Power distribution fire control system based on biological inspiration
By adopting biologically inspired methods in the power distribution fire control system, data is collected and analyzed in real time, and risk assessment models and control strategy optimization systems are built, the existing system is solved in the problem of untimely response in the face of complex fires and electrical faults, and more efficient data processing and intelligent decision-making capabilities are achieved.
Patent Information
- Application Number
- CN202510229098.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-28
- Publication Date
- 2025-05-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing distribution fire control systems lack the ability to handle complex and dynamic environmental changes, and it is difficult to make comprehensive and timely responses in the face of complex fire situations and multiple electrical faults. The system's learning and adaptability are limited, and it has failed to effectively form intelligent data-based identification and decision-making capabilities.
The distribution fire control system based on biological inspiration is adopted, including data acquisition and processing module, biological aspiration selection module, distribution fire classification module and bionic evaluation control module. Data is collected in real time through sensor networks, and data analysis and feature extraction is used for biological heuristic algorithms and convolutional neural networks to build a risk assessment model and control strategy optimization system.
It significantly improves the data collection and processing capabilities of the distribution fire control system, improves the identification and early warning capabilities of fire and electrical faults, enhances the adaptability and intelligence of the system, and can achieve accurate classification and prediction in complex environments.
Smart Images
Figure CN119925867A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power system automation and fire safety, and in particular to a power distribution fire control system based on bioinspiration. Background Art
[0002] In modern society, with the deepening of urbanization and industrialization, the use of electricity is becoming more and more popular, and the distribution network has become particularly important. As an important link in protecting the safety of people and property, distribution fire control is mainly responsible for monitoring fires and electrical faults and making corresponding responses. The current distribution fire control usually adopts a threshold-based detection mechanism, that is, when the reading of a certain sensor exceeds the preset safety range, an alarm is triggered and corresponding control measures are executed. For example, when the temperature monitored by the temperature sensor exceeds the set value, the fire extinguishing system is quickly started and an alarm message is sent, which can help to achieve a quick response in the early stage of a fire and ensure the safety of people and equipment.
[0003] The existing technology has the following deficiencies: the existing power distribution fire control technology mainly relies on pre-set static rules and lacks the ability to handle complex and dynamic environmental changes. In terms of fire safety issues, the causes and characteristics of fires vary greatly and involve the combined effects of multiple factors. Therefore, a single sensor information and rule cannot cover all fire risks, resulting in difficulty in making a comprehensive and timely response when facing complex fire situations and multiple electrical faults. The existing power distribution fire control system lacks advanced data processing capabilities and faces problems of information redundancy and data islands when acquiring data. It is unable to fully utilize the massive data generated by sensors for in-depth analysis. Accordingly, the system's learning and adaptability are limited, and it fails to effectively form data-based intelligent recognition and decision-making capabilities. When encountering nonlinear and dynamically changing scenarios, the response speed and judgment accuracy are often unsatisfactory.
[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention
[0005] The purpose of the present invention is to provide a bio-inspired power distribution fire control system to solve the problems in the above-mentioned background technology.
[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a bio-inspired power distribution fire control system, comprising a data acquisition and processing module, a bio-inspired selection module, a power distribution fire classification module and a bionic evaluation control module; Data acquisition and processing module: collects relevant data of fire power distribution in real time through the sensor network, and performs validity check, abnormal value processing and feature extraction on the relevant data of fire power distribution; Bio-inspired selection module: Analyze power equipment data in real time based on bio-inspired algorithms to select the best solution for fire occurrence and distribution equipment failure data; Power distribution fire classification module: Analyze the optimal fitness data corresponding to fire occurrence and power distribution equipment failure through convolutional neural network, and output the prediction of the current fire occurrence and power distribution equipment failure category; Bionic evaluation control module: Evaluate the performance of fire and electrical fault monitoring through the fitness function, select cells with high fitness to replicate and adjust parameters, and iteratively optimize to select the best control strategy as the optimal solution.
