Campus fire evaluation method and system based on neural network

Abstract

The invention relates to a campus fire evaluation method and system based on a neural network, and relates to the technical field of campus fire, and the method comprises the steps: constructing a campus fire safety risk evaluation index system; the evaluation indexes comprise system management, training management, fire-fighting equipment management and key fire-fighting place management; constructing a judgment matrix through a Delphi expert scoring method, and calculating a weight value of each evaluation index; optimizing the initial weight and threshold of the BP neural network based on a genetic algorithm to obtain a GA-BP neural network model; performing dimension reduction processing on the original input data based on a principal component analysis method; and based on the input data after dimension reduction, through the trained GA-BP neural network model, performing fire hazard safety risk level prediction to obtain a fire hazard risk level.

Description

Technical Field

[0001] The present application relates to the technical field of campus fires, and in particular to a campus fire evaluation method and system based on a neural network. Background Art

[0002] With the development of society and the acceleration of urbanization, the safety of large public facilities such as universities has received increasing attention. Especially in terms of fire safety, due to the dense population, complex buildings and a large number of flammable materials in universities, once a fire occurs, it often causes serious casualties and property losses.

[0003] Traditional fire safety assessment methods mainly rely on manual inspection and empirical judgment, which has many shortcomings. First, manual inspection is difficult to fully cover all potential risk points and is prone to missing important information. Second, empirical judgment is subjective, and different experts may give different evaluation results, resulting in poor consistency and accuracy of the evaluation results. In addition, traditional methods cannot monitor changes in fire risks in real time, making it difficult to detect and respond to emergencies in a timely manner.

[0004] How to solve the above technical problems is a technical problem that needs to be solved by those skilled in the art. Summary of the invention

[0005] In order to at least partially solve the above technical problems, the present application provides a campus fire assessment method and system based on a neural network.

[0006] In a first aspect, the present application provides a neural network-based campus fire assessment method that adopts the following technical solution.

[0007] A campus fire assessment method based on neural network, comprising: Construct a campus fire safety risk assessment index system; the assessment indexes include: system management, training management, fire protection facility and equipment management, and key fire protection site management; The judgment matrix was constructed through the Delphi expert scoring method to calculate the weight value of each evaluation index; The GA-BP neural network model is obtained by optimizing the initial weights and thresholds of the BP neural network based on the genetic algorithm; Perform dimensionality reduction on the original input data based on principal component analysis; Based on the input data after dimensionality reduction, the fire safety risk level is predicted through the trained GA-BP neural network model to obtain the fire risk level.

[0008] Optionally, a judgment matrix is ​​constructed using the Delphi expert scoring method to calculate the weight values ​​of each evaluation indicator, including: Use the scaling method to compare and score each evaluation index in pairs to construct the judgment matrix A of the evaluation index; The weight coefficient is calculated using the sum-product method; the sum-integral formula is: ; is the weight of each indicator, is the element in matrix A; Calculate the maximum eigenvalue of the judgment matrix; ; Performing a consistency check based on the maximum eigenvalue; ; ; Wherein CR is the consistency ratio and CI is the consistency index; when both CR and CI are satisfied, the weight value of each evaluation index is obtained.

[0009] Optionally, the original input data is subjected to dimensionality reduction processing based on principal component analysis, including: According to the characteristics of campus fire safety risks, key variables reflecting the fire safety status on campus are selected; Clean and organize the collected data to achieve data standardization; Measure the partial correlation between the key variables to determine whether principal component analysis is appropriate; Check whether there is a significant correlation between the key variables; Calculate the covariance matrix for key variables that meet partial correlation and significant correlation; Perform eigenvalue decomposition on the covariance matrix to obtain a series of eigenvalues ​​and corresponding eigenvectors; According to the order of eigenvalues, key variables whose cumulative contribution rate reaches a predetermined threshold are selected as principal components.

[0010] Optionally, the GA-BP neural network model is obtained by optimizing the initial weights and thresholds of the BP neural network based on a genetic algorithm, including: Generate an initialization population to generate a random solution set and initialize the weights; Iterate to find the optimal solution through selection, crossover, and mutation operations; The performance of individuals in each generation is scored according to the fitness function until the termination condition is met; Record the best individuals in each generation for setting the parameters of the neural network; The BP neural network model is constructed by combining the initial weights and thresholds optimized by the genetic algorithm.

