for various emergency materials by building a material demand model and comparing historical data. However, the forecasting results of this method are often not convincing enough. In contrast, the indirect forecasting method predicts the demand for emergency materials by assessing the casualties after the earthquake and combining it with the safety stock theory. In recent years, this indirect forecasting method has been increasingly widely used.

[0005] At present, forecasting methods are mainly divided into model-based methods and data-driven methods. Model-based methods rely on historical data and mathematical models such as ARMA and ARIMA, but these models can usually only deal with single variable problems and are difficult to solve nonlinear and multivariate problems in emergency material demand forecasting. With the development of computer technology, artificial neural network (ANN) technology has been widely used. ANN technology is a data-driven method that simulates biological neural networks and uses activation functions to deal with nonlinear problems. It has the advantages of nonlinear calculation, self-adaptation and self-learning. Therefore, ANN performs well in emergency material demand forecasting. At present, BP (back propagation) neural network is the most commonly used forecasting algorithm, but its convergence speed is slow, it is easy to fall into local optimality, and it requires multiple iterations to achieve high accuracy. Summary of the invention

[0006] In view of the slow convergence speed and frequent local optimal problems of traditional BP neural network algorithms, the present invention proposes an improved algorithm, the Sparrow Search Algorithm (SSA), to optimize the BP neural network. SSA optimizes the weights and bias parameters of the BP neural network by simulating the foraging behavior of sparrows. By initializing the population, each sparrow individual is regarded as a set of BP neural network parameters; the fitness of each individual is calculated; the individual position is updated according to the roles of producer, follower and sentinel to achieve a balance between global and local search; iterate until the termination condition is met; and finally select the individual with the highest fitness as the optimal solution for training and predicting the BP neural network. This method significantly improves the prediction accuracy and generalization ability of the BP neural network. Combined with the model for estimating the demand for emergency materials, prediction is performed, and finally the required amount of materials is obtained to solve this problem.

[0007] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0008] A method for predicting earthquake emergency material demand based on SSA optimized BP neural network, comprising the following steps:

[0009] Step 1: Obtain earthquake disaster information prediction index data and label data;

[0010] Step 2: Establish an SSA-BP neural network prediction model, so that it takes the index data as input and the label data as output, and performs iterative fitting prediction training to obtain an ideal model; use the ideal model to predict the earthquake disaster information prediction index data actually collected to obtain the post-earthquake casualty data in the disaster area;

[0011] Step 3: Use the emergency material demand estimation model to predict the post-earthquake casualties data in the disaster area to obtain the emergency material demand required in the disaster area.

[0012] The earthquake disaster information prediction indicators include: earthquake time, magnitude, earthquake intensity, seismic fortification intensity, building damage area, and population density; the label data is casualty data.

[0013] The SSA-BP neural network prediction model includes the following steps:

[0014] Step 21, initialize the sparrow population;

[0015] Step 22, setting the number of iterations, and calculating the fitness of individual sparrows in each iteration, reallocating population roles according to the fitness values, calculating and updating the role positions, until the maximum number of iterations is reached or the iteration is terminated when the objective function reaches the minimum value; the population roles include producers, followers, and guards;

[0016] Step 23, the optimal sparrow individual finally obtained is used as the optimal weight and bias parameter of the BP neural network;

[0017] Step 24: Use the optimal weight and bias parameters to train the BP neural network so that it can perform the prediction task and output the predicted casualty data.

[0018] The initializing the sparrow population includes: defining the positions of the sparrows as the following matrix, where n is the number of sparrows, and d represents the dimension of the variable to be optimized;

[0019]

[0020] The sparrow population is divided into three roles: producers, followers and sentinels; producers are responsible for finding food sources, followers will follow producers to obtain food, and sentinels will send out an alarm signal once they find a predator, and then the producers need to lead all followers to a safe area.

[0021] The calculation of the individual fitness of the sparrows and the reallocation of the population roles according to the fitness values ​​include: calculating the fitness value F of the sparrows x , where n represents the number of sparrows, F x The value of each row in represents the fitness value of an individual;

[0022]

[0023] According to the fitness value in descending order, individuals with higher fitness are selected as producers, and the remaining individuals are divided into followers and guards. The selection of guards accounts for a fixed part of the population in each iteration.