[0007] Preferably, the wireless nodes of the temperature sensor, humidity sensor, smoke sensor and current sensor are used as separate nodes, and data is sent to the aggregation node through wireless signals and received by the central server and stored in the cloud database, wherein the temperature sensor for real-time monitoring of the fire power distribution equipment and the environmental temperature data is installed on the outer shell surface, electrical contacts and air circulation position of the fire power distribution equipment, the humidity sensor for real-time monitoring of the fire power distribution equipment and the environmental humidity data is installed on the indoor wall and high humidity area where the fire power distribution equipment is stored, the smoke sensor for real-time monitoring of the smoke concentration change data in the fire power distribution environment is installed on the top area where the fire power distribution equipment is stored, and the current sensor for real-time detection of the operating status data of the fire power distribution equipment is installed on the line inlet and outlet port of the fire power distribution equipment, and the real-time monitored data is checked for null values and missing values, and whether the real-time monitored data is within the effective range of the temperature sensor, humidity sensor, smoke sensor and current sensor. The real-time monitored data is checked for abnormal values and duplicate values through the constraints of uniqueness and consistency, and different sensor partitions are divided according to different existence situations, and the log object corresponding to each sensor partition is deleted, and the real-time monitored data is filled using the mean filling method to extract the temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics.
[0008] Preferably, each real-time monitoring data is represented as an individual binary string, and multiple binary strings are used to form the entire gene string. A risk assessment model is established to define the quantitative assessment score of each real-time monitoring data for fire occurrence and distribution equipment failure. The specific calculation formula is:
[0009] in Represents the quantitative evaluation score, It represents the score of the fire occurrence and distribution equipment failure i in the pth real-time monitoring data, Indicates the weight of the real-time monitoring data, It represents the total number of data monitored in real time. According to the data characteristics of fire occurrence and power distribution equipment failure, a greedy algorithm is designed in combination with random generation to generate the initial population. The best fitness function of the data of fire occurrence and power distribution equipment failure is defined by the greedy algorithm. The best solution of the data of fire occurrence and power distribution equipment failure that can improve the value of the best fitness function of the data is selected based on the gene string. The best fitness data corresponding to the fire occurrence and power distribution equipment failure is generated step by step and added back to the top of the population.
[0010] Preferably, a convolutional neural network is constructed and the optimal fitness data corresponding to the occurrence of fire and the failure of power distribution equipment is received through the input layer, a convolutional layer is added, and multiple convolution kernels are used to perform sliding local perception on the optimal fitness data corresponding to the occurrence of fire and the failure of power distribution equipment. The temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics are applied, an activation layer is added, ReLU is used as the activation function and nonlinearity is introduced. The specific formula is:
[0011] in represents the feature output of the convolution operation, Indicates the characteristics of fire and power distribution equipment failure. Indicates the weight values of temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics, Respectively represent the row and column indices inside the convolution kernel, The row and column indices represent the temperature change rate, humidity change rate, smoke concentration peak, and current fluctuation characteristics, respectively. The feature output of the convolution operation is downsampled through the pooling layer. The convolution and pooling operations are repeated to flatten each feature output into a one-dimensional vector and connected to the fully connected layer for classification. The specific formula is:
[0012] in, Indicates the output fire occurrence and distribution equipment failure category for a given characteristic output Q is the probability of g, represents the classification activation function value, It represents the degree value of the current fire occurrence and the distribution equipment fault category, and outputs the current fire occurrence and the distribution equipment fault category through each neuron of the fully connected layer connecting all neurons of the convolution layer, activation layer and pooling layer.
[0013] Preferably, a group of immune cells representing the current fire occurrence and distribution equipment failure control strategy is initialized, and the "antibody" parameters of each immune cell represent different control parameters of the control strategy. According to the probability of outputting the fire occurrence and distribution equipment failure category, combined with the actual monitoring of the current fire occurrence and distribution equipment failure category and the difference between the ideal fire and electrical safety values, the performance of each immune cell is evaluated through the fitness function. The specific formula is:
[0014] in, Represents the performance fitness value of each immune cell, represents the probability of fire occurrence category, Indicates the fire safety value, represents the probability of the electrical failure category, It represents the safety value of the electrical appliance. The immune cells with higher fitness values are selected to be replicated to the next generation. The parameters of the modified immune cells are introduced to generate a new generation of immune cells. During the iteration process, the immune cells with the highest fitness are selected as the optimal solution and the algorithm is stopped.
