A method for dispatching and commanding maritime search and rescue forces

Abstract

The present application provides a method for dispatching and commanding a maritime search and rescue force, comprising: dividing the search and rescue site into regions according to a generated visibility distribution map using a clustering algorithm, clustering regions with similar visibility into the same search and rescue sub-region, calculating the average visibility value and environmental characteristic coefficient of each sub-region, and establishing a correspondence table between the search and rescue region and the visibility and environmental characteristics; obtaining available search and rescue resource information from a preset search and rescue force database, including the type, quantity, performance parameters and current status of search and rescue vessels, helicopters and unmanned aerial vehicle equipment, and formulating a preliminary search and rescue force allocation plan in combination with the correspondence table between the search and rescue region and the visibility and environmental characteristics.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for dispatching and commanding maritime search and rescue forces. Background Art

[0002] In the search and rescue operations at sea in rainy days, real-time monitoring of the visibility of the search and rescue site is a key challenge. Although the current marine environment monitoring equipment and sensors can obtain a number of meteorological and hydrological parameters, it is difficult to directly measure and quantify visibility in rainy days. At the same time, the refraction and absorption effects of raindrops on light are complex and changeable, making it difficult to establish an accurate visibility prediction model for rainy days. In addition, the search and rescue site is often wide, and there are significant differences in rainfall intensity and visibility in different areas. How to achieve refined visibility monitoring of the entire search and rescue area is also a technical problem. On the other hand, the dispatching decision of the search and rescue force needs to rely on accurate and timely visibility information, but there is currently a lack of an effective mechanism to combine visibility data with the optimal allocation of search and rescue resources. How to quickly formulate and adjust the search and rescue strategies in different areas according to the dynamically changing visibility conditions and realize the precise deployment of search and rescue forces is an urgent problem to be solved. At the same time, under complex sea conditions, the detection capability of the search and rescue equipment will be significantly affected. How to ensure the efficiency and safety of search and rescue in low visibility environments also puts higher requirements on search and rescue technology and equipment. Therefore, it is necessary to research and develop a method for dispatching and commanding maritime search and rescue forces that integrates environmental perception, data analysis, and decision support functions to meet the challenges of search and rescue in low visibility conditions in rainy days. Summary of the invention

[0003] The present invention provides a method for dispatching and commanding a maritime search and rescue force, which mainly includes:

[0004] First, deploy marine environment monitoring equipment and sensor networks in various areas of the search and rescue site to collect multiple environmental parameter data including raindrop refractive index, absorption rate, and light intensity. The collected real-time data is transmitted to the central processing system through wireless transmission. The data is classified and sorted according to the data collection timestamp and geographic location information to form an environmental data set for the search and rescue site.

[0005] Using the pre-trained rain visibility prediction model, the raindrop refractive index, absorptivity and light intensity parameters in the sorted environmental data set are input to calculate the real-time visibility values ​​of each area at the search and rescue site. The calculated results are verified, outliers are removed, and the visibility distribution map of the search and rescue site is generated.

[0006] According to the generated visibility distribution map, a clustering algorithm is used to divide the search and rescue site into regions, and regions with similar visibility are clustered into the same search and rescue sub-regions. The average visibility value and environmental characteristic coefficient of each sub-region are calculated, and a corresponding relationship table between the search and rescue region and visibility and environmental characteristics is established.

[0007] Obtain available search and rescue resource information from the preset search and rescue force database, including the type, quantity, performance parameters and current status of search and rescue vessels, helicopters and drones, and formulate a preliminary search and rescue force allocation plan based on the corresponding relationship table between the search and rescue area and visibility and environmental characteristics;

[0008] Establish a search and rescue force dispatch priority assessment model, input the average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area, calculate the search and rescue difficulty coefficient and priority of each sub-area, sort the sub-areas according to the priority, determine the priority of search and rescue force input in combination with the preliminary search and rescue force allocation plan, and determine the final search and rescue plan;

[0009] Build a real-time communication network to connect the search and rescue command center with various search and rescue units. Use encrypted transmission protocol to send the final search and rescue plan and dispatch instructions. After receiving the instructions, the search and rescue units send confirmation information to the command center. The system records the time when the instructions are issued and confirmed, and monitors the response speed of the search and rescue forces.