[0024] The calculation and update of the producer position includes:

[0025]

[0026] in represents the value of the jth dimension of the i-th sparrow at the t+1th iteration, where t and j represent the dimensions of the current iteration and the problem to be optimized, respectively, and α∈(0,1] is a random number; iter max represents the maximum number of iterations; Q is a uniformly distributed random number used to simulate the random flight behavior of sparrows; L is a matrix whose elements are all constant 1, used to adjust the amplitude of the sparrow position update; R 2 and ST represent the alarm value and safety value respectively, where R 2 is a random number used to simulate changes in the surrounding environment. ST is a pre-set constant used to determine the threshold for alarming.

[0027] If R 2When ST, the sparrow judges that the current environment is safe, and the producer forages and searches near the current position; conversely, if R 2 ≥ST, it means that the sparrow has discovered a predator and believes that the current environment is dangerous, and all sparrows need to quickly fly to other safe areas.

[0028] The calculation of the follower's position and update includes:

[0029]

[0030] Among them, X p represents the optimal position currently occupied by the producer; X worst represents the current global worst position; A + is a 1×d matrix with all elements being 1 or -1, used to adjust the frequency of position update; L represents a constant associated with the problem scale, used to adjust the amplitude of position update; n is the total number of sparrows;

[0031] When i > n / 2, it means that the current sparrow is an individual in the second half of the group. Because of its poor fitness, it may be in a state of hunger, and these individuals usually need to conduct a large-scale exploration; conversely, it means that the current sparrow is an individual in the first half of the group. These individuals have better fitness and can conduct local searches around the producer.

[0032] The calculation of the vigilant's position and update includes:

[0033]

[0034] Among them, X best refers to the current global optimal position. β is the control coefficient of the step size, which is a normal distribution of random numbers with a mean of 0 and a variance of 1. K is a random number from -1 to 1. f i represents the fitness value of the current sparrow, f g and f w are the current global optimal and worst fitness values respectively. ε is a very small integer set to avoid division by zero error;

[0035] When f i > f g , it means that the sparrow is at the edge of the group and is in danger of being captured by the predator; when f i = f g , it means that the sparrow perceives danger and needs to fly towards other sparrows to reduce the risk of being captured.

[0036] Estimate the emergency material demand according to the safety inventory theory. The objective function is:

[0037]

[0038] Where k represents a certain emergency material; P i (t) represents the number of injured persons at disaster site i at time t; a k represents the average hourly demand for emergency supplies k by the people in the earthquake-stricken area; LT represents the lead time for supplying supplies to the disaster area; represents the quantity of emergency supplies k stored in disaster area i before the earthquake; is the quantity of material k that has been delivered to disaster site i within time t-ε; Q k Indicates different service levels and the supply of emergency supplies under the lead time LT; through this objective function, the quantity D of emergency supplies k that should be supplied to the disaster site i within a certain period of time can be estimated i (t), where k 1 Indicates food and drink, k 2 Indicates cold-weather items, k 3 Indicates medical supplies;

[0039] a. The number of survivors S is calculated according to the formula i (t);

[0040] S i (t) = δ i (t)-X i (t)

[0041] The time when rescue starts after the earthquake is recorded as 0, and the time from this time is measured in hours, and recorded as t (t=0,1,2,...); let δ i is the total population of disaster area i, X i (t) represents the casualties data of disaster area i at time t;

[0042] b. Calculate the service level by formula and the situation of emergency material supply under lead time LT k ;

[0043]

[0044] in It indicates the service level of emergency material supply, that is, the degree to which the material supply meets the needs of the people in the disaster area; The service level is The service level coefficient corresponding to the case; LT represents the lead time for supplying materials to the disaster area;

[0045] c. Calculate the standard deviation of the average hourly demand for emergency supplies k at disaster site i up to time t

[0046]

[0047] d. Calculate the average hourly demand for relief supplies k at disaster site i before time t

[0048]

[0049] The present invention has the following advantages and beneficial effects:

[0050] 1. Improve prediction accuracy: The global search capability of the SSA algorithm is used to optimize the weight and bias parameters of the BP neural network, which significantly improves the accuracy of predicting the demand for earthquake emergency materials.