[0015] In the above technical solution, the technical effects and advantages provided by the present invention are: 1. By using the wireless nodes of temperature, humidity, smoke and current sensors as independent modules, using wireless signals to send real-time monitoring data to the aggregation node, and storing the data in the cloud database through the central server, the data collection and processing capabilities of the power distribution fire control system have been significantly improved. By installing sensors at key locations of fire distribution equipment, the system can monitor the ambient temperature, humidity, smoke concentration and current status in real time, thereby achieving comprehensive monitoring of fire and electrical faults. This not only improves the breadth and accuracy of data collection, but also enhances the real-time response capability of the system, and can promptly discover potential safety hazards. Strict quality checks, including the detection of null values, missing values, outliers and duplicate values, ensure the integrity and reliability of the data. The mean filling method is used to process missing data and extract the temperature change rate, humidity change rate, smoke concentration peak and current fluctuation characteristics, which can effectively improve the prediction accuracy of the model. By representing each real-time monitored data as a binary string and using the greedy algorithm to generate the initial population, a risk assessment model is constructed, which realizes the quantitative assessment of fire occurrence and distribution equipment failure. It can automatically identify and optimize data, and gradually iterate to generate the best fitness data, thereby improving the system's ability to identify and warn of fire and electrical failures.
[0016] 2. By constructing a convolutional neural network and using the input layer to receive the best fitness data, using the convolution layer and pooling layer for feature extraction, and combining the activation layer to introduce nonlinearity, the learning ability and classification accuracy of the convolutional neural network are further improved. The design of the fully connected layer enables the system to effectively map the extracted features to the specific categories of fire and electrical failure, thereby achieving accurate classification and prediction. It performs better than the traditional rule-based system in complex environments and can better adapt to the ever-changing fire and electrical failure scenarios. An immune algorithm is used to initialize a group of immune cells representing the control strategy, and the performance of each cell is evaluated according to the fitness function, which can optimize the control strategy. By selecting cells with high fitness for replication and parameter adjustment, the system can continuously evolve and find the best control strategy, thereby improving the system's adaptability and intelligence level. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0018] Figure 1 This is a method flow chart of the bio-inspired power distribution fire control system of the present invention.
[0019] Figure 2 Schematic diagram of the module of the bio-inspired power distribution fire control system of the present invention DETAILED DESCRIPTION
[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.
[0021] The present invention provides Figure 2 The bio-inspired power distribution fire control system shown includes a data acquisition and processing module, a bio-inspired selection module, a power distribution fire classification module, and a bionic evaluation and control module; Data acquisition and processing module: collects relevant data of fire power distribution in real time through the sensor network, and performs validity check, abnormal value processing and feature extraction on the relevant data of fire power distribution; The wireless nodes of temperature sensor, humidity sensor, smoke sensor and current sensor are used as separate nodes, and data is sent to the aggregation node through wireless signals. The data is received by the central server and stored in the cloud database. The temperature sensor used for real-time monitoring of fire power distribution equipment and environmental temperature data is installed on the shell surface, electrical contacts and air circulation position of the fire power distribution equipment. The humidity sensor used for real-time monitoring of fire power distribution equipment and environmental humidity data is installed on the indoor wall and high humidity area where the fire power distribution equipment is stored. The smoke sensor used for real-time monitoring of smoke concentration change data in the fire power distribution environment is installed on the top area where the fire power distribution equipment is stored. The current sensor used for real-time detection of the operating status data of the fire power distribution equipment is installed on the line inlet and outlet ports of the fire power distribution equipment. Check whether there are null values and missing values in the real-time monitored data. Check whether the real-time monitored data is within the effective range of the temperature sensor, humidity sensor, smoke sensor and current sensor. Check whether there are abnormal values and duplicate values in the real-time monitored data through the constraints of uniqueness and consistency. Different sensor partitions are divided according to different existence situations, and the log object corresponding to each sensor partition is deleted. The real-time monitored data is filled using the mean filling method to extract the temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics.
[0022] In this embodiment, what needs to be specifically explained is the data acquisition and processing module, which ensures the accuracy and reliability of real-time monitoring data through efficient sensor configuration, wireless data transmission, strict data quality inspection and feature extraction, provides a solid foundation for early warning of fire and electrical failure, and helps to improve the overall safety and intelligence level of the distribution fire control system.