[0011] Optionally, the method further comprises: triggering an early warning when the fire risk level reaches a set threshold; Generate evacuation routes based on the current fire risk level and campus layout, and display and voice broadcast the generated evacuation routes through smart terminals on campus; Generate evacuation routes based on current fire risk level and campus layout, including: Obtain campus layout, building location information, and emergency exit locations; Matching a corresponding priority factor based on the current crowd density in the building; the higher the priority, the greater the priority factor; Traversing all routes to obtain a route with the largest comprehensive score value based on the priority factors and the path scores; wherein the shorter the path, the higher the path score; When there are at least two routes with the largest comprehensive score values, one of the routes is selected as the evacuation route based on at least one parameter of the width and the number of turns of the path.

[0012] Optionally, traversing all routes to obtain a route with the maximum comprehensive score value based on the priority factor and the path score includes: Match the corresponding priority factors according to the current crowd density in each building on campus; For each route from the current node to the safe exit, the path score is calculated based on the route length; The comprehensive score value is calculated based on the priority factor and path score corresponding to the route; Traverse all routes in turn, compare the comprehensive score values ​​of each route, and select the route with the largest comprehensive score value.

[0013] Optionally, the method further includes: Monitor the crowd density of each building on campus in real time, and obtain the building location information of the current crowded area when the route is detected to be crowded; Generate alternate evacuation routes based on building location information in the current congested area; The backup evacuation route is displayed and voice broadcasted on a smart terminal near the congestion site.

[0014] In the second aspect, the neural network-based campus fire assessment system provided in this application adopts the following technical solution.

[0015] A campus fire evaluation system based on neural network, comprising: The first processing module is used to: construct a campus fire safety risk evaluation index system; the evaluation indicators include: system management, training management, fire protection facilities and equipment management, and key fire protection site management; The second processing module is used to: construct a judgment matrix through the Delphi expert scoring method and calculate the weight value of each evaluation index; The third processing module is used to: optimize the initial weights and thresholds of the BP neural network based on the genetic algorithm to obtain a GA-BP neural network model; The fourth processing module is used to: perform dimensionality reduction processing on the original input data based on the principal component analysis method; The fifth processing module is used to predict the fire safety risk level based on the input data after dimension reduction through the trained GA-BP neural network model to obtain the fire risk level.

[0016] In a third aspect, the present application discloses an electronic device, comprising a memory and a processor, wherein the memory stores a computer program that is loaded by the processor and executes any of the above methods.

[0017] In a fourth aspect, the present application discloses a computer-readable storage medium storing a computer program that can be loaded by a processor and execute any of the above methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a campus fire assessment method based on a neural network according to an embodiment of the present application; Figure 2 It is a system block diagram of a campus fire assessment method based on a neural network according to an embodiment of the present application; In the figure, 201 is a first processing module; 202 is a second processing module; 203 is a third processing module; 204 is a fourth processing module; 205 is a fifth processing module. DETAILED DESCRIPTION

[0019] The following is combined with Figure 1-2 The present application is further described with specific embodiments: The present application embodiment discloses a campus fire assessment method based on a neural network, comprising the following steps: Step 101, construct a campus fire safety risk evaluation index system; the evaluation indicators include: system management, training management, fire protection facilities and equipment management and key fire protection site management.

[0020] Step 102: construct a judgment matrix using the Delphi expert scoring method and calculate the weight value of each evaluation index.

[0021] Step 103: Optimize the initial weights and thresholds of the BP neural network based on the genetic algorithm to obtain a GA-BP neural network model.

[0022] Step 104: Perform dimensionality reduction processing on the original input data based on principal component analysis.

[0023] Step 105: Based on the input data after dimensionality reduction, the fire safety risk level is predicted by using the trained GA-BP neural network model to obtain the fire risk level.

[0024] Specifically, this application constructs a comprehensive campus fire safety risk evaluation index system, covering multiple aspects such as system management, training management, fire protection facilities and equipment management, and key fire protection site management. It uses the Delphi expert scoring method to construct a judgment matrix and calculate the weight value of each evaluation index; based on the genetic algorithm, the initial weights and thresholds of the BP neural network are optimized to obtain the GA-BP neural network model. The original input data is reduced in dimension by the principal component analysis method, which effectively reduces data redundancy; based on the input data after dimensionality reduction, the fire safety risk level is predicted by the trained GA-BP neural network model, which can accurately evaluate the fire safety risk level on campus and improve the accuracy and reliability of fire safety risk assessment.

