Communication support method for emergency command and disposal based on neural network

Abstract

The present invention relates to the field of communication guarantee for emergency command and disposal, and specifically to a communication guarantee method for emergency command and disposal based on a neural network. The method includes establishing a multi-level redundant communication network to provide guarantee for the emergency site; constructing a neural network model, using the convolutional neural network (CNN) architecture to adapt to the characteristics of emergency communication data; acquiring communication data in real time and performing feature extraction and screening; using a trained CNN model to evaluate the current communication quality and predict future trends; and automatically allocating communication resources based on the evaluation and prediction results. This method can provide reliable communication guarantee in the case of "three disconnections" after a major disaster, automatically extract data features, improve the accuracy of communication quality evaluation and prediction, provide a reliable basis for decision-making, improve the efficiency and effect of emergency command and disposal, reduce losses, and protect people's lives and property.

Description

Technical Field

[0001] The present invention relates to the field of communication support for emergency command and disposal, and in particular to a communication support method for emergency command and disposal based on a neural network. Background Art

[0002] In emergency command and disposal, it is crucial to achieve two-way audio and video interconnection between the emergency response site and the emergency command center. However, in actual situations, we often encounter the "three disconnections" after a major disaster, namely, circuit disconnection, power outage, and network disconnection, which leads to poor communication and seriously affects the efficiency of emergency response command.

[0003] Existing communication guarantee methods often show many shortcomings when facing such complex situations. They are difficult to effectively adapt to the complex characteristics of emergency communication data, and cannot accurately evaluate and predict the changing trend of communication quality. This makes the allocation of communication resources not reasonable and efficient, and cannot meet the urgent needs of emergency command and disposal. Therefore, in order to effectively solve the problems of poor communication in the "three cuts" situation after a major disaster and improve the efficiency of emergency disposal and command, there is an urgent need for a communication guarantee method for emergency command and disposal based on a neural network, which can intelligently evaluate and predict communication quality, reasonably allocate communication resources, and effectively improve the efficiency and effectiveness of emergency disposal and command. Summary of the invention

[0004] In view of the technical problems existing in the above-mentioned existing communication guarantee methods in the face of communication restoration after a major disaster, the present invention proposes an artificial intelligence technology for rationally allocating communication resources and effectively improving the efficiency and effectiveness of emergency response command.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is a communication guarantee method for emergency command and disposal based on a neural network, comprising the following steps:

[0006] S1. First, establish a multi-level redundant communication network to provide temporary communication support for the emergency site;

[0007] S2, then build a neural network model, collect historical communication data and use the convolutional neural network (CNN) architecture to adapt to the complex characteristics of emergency communication data;

[0008] S3, secondly, obtain the communication data of the emergency scene in real time through sensors and communication equipment, and perform feature extraction and screening;

[0009] S4, then use the trained convolutional neural network (CNN) model to evaluate the current communication quality, and comprehensively consider multiple indicators to predict the change trend of communication quality in the future;

[0010] S5. Finally, based on the communication quality evaluation and prediction results, communication resources are automatically allocated to effectively improve the efficiency and effectiveness of emergency response command;

[0011] The process of using the trained convolutional neural network CNN model in step S4 is:

[0012] S41. First, input the data into the CNN model and perform forward propagation calculations. Among them, y is the output of the model, X is the input data, is the weight parameter of the model;

[0013] S42, secondly, generating a prediction, outputting the evaluation result of the current communication quality including the prediction value and the related confidence and probability distribution;

[0014] S43, then comprehensively consider multiple indicators to predict future communication quality, use sliding window technology to predict future time periods, slide the window forward and continuously update the prediction results in, is the predicted value at a future time point. is the current forecast value, It is characteristic of the present moment;

[0015] S44, the final comprehensive evaluation analysis, combining multiple evaluation indicators, comprehensively evaluates the communication quality and calculates the weighted average in, is the weight of each indicator, It is the indicator value; in addition, the forecast results and historical data are used for trend analysis to observe the changing trend of future communication quality. in, is the predicted value of the minister in the future N time. is the predicted value at the current time.