[0051] 2. Accelerate convergence speed: It solves the problem of slow convergence speed and easy falling into local optimality of traditional BP neural network, making the prediction process more efficient.

[0052] 3. Improve resource allocation efficiency: More accurately predict casualties after an earthquake, making the forecast of emergency material demand more precise, thereby improving the efficiency of rescue resource allocation and post-disaster recovery.

[0053] 4. Intelligent decision support: Through automated monitoring and prediction by information systems, human resource consumption and errors are reduced, providing scientific decision support. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A flow chart of the SSA-BP earthquake emergency material demand forecasting method of the present invention; DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the invention, so the present invention is not limited by the specific implementation disclosed below.

[0056] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0057] Step 1: Obtain earthquake disaster information prediction indicators: including earthquake time, magnitude, earthquake intensity, seismic fortification intensity, affected population (10,000 people), house damage (km 2 ), population density, etc.

[0058] Step 2: Predict the casualties after the earthquake: Obtain the casualties after the earthquake in the disaster area through the SSA-BP neural network prediction model.

[0059] Step 3: Estimate the demand for emergency supplies: Based on the predicted post-earthquake casualties data, the emergency supplies demand estimation model is used to obtain the emergency supplies needed in the disaster area.

[0060] Among them, the SSA-BP neural network optimization steps are:

[0061] Initialize the population and regard each sparrow individual as a set of BP neural network parameters.

[0062] Calculate the fitness value of each individual and update the individual position according to the roles of producer, follower and sentinel.

[0063] The cycle is iterated until the termination condition is met, and finally the individual with the highest fitness is extracted as the optimal solution for training and predicting the BP neural network.

[0064] Among them, the estimated demand for emergency supplies is:

[0065] Based on the post-earthquake casualties (including deaths and injuries), the demand for emergency supplies is estimated through a formula, taking into account factors such as the number of survivors, supply demand, supply lead time, service level, etc.

[0066] The final output is the forecast result, which provides a basis for emergency response and resource allocation. The forecast result is the consumption of certain items in the categories of food, cold-weather items, and medical supplies.

[0067] Specifically, a method for predicting earthquake emergency material demand based on a back propagation (BP) neural network optimized by a sparrow search algorithm (SSA) comprises the following steps:

[0068] 1. Obtain earthquake disaster information prediction indicators;

[0069] 2. Obtain the data of casualties in the disaster area after the earthquake through the SSA-BP neural network prediction model;

[0070] 3. Obtain the emergency material demand required in the disaster area through the emergency material demand estimation model;

[0071] The earthquake disaster information prediction indicators include: earthquake time, magnitude, earthquake intensity, seismic fortification intensity, affected population (10,000 people), house damage (km 2 ), population density;

[0072] Among them, the steps of the back propagation (BP) neural network optimized by the sparrow search algorithm (SSA) are as follows:

[0073] a) Initialize the population

[0074] The position of the sparrows can be represented by the following matrix, where n is the number of sparrows and d represents the dimension of the variable to be optimized:

[0075]

[0076] Sparrow populations are divided into three roles: producers, followers and guards.

[0077] Producers are responsible for finding food sources;

[0078] Followers will follow producers to obtain food;

[0079] Once the sentinel spots a predator, it will send out an alarm signal, and then the producer needs to lead all the followers to a safe area.

[0080] In a population, producers and followers can switch to each other. The switching mechanism is based on the fitness value and the algorithm's iterative process. In each iteration, the sparrow individuals are sorted according to their fitness values, and individuals with higher fitness are selected as producers, while the remaining individuals are divided into followers and guards. The selection of guards is fixed to a certain part of the population in each iteration.