[0025] As a specific implementation of a campus fire evaluation method based on a neural network, a judgment matrix is ​​constructed by using the Delphi expert scoring method to calculate the weight value of each evaluation index, including: Use the scaling method to compare and score each evaluation index in pairs to construct the judgment matrix A of the evaluation index; The weight coefficient is calculated using the sum-product method; the sum-integral formula is: ; is the weight of each indicator, is the element in matrix A; Calculate the maximum eigenvalue of the judgment matrix; ; Performing a consistency check based on the maximum eigenvalue; ; ; Wherein CR is the consistency ratio and CI is the consistency index; when both CR and CI are satisfied, the weight value of each evaluation index is obtained.

[0026] Specifically, the scaling method was used to compare and score each evaluation index in pairs, and a judgment matrix A reflecting the expert opinions was constructed; the sum-product method was used to calculate the weight coefficient and the weight of each index; the maximum eigenvalue of the judgment matrix was calculated, and a consistency test was performed based on the maximum eigenvalue, and the consistency ratio CR and consistency index CI were used to verify the consistency of the judgment matrix. When both CR and CI meet the set threshold, the weight values ​​of each evaluation index obtained not only have high reliability, but also can effectively avoid the deviation caused by differences in expert subjective judgment.

[0027] As a specific implementation of a campus fire assessment method based on a neural network, the original input data is subjected to dimensionality reduction processing based on the principal component analysis method, including: According to the characteristics of campus fire safety risks, key variables reflecting the fire safety status on campus are selected; Clean and organize the collected data to achieve data standardization; Measure the partial correlation between the key variables to determine whether principal component analysis is appropriate; Check whether there is a significant correlation between the key variables; Calculate the covariance matrix for key variables that meet partial correlation and significant correlation; Perform eigenvalue decomposition on the covariance matrix to obtain a series of eigenvalues ​​and corresponding eigenvectors; According to the order of eigenvalues, key variables whose cumulative contribution rate reaches a predetermined threshold are selected as principal components.

[0028] Specifically, according to the characteristics of campus fire safety risks, key variables that can reflect the fire safety status on campus are selected; the collected data are cleaned and sorted to achieve data standardization; by measuring the partial correlation between key variables, it is determined whether it is suitable for principal component analysis. For key variables that meet partial correlation and significant correlation, their covariance matrix is ​​calculated, and by performing eigenvalue decomposition on the covariance matrix, a series of eigenvalues ​​and corresponding eigenvectors are obtained. According to the eigenvalue size sorting, the key variables whose cumulative contribution rate reaches a predetermined threshold are selected as the main components, which not only effectively reduces the data dimension, but also retains the main information in the data. The scheme of the present application not only improves the efficiency of data processing and reduces the computational complexity, but also enhances the explanatory power and prediction accuracy of the model.

[0029] As a specific implementation of a campus fire assessment method based on a neural network, a GA-BP neural network model is obtained by optimizing the initial weights and thresholds of a BP neural network based on a genetic algorithm, including: Generate an initialization population to generate a random solution set and initialize the weights; Iterate to find the optimal solution through selection, crossover, and mutation operations; The performance of individuals in each generation is scored according to the fitness function until the termination condition is met; Record the best individuals in each generation for setting the parameters of the neural network; The BP neural network model is constructed by combining the initial weights and thresholds optimized by the genetic algorithm.

[0030] Specifically, the initial weights and thresholds of the BP neural network are optimized based on the genetic algorithm to construct a GA-BP neural network model. The population is initialized to generate a random solution set and initialize the weights. Operations such as selection, crossover, and mutation are used to iterate and gradually find the optimal solution. The performance of each generation of individuals is scored according to the fitness function; after the termination condition is met, the best individuals in each generation are recorded. These best individuals are used to set the parameters of the neural network, thereby avoiding the problem that traditional BP neural networks are prone to fall into local optimality.

[0031] As one implementation of a campus fire assessment method based on a neural network, the method further includes: triggering an early warning when the fire risk level reaches a set threshold; Generate evacuation routes based on the current fire risk level and campus layout, and display and voice broadcast the generated evacuation routes through smart terminals on campus; Generate evacuation routes based on current fire risk level and campus layout, including: Obtain campus layout, building location information, and emergency exit locations; Matching a corresponding priority factor based on the current crowd density in the building; the higher the priority, the greater the priority factor; Traversing all routes to obtain a route with the largest comprehensive score value based on the priority factors and the path scores; wherein the shorter the path, the higher the path score; When there are at least two routes with the largest comprehensive score values, one of the routes is selected as the evacuation route based on at least one parameter of the width and the number of turns of the path.