[0023] Bio-inspired selection module: Analyze power equipment data in real time based on bio-inspired algorithms to select the best solution for fire occurrence and distribution equipment failure data; Each real-time monitoring data is represented as an individual binary string. The binary number 0000 represents the fire power distribution equipment and environmental temperature data, 0001 represents the fire power distribution equipment and environmental humidity data, 0010 represents the smoke concentration change data in the fire power distribution environment, and 0011 represents the fire power distribution equipment operation status data. Multiple binary strings are used to form the entire gene string to establish a risk assessment model and define the quantitative assessment score of each real-time monitoring data for fire occurrence and power distribution equipment failure. The specific calculation formula is:
[0024] in Represents the quantitative evaluation score, It represents the score of the fire occurrence and distribution equipment failure i in the pth real-time monitoring data, Indicates the weight of the real-time monitoring data, It represents the total number of data monitored in real time. According to the data characteristics of fire occurrence and power distribution equipment failure, a greedy algorithm is designed in combination with random generation to generate the initial population. The best fitness function of the data of fire occurrence and power distribution equipment failure is defined by the greedy algorithm. The best solution of the data of fire occurrence and power distribution equipment failure that can improve the value of the best fitness function of the data is selected based on the gene string. The best fitness data corresponding to the fire occurrence and power distribution equipment failure is generated step by step and added back to the top of the population.
[0025] In this embodiment, what needs to be specifically explained is the biological inspiration selection module, which encodes the monitoring data into a binary string and optimizes the data in combination with the greedy algorithm and the genetic algorithm, so as to achieve quantitative evaluation of fire and electrical failure, which not only improves the accuracy of the risk assessment model, but also enhances the system's adaptability to environmental changes, and provides strong support for real-time monitoring and early warning. Assuming that the score of fire occurrence and power distribution equipment failure in the fire distribution equipment and ambient temperature data is 0.9, its weight is 0.4, the score of fire occurrence and power distribution equipment failure in the fire distribution equipment and ambient humidity data is 0.85, its weight is 0.3, the score of fire occurrence and power distribution equipment failure in the smoke concentration change data in the fire distribution environment is 0.95, its weight is 0.2, the score of fire occurrence and power distribution equipment failure in the fire distribution equipment operating status data is 0.8, its weight is 0.1, then the quantitative evaluation score is 4.5, so as to provide an intuitive risk indication of fire occurrence and power distribution equipment failure.
[0026] Power distribution fire classification module: Analyze the optimal fitness data corresponding to fire occurrence and power distribution equipment failure through convolutional neural network, and output the prediction of the current fire occurrence and power distribution equipment failure category; Construct a convolutional neural network and receive the best fitness data corresponding to the occurrence of fire and distribution equipment failure through the input layer. Add a convolution layer, use multiple convolution kernels to perform sliding local perception on the best fitness data corresponding to the occurrence of fire and distribution equipment failure, apply the temperature change rate, humidity change rate, smoke concentration peak and current fluctuation characteristics, add an activation layer, use ReLU as the activation function and introduce nonlinearity. The specific formula is:
[0027] in represents the feature output of the convolution operation, Indicates the characteristics of fire and power distribution equipment failure. Indicates the weight values of temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics, Respectively represent the row and column indices inside the convolution kernel, The row and column indices represent the temperature change rate, humidity change rate, smoke concentration peak, and current fluctuation characteristics, respectively. The feature output of the convolution operation is downsampled through the pooling layer. The convolution and pooling operations are repeated to flatten each feature output into a one-dimensional vector and connected to the fully connected layer for classification. The specific formula is:
[0028] in, Indicates the output fire occurrence and distribution equipment failure category for a given characteristic output Q is the probability of g, represents the classification activation function value, It represents the degree value of the current fire occurrence and the distribution equipment fault category, and outputs the current fire occurrence and the distribution equipment fault category through each neuron of the fully connected layer connecting all neurons of the convolution layer, activation layer and pooling layer.
[0029] In this embodiment, what needs to be specifically explained is the power distribution fire classification module, which uses a convolutional neural network to efficiently detect fire occurrences and power distribution equipment failures and capture important features in real time, which not only improves monitoring accuracy, but also provides intelligent and automated support for equipment safety management, helps reduce the risks of equipment failures and fire accidents, and improves the overall safety level.