[0081] b) Fitness evaluation

[0082] The fitness values ​​of all sparrows can be represented by the following vector, where n represents the number of sparrows, F x The value of each row in represents the fitness value of the individual:

[0083]

[0084] c) Producers

[0085] Generally speaking, producers are usually individuals with higher fitness. The producer's responsibility is to provide food areas or directions for all foragers, and it is responsible for finding where there is abundant food. The producer position update formula is as follows:

[0086]

[0087] in, represents the value of the jth dimension of the i-th sparrow at the t+1th iteration, where t and j represent the dimensions of the current iteration and the problem to be optimized, respectively. α∈(0,1] is a random number; iter max represents the maximum number of iterations; Q is a uniformly distributed random number used to simulate the random flight behavior of sparrows; L is a matrix whose elements are all constant 1, used to adjust the amplitude of the sparrow position update; R 2 and ST represent the alarm value and safety value respectively, where R 2is a random number used to simulate changes in the surrounding environment. ST is a pre-set constant and is the threshold for determining whether to give an alarm. If R 2 < ST, the sparrow determines that the current environment is safe, and the producer can forage and search near the current location. Conversely, if R 2 ≥ ST, it means that the sparrow has detected a predator and considers the current environment dangerous. All sparrows need to quickly fly to other safe areas.

[0088] d) Followers

[0089] Followers will frequently observe the behavior of the producer. Once they find that the producer has discovered food, the followers will immediately leave their current positions to compete for the food. If the competition is successful, the followers will obtain the food discovered by the producer. Otherwise, they will continue to update their positions. The position update formula for followers is as follows:

[0090]

[0091] where, X p represents the optimal position currently occupied by the producer, and X worst represents the current globally worst position (calculated according to the fitness in step b). A + is a 1×d matrix with all elements being 1 or -1, used to adjust the frequency of position update. L represents a constant associated with the problem scale, used to adjust the amplitude of position update. n is the total number of sparrows. When i > n / 2, it means that the current sparrow is an individual in the second half of the group. Because of its poor fitness, it may be in a hungry state, and these individuals usually need to conduct a larger range of exploration. Conversely, it means that the current sparrow is an individual in the first half of the group. These individuals have better fitness and can conduct local searches around the producer.

[0092] e) Guardians

[0093] Guardians account for a fixed part of the group, generally accounting for 10% to 20%. These guardians are randomly generated at the initial position of the population. The position update formula for guardians is as follows:

[0094]

[0095] where, X best refers to the current globally optimal position. β is the control coefficient of the step size, which is a normal distribution of random numbers with a mean of 0 and a variance of 1. K is a random number from -1 to 1. f i represents the fitness value of the current sparrow, f g and f w are the current globally optimal and worst fitness values respectively. ε is a very small integer set to avoid division by zero error. When f i>f g When the sparrow is at the edge of the group, it is in danger of being captured by predators. i =f g , it means the sparrow senses danger and needs to fly toward other sparrows to reduce the risk of being captured.

[0096] f) Iterate step b) to step e) until the maximum number of iterations is reached or the objective function is minimized in advance. The objective function may be a root mean square error function RMSE.

[0097] g) Output the optimal sparrow individual obtained in the end, that is, the optimal weight and bias parameters of the BP neural network. Use the optimal weight and bias parameters to train the BP neural network and make predictions.

[0098] Specifically, the earthquake disaster information prediction indicators (earthquake time, magnitude, earthquake intensity, seismic fortification intensity, house damage (km 2 ) and population density) are used as the input of the BP neural network, and the disaster-affected population (10,000 people) is used as the predicted output of the BP neural network for training.

[0099] The affected population is then predicted, including the number of deaths and injuries.

[0100] According to the safety stock theory, the demand for emergency supplies is estimated, and the objective function is:

[0101]

[0102] Among them, k represents some emergency supplies, such as drinking water, sleeping bags, tents, etc.; P i (t) represents the number of deaths and injuries at the disaster site i at time t; a k represents the average hourly demand for emergency supplies k by the people in the earthquake-stricken area; LT represents the lead time for supplying supplies to the disaster area; represents the quantity of emergency supplies k stored in disaster area i before the earthquake; is the quantity of material k that has been delivered to disaster site i within time t-ε; Q k Indicates different service levels and the supply of emergency supplies under the lead time LT; through this objective function, the quantity D of emergency supplies k that should be supplied to the disaster site i within a certain period of time can be estimated i (t), where k 1 Indicates food and drink, k 2 Indicates cold-weather items, k 3 Indicates medical supplies.

[0103] in:

[0104] S i (t) = δi (t)-X i (t) (7)

[0105] The time when rescue started after the earthquake was recorded as 0, and the time from here on was recorded as hours, and t (t = 0, 1, 2, ...). Let δ i is the total population of disaster area i, X i (t) represents the number of deaths in disaster area i at time t, then the number of survivors S can be obtained through this formula i (t).