[0032] Specifically, by real-time monitoring and evaluation of the fire risk level, once the preset threshold is reached, the system immediately triggers an early warning; the system obtains the campus layout, building location information and safe exit location, and matches the corresponding priority factor in combination with the current crowd density in the building. Traverse all possible evacuation routes, and calculate the comprehensive score value based on the priority factor and path score, and select the route with the largest comprehensive score value as the optimal evacuation route. When there are multiple routes with the same comprehensive score value, further consider parameters such as path width and number of turns to select the optimal route. The evacuation route is displayed and voice broadcasted through the smart terminals on campus, providing teachers and students with intuitive and clear evacuation guidance.

[0033] As one implementation of a campus fire assessment method based on a neural network, traversing all routes to obtain a route with the largest comprehensive score value based on the priority factor and the path score includes: Match the corresponding priority factors according to the current crowd density in each building on campus; For each route from the current node to the safe exit, the path score is calculated based on the route length; The comprehensive score value is calculated based on the priority factor and path score corresponding to the route; Traverse all routes in turn, compare the comprehensive score values ​​of each route, and select the route with the largest comprehensive score value.

[0034] Specifically, the corresponding priority factors are matched according to the current crowd density in each building on campus, so that the evacuation routes in high crowd density areas have higher priorities; for each route from the current node to the safe exit, the path score is calculated according to the route length, so that shorter paths get higher scores; the comprehensive score value is calculated by combining the priority factor and path score corresponding to the route, which takes into account both the physical length of the path and dynamic factors such as crowd density. All possible routes are traversed and the comprehensive score values ​​of each route are compared, and finally the route with the largest comprehensive score value is selected as the optimal evacuation route. This method can not only find the shortest evacuation route that best meets actual needs, but also respond quickly in an emergency, ensuring the efficiency of the evacuation process and improving the reliability and effectiveness of the overall emergency evacuation.

[0035] As one implementation of a campus fire assessment method based on a neural network, the method further includes: Monitor the crowd density of each building on campus in real time, and obtain the building location information of the current crowded area when the route is detected to be crowded; Generate alternate evacuation routes based on building location information in the current congested area; The backup evacuation route is displayed and voice broadcasted on a smart terminal near the congestion site.

[0036] Specifically, the crowd density of each building on campus is monitored in real time. When a route is detected to be congested, the location information of the buildings in the current congested area is obtained, and an alternative evacuation route is generated based on this information. By displaying and voice broadcasting the alternative evacuation route on the smart terminal near the congested area, teachers and students are provided with intuitive and clear evacuation guidance, effectively avoiding evacuation problems caused by crowds.

[0037] The present application also provides a campus fire assessment system based on a neural network, comprising: The first processing module 201 is used to: construct a campus fire safety risk evaluation index system; the evaluation indexes include: system management, training management, fire protection facility and equipment management, and key fire protection site management; The second processing module 202 is used to: construct a judgment matrix through the Delphi expert scoring method and calculate the weight value of each evaluation index; The third processing module 203 is used to: optimize the initial weights and thresholds of the BP neural network based on the genetic algorithm to obtain a GA-BP neural network model; The fourth processing module 204 is used to: perform dimensionality reduction processing on the original input data based on the principal component analysis method; The fifth processing module 205 is used to predict the fire safety risk level based on the input data after dimension reduction through the trained GA-BP neural network model to obtain the fire risk level.

[0038] The embodiment of the present application also discloses an electronic device.

[0039] Specifically, the device includes a memory and a processor, and the memory stores a computer program that can be loaded by the processor and execute any of the above-mentioned neural network-based campus fire assessment methods.

[0040] The embodiment of the present application also discloses a computer-readable storage medium. Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and execute any of the above-mentioned campus fire assessment methods based on neural networks, and the computer-readable storage medium includes, for example: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0041] It should be noted that the above embodiments are only used to illustrate the present application and are not intended to limit the technical solutions described in the present application. Although the present application has been described in detail in this specification with reference to the above embodiments, a person of ordinary skill in the art should understand that a person of ordinary skill in the art can still modify or make equivalent substitutions to the present application, and all technical solutions and improvements thereof that do not depart from the spirit and scope of the present application should be included in the scope of the claims of the present application.

Claims

1. A campus fire assessment method based on neural network, characterized in that: include: Construct a campus fire safety risk assessment index system; the assessment indexes include: system management, training management, fire protection facility and equipment management, and key fire protection site management; The judgment matrix was constructed through the Delphi expert scoring method to calculate the weight value of each evaluation index; The GA-BP neural network model is obtained by optimizing the initial weights and thresholds of the BP neural network based on the genetic algorithm; Perform dimensionality reduction on the original input data based on principal component analysis; Based on the input data after dimensionality reduction, the fire safety risk level is predicted through the trained GA-BP neural network model to obtain the fire risk level.