[0016] Preferably, the step S1 of establishing a multi-level redundant communication network includes deploying satellite communication equipment at the emergency site and the command center, using wireless ad hoc network technology to quickly establish a local communication network and deploy emergency communication vehicles at the emergency site, equipped with a variety of communication means including satellite communication and radio, as mobile communication nodes, to provide temporary communication support for the emergency site.

[0017] Preferably, the main operations of constructing the convolutional neural network (CNN) model in step S2 are as follows:

[0018] S21, first is the convolution layer, which extracts the local features of the data. The convolution operation formula is in, is the bias of the kth convolution kernel, is the value of the output feature at position (i, j) and the kth channel, is the weight of the convolution kernel at position (m,n) and the kth channel, is the value of the input data at position (i+m,j+n);

[0019] S22, followed by the activation function, uses Leaky ReLU, which allows the gradient of negative input values ​​to exist. The formula is Where x is the input value, is the negative slope coefficient, which is 0.01;

[0020] S23, followed by the pooling layer, downsampling to reduce the data dimension, using maximum pooling Among them, s is the stride, which determines the sliding distance of the pooling window on the input feature map;

[0021] S24, followed by a fully connected layer, is used to map the flattened feature vector to the target space in, is the i-th element of the input feature vector, is the weight of the jth node, is the bias of the jth output node;

[0022] S25, then train and optimize, use Adam to optimize model parameters, adjust convolution kernel size, stride, pooling window size, and learning rate hyperparameters to obtain the best performance;

[0023] S26. Finally, the cross-validation method is used to evaluate the generalization ability of the model.

[0024] Preferably, in S3, the communication data of the emergency site obtained in real time through sensors and communication equipment include signal strength, delay, bit error rate, data throughput, network jitter and packet loss rate.

[0025] Preferably, the feature extraction in step S3 includes time domain features, extracting statistical features such as mean, standard deviation, skewness and kurtosis to capture the basic trends and fluctuations of the data; frequency domain features, calculating frequency domain features through Fourier transform and power spectral density to reveal the periodicity and frequency distribution in the data; and time-frequency features, applying short-time Fourier transform to extract time-frequency features to capture the time and frequency changes in the data.

[0026] Preferably, the feature screening in step S3 includes correlation analysis, calculating the correlation between features, removing redundant features to reduce the complexity of data; recursive feature elimination, gradually eliminating unimportant features based on the importance score of the model, and optimizing the feature subset.

[0027] Compared with the prior art, the advantages and positive effects of the present invention are:

[0028] 1. Establish a multi-level redundant communication network that can provide reliable temporary communication support for the emergency site in the event of "three cuts" after a major disaster, and ensure two-way interconnection of audio and video between the emergency response site and the emergency command center.

[0029] 2. The convolutional neural network (CNN) architecture can automatically extract local features of emergency communication data, effectively adapt to its complex characteristics, and improve the accuracy of communication quality assessment and prediction.

[0030] 3. Real-time acquisition and analysis of various communication data at the emergency site, comprehensive feature extraction and screening, providing a more reliable basis for communication support decision-making, and helping to improve the efficiency and effectiveness of emergency command and disposal. DETAILED DESCRIPTION

[0031] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0032] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0033] Embodiment: After a disaster, the "three breaks" often occur, namely, circuit break, power failure, and network failure, which leads to poor communication and seriously affects the efficiency of emergency response and command. Existing communication guarantee methods often show many shortcomings when facing such complex situations. Therefore, the present invention proposes a communication guarantee method for emergency command and disposal based on neural networks. First, a multi-level redundant communication network is established, including satellite communication equipment, wireless ad hoc networks, and emergency communication vehicles. Satellite communication equipment is deployed at the emergency site and command center. Satellite communication does not rely on ground infrastructure and can provide stable communication in the event of circuit break, power failure, and network failure; wireless ad hoc network technology is used to quickly establish a local communication network at the emergency site. Ad hoc networks can realize direct communication between devices without fixed infrastructure; emergency communication vehicles are deployed, equipped with a variety of communication means, which can be used as mobile communication nodes to provide temporary communication guarantees for emergency sites.