[0106]

[0107] in, It indicates the service level of emergency material supply, that is, the degree to which the material supply meets the needs of the people in the disaster area; The service level is The service level coefficient corresponding to the case; LT represents the lead time for supplying materials to the disaster area. This formula can be used to reflect the different service levels. and the situation of emergency material supply under lead time LT k .

[0108]

[0109] in It represents the standard deviation of the average hourly demand for emergency supplies k at disaster site i up to time t.

[0110]

[0111] in, It represents the average hourly demand for relief supplies k at disaster site i before time t.

[0112] This is the disaster data of the 7.1 magnitude earthquake that occurred in Zaduo County, Qinghai Province on April 14, 2010. According to the prediction indicators, the data collected are the magnitude, occurrence time, fortification intensity, damage intensity, affected population, housing destruction and population density. The data are shown in Table 1 below. These prediction indicators are expressed in C i (i=1,2,3…7) to represent:

[0113] Table 1

[0114] area <![CDATA[c 1 ]]> <![CDATA[c 2 ]]> <![CDATA[c 3 ]]> <![CDATA[c 4 ]]> <![CDATA[c 5 ]]> <![CDATA[c 6 ]]> <![CDATA[c 7 ]]> Zaduo County 21:27:00 7 6 7 0.2 2491 0.4

[0115] The data in Table 1 are used as the input vector of the SSA-BP neural network model, and the predicted casualty data are shown in Table 2.

[0116] Table 2

[0117] area Number of injured Death toll Zaduo County 312 57

[0118] In the early stage after an earthquake, the demand for emergency supplies is the most urgent. Therefore, the focus of the present invention is to predict the demand for emergency supplies at this stage, and specifically calculate the demand for emergency supplies at time t=0. From Table 2, it can be seen that 312 people were injured and 57 people died. After inquiry, the total population of Zaduo County is 45,000. The materials to be used in the present invention are drinking water, compressed biscuits, tents, sleeping bags, and anti-inflammatory drugs, each with a i (i=1,2,3,...,5) indicates that the estimation is performed according to formulas (6) to (10). The relevant parameters and estimation results are shown in Table 3:

[0119] Table 3

[0120]

[0121] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may be modified, changed and expanded in various ways. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting earthquake emergency material demand based on SSA optimized BP neural network, characterized in that: It includes the following steps: Step 1: Obtain the seismic disaster information prediction index data and label data; Step 2: Establish an SSA-BP neural network prediction model, which takes the index data as input and the label data as output, and performs iterative fitting prediction training to obtain an ideal model; use the ideal model to predict the actually collected seismic disaster information prediction index data to obtain the post-earthquake casualty data in the disaster area; Step 3: Predict the post-earthquake casualty data in the disaster area through an emergency material demand estimation model to obtain the required emergency material demand in the disaster area.

2. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 1 is characterized in that: The seismic disaster information prediction indexes include: earthquake occurrence time, magnitude, seismic intensity, seismic fortification intensity, damaged building area, population density; the label data is casualty data.

3. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 1 is characterized in that: The SSA-BP neural network prediction model includes the following steps: Step 21: Initialize the sparrow population; Step 22: Set the number of iterations, and calculate the fitness of each sparrow individual during each iteration. Reassign the population roles according to the fitness value, calculate the role positions and update them until the maximum number of iterations is reached or the iteration stops when the objective function reaches the minimum value; the population roles include producers, followers, and vigilant birds; Step 23: The finally obtained optimal sparrow individual is used as the optimal weight and bias parameters of the BP neural network; Step 24: Use the optimal weight and bias parameters to train the BP neural network so that it can perform the prediction task and output the predicted casualty data.

4. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 3 is characterized in that: The initialization of the sparrow population includes: defining the position of the sparrow as the following matrix, where n is the number of sparrows and d represents the dimension of the variable to be optimized; The sparrow population is divided into three roles, namely producers, followers, and vigilant birds; producers are responsible for finding food sources, followers will follow the producers to obtain food, and vigilant birds will send out alarm signals once they detect predators, and then the producers need to lead all the followers to a safe area.

5. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 3 is characterized in that: The calculation of the individual fitness of the sparrows and the reallocation of the population roles according to the fitness values ​​include: calculating the fitness value F of the sparrows x , where n represents the number of sparrows, F x The value of each row in represents the fitness value of an individual; Arrange in descending order according to the fitness value, select the individuals with higher fitness as producers, and the remaining individuals are divided into followers and vigilant birds. The selection of vigilant birds is fixed at a certain proportion of the population during each iteration process.

6. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 3 is characterized in that: The calculation of the producer position and update includes: in represents the value of the jth dimension of the i-th sparrow at the t+1th iteration, where t and j represent the dimensions of the current iteration and the problem to be optimized, respectively, and α∈(0,1] is a random number; iter max represents the maximum number of iterations; Q is a uniformly distributed random number used to simulate the random flight behavior of sparrows; L is a matrix whose elements are all constant 1, used to adjust the amplitude of the sparrow position update; R2 and ST represent the alarm value and safety value respectively, where R2 is a random number used to simulate the changes in the surrounding environment, and ST is a pre-set constant used to determine the threshold for alarming; If R2 < ST, the sparrow judges that the current environment is safe, and the producer forages and searches near the current position; otherwise, if R2 ≥ ST, it means that the sparrow has detected a predator and considers the current environment dangerous, and all sparrows need to quickly fly to other safe areas.

7. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 3 is characterized in that: The calculation of the follower position and update includes: Where X p Represents the optimal position currently occupied by the producer; X worst represents the worst position in the world at present; A + is a 1×d matrix whose elements are all 1 or -1, which is used to adjust the frequency of position updates; L represents a constant associated with the problem scale, which is used to adjust the amplitude of position updates; n is the total number of sparrows; When i > n / 2, it means that the current sparrow is an individual in the second half of the group. Because of its poor fitness, it may be in a state of hunger, and these individuals usually need to explore in a larger range; otherwise, it means that the current sparrow is an individual in the first half of the group. These individuals have better fitness and can conduct local searches around the producer.

8. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 3 is characterized in that: The calculation of the vigilant bird position and update includes: Where X best refers to the current global optimal position, β is the control coefficient of the step size, which is a normal distribution of random numbers with a mean of 0 and a variance of 1, K is a random number from -1 to 1, and f i Indicates the current sparrow's fitness value, f g and f w are the current global optimal and worst fitness values, respectively, and ε is a very small integer set to avoid zero division error; When f i >f g When , it means that the sparrow is on the edge of the group and is in danger of being captured by predators; when f i =f g When the sparrow senses danger, it means it needs to fly toward other sparrows to reduce the risk of being captured.

9. The earthquake emergency material demand forecasting method based on SSA optimized BP neural network according to claim 1 is characterized in that: Estimate the emergency material demand according to the safety stock theory, and the objective function is: Where k represents a certain emergency material; P i (t) represents the number of injured persons at disaster site i at time t; a k represents the average hourly demand for emergency supplies k by the people in the earthquake-stricken area; LT represents the lead time for supplying supplies to the disaster area; represents the quantity of emergency supplies k stored in disaster area i before the earthquake; is the quantity of material k that has been delivered to disaster site i within time t-ε; Q k Indicates different service levels and the supply of emergency materials under the lead time LT; through this objective function, the quantity D of emergency materials k that should be supplied to the disaster site i within a certain period of time can be estimated i (t), where k1 represents food and drink, k2 represents cold-weather items, and k3 represents medical supplies; a. Where, Calculate the number of survivors S according to the formula i (t); S i (t)=δ i (t)-X i (t) The time when rescue starts after the earthquake is recorded as 0, and the time from this time is measured in hours, and recorded as t (t = 0, 1, 2, ...); let δ i is the total population of disaster area i, X i (t) represents the casualties data of disaster area i at time t; b. Calculate the service level by formula and the situation of emergency material supply under lead time LT k ; in It indicates the service level of emergency material supply, that is, the degree to which the material supply meets the needs of the people in the disaster area; The service level is The service level coefficient corresponding to the case; LT represents the lead time for supplying materials to the disaster area; c. Calculate the standard deviation of the average hourly demand for emergency supplies k at disaster site i up to time t d. Calculate the average hourly demand for relief supplies k at disaster site i before time t