[0030] Bionic evaluation control module: evaluates the performance of fire and electrical fault monitoring through fitness function, selects cells with high fitness to replicate and adjust parameters, and iteratively optimizes and selects the best control strategy as the optimal solution; Initialize a group of immune cells representing the current fire occurrence and distribution equipment failure control strategy, and use the "antibody" parameters of each immune cell to represent different control parameters of the control strategy. According to the probability of outputting the fire occurrence and distribution equipment failure category, combined with the actual monitoring of the current fire occurrence and distribution equipment failure category and the difference between the ideal fire and electrical safety values, the performance of each immune cell is evaluated through the fitness function. The specific formula is:
[0031] in, Represents the performance fitness value of each immune cell, represents the probability of fire occurrence category, Indicates the fire safety value, represents the probability of failure of the power distribution equipment, It represents the safety value of the distribution equipment, which reflects the quality of the current control strategy. The smaller the error, the higher the fitness value. The immune cells with higher fitness values are selected to be replicated to the next generation, and the parameters of the modified immune cells are introduced to generate a new generation of immune cells. During the iteration process, the immune cells with the highest fitness are selected as the optimal solution and the algorithm is stopped.
[0032] In this embodiment, what needs to be specifically explained is the bionic evaluation control module. It is assumed that the probability of the fire occurrence category in the system is 0.8 and the fire safety value is 0.75; the probability of the current distribution equipment failure is 0.2, and the distribution equipment safety value is 0.1. The fitness value of the current fire occurrence and distribution equipment failure control strategy is calculated through the fitness function, and the control strategy is optimized in the immune algorithm to minimize the error between the predicted value and the actual value, thereby ensuring that the control strategy for fire occurrence and distribution equipment failure achieves the optimal performance.
[0033] By using the wireless nodes of temperature, humidity, smoke and current sensors as independent modules, using wireless signals to send real-time monitoring data to the aggregation node, and storing the data in the cloud database through the central server, the data collection and processing capabilities of the distribution fire control system are significantly improved. By installing sensors at key locations of fire distribution equipment, the system can monitor the ambient temperature, humidity, smoke concentration and current status in real time, thereby realizing comprehensive monitoring of fire and electrical faults. This not only improves the breadth and accuracy of data collection, but also enhances the real-time response capability of the system, and can timely discover potential safety hazards. By conducting strict quality checks on real-time monitoring data, including the detection of null values, missing values, outliers and duplicate values, the integrity and reliability of the data are ensured. The missing data are processed using the mean filling method, and the temperature change rate, humidity change rate, smoke concentration peak and current fluctuation characteristics are extracted, which can effectively improve the prediction accuracy of the model. By representing each real-time monitoring data as a binary string and using the greedy algorithm to generate the initial population, a risk assessment model is constructed, which realizes the quantitative evaluation of fire occurrence and distribution equipment failure. The data can be automatically identified and optimized, and the optimal fitness data is gradually iterated to improve the system's identification and early warning capabilities for fire and electrical faults.
[0034] By constructing a convolutional neural network and using the input layer to receive the best fitness data, using the convolution layer and pooling layer for feature extraction, and combining the activation layer to introduce nonlinearity, the learning ability and classification accuracy of the convolutional neural network are further improved. The design of the fully connected layer enables the system to effectively map the extracted features to the specific categories of fire and electrical failure, thereby achieving accurate classification and prediction. It performs better than the traditional rule-based system in complex environments and can better adapt to the ever-changing fire and electrical failure scenarios. An immune algorithm is used to initialize a group of immune cells representing the control strategy, and the performance of each cell is evaluated according to the fitness function, which can optimize the control strategy. By selecting cells with high fitness for replication and parameter adjustment, the system can continuously evolve and find the best control strategy, thereby improving the system's adaptability and intelligence level.