[0034] Considering the complex characteristics of communication data, the present invention uses a convolutional neural network (CNN) architecture to achieve this goal. The convolutional neural network can automatically extract local features of the data, effectively capture complex patterns and changes in emergency communication data, and process large-scale data to adapt to the large amount and diversity of data in emergency communications. The main operation of constructing a convolutional neural network (CNN) model is first the convolution layer to extract local features of the data. The convolution operation formula is: in, is the bias of the kth convolution kernel, is the value of the output feature at position (i, j) and the kth channel, is the weight of the convolution kernel at position (m,n) and the kth channel, is the value of the input data at position (i+m,j+n); secondly, the activation function uses Leaky ReLU, which allows the gradient of negative input values ​​to exist, and the formula is Where x is the input value, is the negative slope coefficient, which is 0.01; followed by the pooling layer, downsampling to reduce the data dimension, using the maximum pooling Among them, s is the stride, which determines the sliding distance of the pooling window on the input feature map; then the fully connected layer is used to map the flattened feature vector to the target space in, is the i-th element of the input feature vector, is the weight of the jth node, is the bias of the jth output node; then training and optimization are performed, using Adam to optimize the model parameters, adjusting the convolution kernel size, stride, pooling window size, and learning rate hyperparameters to obtain the best performance; finally, the cross-validation method is used to evaluate the generalization ability of the model.

[0035] Considering that the complexity of data will affect the accuracy of subsequent emergency command, the communication data of the emergency site including signal strength, delay, bit error rate, data throughput, network jitter and packet loss rate are obtained in real time through sensors and communication equipment. The communication data is subjected to feature extraction and feature screening to more comprehensively understand the data and improve the accuracy of decision-making. The feature extraction includes time domain features, extracting statistical features such as mean, standard deviation, skewness and kurtosis to capture the basic trend and fluctuation of the data; frequency domain features, calculating frequency domain features through Fourier transform and power spectral density to reveal the periodicity and frequency distribution in the data; and time-frequency features, applying short-time Fourier transform to extract time-frequency features to capture the time and frequency changes in the data. Feature screening includes correlation analysis, calculating the correlation between features, removing redundant features to reduce the complexity of data; recursive feature elimination, gradually eliminating unimportant features based on the importance score of the model, and optimizing feature subsets. The communication data after feature extraction and screening can more accurately reflect the actual situation, provide a more reliable basis for communication guarantee decision-making, and help improve the efficiency and effectiveness of emergency command and disposal.

[0036] The further implementation process of using the trained convolutional neural network CNN model is as follows: first input data into the CNN model and perform forward propagation calculations Among them, y is the output of the model, X is the input data, is the weight parameter of the model; the second is to generate predictions and output the evaluation results of the current communication quality, including the predicted value and the related confidence and probability distribution; then comprehensively consider multiple indicators to predict the future communication quality, use the sliding window technology to predict the future time period, slide the window forward and continuously update the prediction results in, is the predicted value at a future time point. is the current forecast value, is the characteristic of the current moment; finally, comprehensive evaluation and analysis, combining multiple evaluation indicators, comprehensively evaluate the communication quality and calculate the weighted average in, is the weight of each indicator, It is the indicator value; in addition, the forecast results and historical data are used for trend analysis to observe the changing trend of future communication quality. in, is the predicted value of the minister in the future N time. is the predicted value at the current time.

[0037] Finally, based on precise communication quality assessment and accurate prediction results, communication resources are deployed. This process includes dynamically adjusting spectrum resource allocation to ensure stable and smooth communication signals; optimizing base station power settings to better adapt to actual communication needs; and timely enabling backup communication links to respond to sudden communication failures or congestion. Through such an automatic deployment mechanism, the efficiency and effectiveness of emergency response command can be effectively improved. In an emergency, communication resources can be optimally configured, and information can be transmitted quickly and accurately, so that commanders can make decisions in a timely manner, and various departments can collaborate and efficiently perform tasks, thereby greatly improving the ability to respond to emergencies, minimizing losses, and protecting people's lives and property.