2. A campus fire assessment method based on neural network according to claim 1, characterized in that: The judgment matrix was constructed by the Delphi expert scoring method to calculate the weight value of each evaluation index, including: Use the scaling method to compare and score each evaluation index in pairs to construct the judgment matrix A of the evaluation index; The weight coefficient is calculated using the sum-product method; the sum-integral formula is: ; is the weight of each indicator, is the element in matrix A; Calculate the maximum eigenvalue of the judgment matrix; ; Performing a consistency check based on the maximum eigenvalue; ; ; Wherein CR is the consistency ratio and CI is the consistency index; when both CR and CI are satisfied, the weight value of each evaluation index is obtained.

3. A campus fire assessment method based on neural network according to claim 2, characterized in that: The original input data is processed for dimensionality reduction based on principal component analysis, including: According to the characteristics of campus fire safety risks, key variables reflecting the fire safety status on campus are selected; Clean and organize the collected data to achieve data standardization; Measure the partial correlation between the key variables to determine whether principal component analysis is appropriate; Check whether there is a significant correlation between the key variables; Calculate the covariance matrix for key variables that meet partial correlation and significant correlation; Perform eigenvalue decomposition on the covariance matrix to obtain a series of eigenvalues ​​and corresponding eigenvectors; According to the order of eigenvalues, key variables whose cumulative contribution rate reaches a predetermined threshold are selected as principal components.

4. A campus fire assessment method based on neural network according to claim 3, characterized in that: The GA-BP neural network model is obtained by optimizing the initial weights and thresholds of the BP neural network based on the genetic algorithm, including: Generate an initialization population to generate a random solution set and initialize the weights; Iterate to find the optimal solution through selection, crossover, and mutation operations; The performance of individuals in each generation is scored according to the fitness function until the termination condition is met; Record the best individuals in each generation for setting the parameters of the neural network; The BP neural network model is constructed by combining the initial weights and thresholds optimized by the genetic algorithm.

5. A campus fire assessment method based on neural network according to claim 4, characterized in that: The method further includes: triggering an early warning when the fire risk level reaches a set threshold; Generate evacuation routes based on the current fire risk level and campus layout, and display and voice broadcast the generated evacuation routes through smart terminals on campus; Generate evacuation routes based on current fire risk level and campus layout, including: Obtain campus layout, building location information, and emergency exit locations; Matching a corresponding priority factor based on the current crowd density in the building; the higher the priority, the greater the priority factor; Traversing all routes to obtain a route with the largest comprehensive score value based on the priority factors and the path scores; wherein the shorter the path, the higher the path score; When there are at least two routes with the largest comprehensive score values, one of the routes is selected as the evacuation route based on at least one parameter of the width and the number of turns of the path.

6. A campus fire assessment method based on neural network according to claim 5, characterized in that: Traverse all routes based on the priority factors and path scores to obtain the route with the largest comprehensive score value, including: Match the corresponding priority factors according to the current crowd density in each building on campus; For each route from the current node to the safe exit, the path score is calculated based on the route length; The comprehensive score value is calculated based on the priority factor and path score corresponding to the route; Traverse all routes in turn, compare the comprehensive score values ​​of each route, and select the route with the largest comprehensive score value.

7. A campus fire assessment method based on neural network according to claim 6, characterized in that: The method further comprises: Monitor the crowd density of each building on campus in real time, and obtain the building location information of the current crowded area when the route is detected to be crowded; Generate alternate evacuation routes based on building location information in the current congested area; The backup evacuation route is displayed and voice broadcasted on a smart terminal near the congestion site.

8. A campus fire assessment system based on neural network, characterized in that: include: The first processing module is used to: construct a campus fire safety risk evaluation index system; the evaluation indicators include: system management, training management, fire protection facilities and equipment management, and key fire protection site management; The second processing module is used to: construct a judgment matrix through the Delphi expert scoring method and calculate the weight value of each evaluation index; The third processing module is used to: optimize the initial weights and thresholds of the BP neural network based on the genetic algorithm to obtain a GA-BP neural network model; The fourth processing module is used to: perform dimensionality reduction processing on the original input data based on the principal component analysis method; The fifth processing module is used to predict the fire safety risk level based on the input data after dimension reduction through the trained GA-BP neural network model to obtain the fire risk level.

Images