[0035] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. A bio-inspired power distribution fire control system, characterized in that: It includes data acquisition and processing module, biological inspiration selection module, power distribution and fire protection classification module and bionic evaluation and control module; Data acquisition and processing module: collects relevant data of fire power distribution in real time through the sensor network, and performs validity check, abnormal value processing and feature extraction on the relevant data of fire power distribution; Bio-inspired selection module: Analyze power equipment data in real time based on bio-inspired algorithms to select the best solution for fire occurrence and distribution equipment failure data; Power distribution fire classification module: Analyze the optimal fitness data corresponding to fire occurrence and power distribution equipment failure through convolutional neural network, and output the prediction of the current fire occurrence and power distribution equipment failure category; Bionic evaluation control module: Evaluate the performance of fire and electrical fault monitoring through the fitness function, select cells with high fitness to replicate and adjust parameters, and iteratively optimize to select the best control strategy as the optimal solution.
2. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the data acquisition and processing module, the sensor network includes: a temperature sensor for real-time monitoring of fire power distribution equipment and environmental temperature data, a humidity sensor for real-time monitoring of fire power distribution equipment and environmental humidity data, a smoke sensor for real-time monitoring of smoke concentration change data in the fire power distribution environment, and a current sensor for real-time detection of fire power distribution equipment operating status data.
3. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the data acquisition and processing module, the specific steps of validity check and outlier processing are: check whether there are null values and missing values in the real-time monitoring data, check whether the real-time monitoring data is within the effective range of the temperature sensor, humidity sensor, smoke sensor and current sensor, check whether there are outliers and duplicate values in the real-time monitoring data through uniqueness and consistency constraints, divide different sensor partitions according to different existence situations, delete the log object corresponding to each sensor partition, and fill the real-time monitoring data with the mean filling method.
4. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the biological inspiration selection module, a quantitative evaluation score of each real-time monitoring data for fire occurrence and power distribution equipment failure is defined, and the specific calculation formula is: ； in Represents the quantitative evaluation score, It represents the score of the fire occurrence and distribution equipment failure i in the pth real-time monitoring data, Indicates the weight of the real-time monitoring data, Indicates the total amount of data monitored in real time.
5. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the biologically inspired selection module, the specific steps for selecting the best solution for the data of fire occurrence and power distribution equipment failure are as follows: according to the data characteristics of fire occurrence and power distribution equipment failure combined with random generation, a greedy algorithm is designed to generate an initial population, the best fitness function for the data of fire occurrence and power distribution equipment failure is defined by the greedy algorithm, and the best solution for the data of fire occurrence and power distribution equipment failure that can improve the value of the best fitness function of the data is selected based on the gene string, and the best fitness data corresponding to the fire occurrence and power distribution equipment failure is generated step by step and iteratively, and added and returned to the top of the population.
6. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the power distribution fire classification module, the specific steps of the convolutional neural network are: constructing a convolutional neural network and receiving the best fitness data corresponding to the fire occurrence and the power distribution equipment failure through the input layer, adding a convolutional layer, using multiple convolution kernels to perform sliding local perception on the best fitness data corresponding to the fire occurrence and the power distribution equipment failure, applying the temperature change rate, humidity change rate, smoke concentration peak and current fluctuation characteristics, adding an activation layer, using ReLU as the activation function and introducing nonlinearity, downsampling the feature output of the convolution operation through the pooling layer, repeating the iterative convolution and pooling operations to flatten each feature output into a one-dimensional vector, and connecting to the fully connected layer for classification, and connecting all neurons of the convolution layer, activation layer and pooling layer through each neuron of the fully connected layer to output the current fire occurrence and power distribution equipment failure category.
7. The bio-inspired power distribution fire control system according to claim 6, characterized in that: The specific formula of ReLU as an activation function is: ； in represents the feature output of the convolution operation, Indicates the characteristics of fire and power distribution equipment failure. Indicates the weight values of temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics, Respectively represent the row and column indices inside the convolution kernel, The indexes of the rows and columns representing the temperature change rate, humidity change rate, smoke concentration peak value and current fluctuation characteristics respectively, and the specific formula for classification by the fully connected layer is: ； in, Indicates the output fire occurrence and distribution equipment failure category for a given characteristic output Q is the probability of g, represents the classification activation function value, Indicates the current severity of fire and power distribution equipment failure. Indicates the current severity of fire and power distribution equipment failure.
8. The bio-inspired power distribution fire control system according to claim 1, characterized in that: In the bionic evaluation control module, the specific formula of the fitness function is: ； in, Represents the performance fitness value of each immune cell, represents the probability of fire occurrence category, Indicates the fire safety value, represents the probability of the electrical failure category, Indicates the electrical safety value.