[0038] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A communication guarantee method for emergency command and disposal based on a neural network, characterized in that: The following steps are involved: S1. First, establish a multi-level redundant communication network to provide temporary communication support for the emergency site; S2, then build a convolutional neural network model, collect historical communication data and use the convolutional neural network CNN architecture to adapt to the complex characteristics of emergency communication data; S3, secondly, obtain the communication data of the emergency scene in real time through sensors and communication equipment, and perform feature extraction and screening; S4, then use the trained convolutional neural network (CNN) model to evaluate the current communication quality, and comprehensively consider multiple indicators to predict the change trend of communication quality in the future; S5. Finally, communication resources are automatically allocated based on the communication quality evaluation and prediction results; The process of using the trained convolutional neural network CNN model in step S4 is: S41. First, input data into the CNN model and perform forward propagation to calculate y=f(X,θ), where y is the output of the model, X is the input data, and θ is the weight parameter of the model; S42, secondly, generating a prediction, outputting the evaluation result of the current communication quality including the prediction value and the related confidence and probability distribution; S43, then comprehensively consider multiple indicators to predict future communication quality, use sliding window technology to predict future time periods, slide the window forward and continuously update the prediction results in, is the predicted value at a future time point. is the current predicted value, Features t It is characteristic of the present moment; S44, the final comprehensive evaluation analysis, combining multiple evaluation indicators, comprehensively evaluates the communication quality and calculates the weighted average Among them, ω i is the weight of each indicator, x i It is the indicator value; in addition, the forecast results and historical data are used for trend analysis to observe the changing trend of future communication quality. Among them, y t+N is the predicted value of the minister in the future N time, y t is the predicted value at the current time.

2. The communication guarantee method for emergency command and disposal based on neural network according to claim 1 is characterized in that: The step S1 of establishing a multi-level redundant communication network includes deploying satellite communication equipment at the emergency site and the command center, using wireless ad hoc network technology to quickly establish a local communication network and deploy emergency communication vehicles at the emergency site, equipped with communication means including satellite communication and radio, as mobile communication nodes, to provide temporary communication guarantee for the emergency site.

3. The communication guarantee method for emergency command and disposal based on neural network according to claim 1 is characterized in that: The operation of constructing the convolutional neural network CNN model in step S2 is as follows: S21, first is the convolution layer, which extracts the local features of the data. The convolution operation formula is Among them, b k is the bias of the kth convolution kernel, y i,j,k is the value of the output feature at position (i, j) and the kth channel, ω m,n,k is the weight of the convolution kernel at position (m,n) and the kth channel, x i+m,j+n is the value of the input data at position (i+m,j+n); S22, followed by the activation function, uses Leaky ReLU, which allows the gradient of negative input values ​​to exist. The formula is Where x is the input value, α is the negative slope coefficient, which is 0.01; S23, followed by the pooling layer, downsampling to reduce the data dimension, using the maximum pooling y i,j,k =max m,n (X i·s+m,j·s+n,k ) where s is the stride, which determines the sliding distance of the pooling window on the input feature map; S24, followed by a fully connected layer, is used to map the flattened feature vector to the target space Among them, x i is the i-th element of the input feature vector, ω i,j is the weight of the jth node, b j is the bias of the jth output node; S25, then train and optimize, use Adam to optimize model parameters, adjust convolution kernel size, stride, pooling window size, and learning rate hyperparameters to obtain the best performance; S26. Finally, the cross-validation method is used to evaluate the generalization ability of the model.

4. The communication guarantee method for emergency command and disposal based on neural network according to claim 1 is characterized in that: In S3, communication data of the emergency scene including signal strength, delay, bit error rate, data throughput, network jitter and packet loss rate are obtained in real time through sensors and communication equipment.

5. The communication guarantee method for emergency command and disposal based on neural network according to claim 1 is characterized in that: The feature extraction in step S3 includes time domain features, extracting mean, standard deviation, skewness and kurtosis to capture the basic trend and fluctuation of the data; frequency domain features, calculating frequency domain features through Fourier transform and power spectral density to reveal the periodicity and frequency distribution in the data; As well as time-frequency features, short-time Fourier transform is applied to extract time-frequency features to capture the time and frequency changes in the data.

6. The communication guarantee method for emergency command and disposal based on neural network according to claim 1 is characterized in that: The feature screening in step S3 includes correlation analysis, calculating the correlation between features, and removing redundant features to reduce the complexity of the data; Recursive feature elimination gradually eliminates unimportant features based on the model's importance score and optimizes the feature subset.

Images