[0010] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0011] The present invention discloses a method for dispatching and commanding a search and rescue force at sea. First, by efficiently and accurately collecting and processing the marine environment data of the search and rescue site, a valuable environmental data set is formed. Then, the visibility distribution of the search and rescue site is accurately predicted and drawn by using models and algorithms. At the same time, the data is verified and outliers are processed to ensure the accuracy and reliability of the results. In addition, by reasonably dividing the search and rescue site into regions, a complexity distribution map at sea is constructed to provide a basis for the subsequent allocation of search and rescue forces. A preliminary search and rescue force allocation plan is formulated in combination with the available search and rescue resource information and the on-site situation, so that the reasonable allocation of resources can be achieved. At the same time, by determining the priority of the search and rescue force investment, a final search and rescue plan is formed, and the search and rescue force can be invested according to the needs and priorities of different regions to improve the search and rescue efficiency and effect. Finally, by building real-time effective communication and command transmission, the response speed of the search and rescue force is monitored, and the efficient communication and collaborative work between the command center and the search and rescue units are guaranteed, and the response speed and execution efficiency are improved. In summary, the present invention realizes rapid and effective search and rescue under complex weather conditions by integrating environmental monitoring, data processing, resource scheduling and communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flow chart of a method for dispatching and commanding maritime search and rescue forces.

[0013] Figure 2 It is a schematic diagram of a method for dispatching and commanding maritime search and rescue forces according to the present invention.

[0014] Figure 3 It is another schematic diagram of a method for dispatching and commanding maritime search and rescue forces according to the present invention. DETAILED DESCRIPTION

[0015] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0016] like Figure 1-3 In this embodiment, a method for dispatching and commanding a maritime search and rescue force may specifically include:

[0017] S101. First deploy marine environment monitoring equipment and sensor networks in various areas of the search and rescue site to collect multiple environmental parameter data including raindrop refractive index, absorption rate, and light intensity. The collected real-time data is transmitted to the central processing system via wireless transmission. The data is classified and sorted according to the data collection timestamp and geographic location information to form an environmental data set for the search and rescue site.

[0018] A plurality of marine environment monitoring devices are arranged in different areas of the search and rescue site, and the marine environment monitoring devices obtain multiple environmental parameters such as raindrop refractive index, absorption rate and light intensity; the environmental parameter data collected by the marine environment monitoring devices are transmitted to a central processing system by using LoRaWAN communication, and the central processing system classifies and organizes the received environmental parameter data; outlier processing is performed on the environmental parameter data according to the timestamp information contained in the environmental parameter data and the device geographical location information obtained in real time by the built-in GPS module; the processed environmental parameter data are grouped by using the K-means clustering algorithm to obtain grouped environmental parameter data; the grouped environmental parameter data are filled in the spatial dimension by using the Kriging interpolation method to form a search and rescue site environmental data model.

[0019] Specifically, multiple marine environment monitoring devices are set up in different areas of the search and rescue site. The marine environment monitoring devices include sensor modules for collecting multiple environmental parameters such as raindrop refractive index, absorption rate, light intensity, etc. Different sampling frequencies are set according to the changing characteristics of different environmental parameters. The light intensity is collected every 5 seconds, and the raindrop refractive index is collected every 30 seconds. The environmental parameter values ​​are continuously collected, and the corresponding timestamp information is recorded while collecting data. LoRaWAN communication technology is used for data transmission. This technology has the characteristics of long distance and low power consumption, which is suitable for the complex environment of the search and rescue site. The real-time environmental parameter data collected by the marine environment monitoring equipment is transmitted to the central processing system. The data is encrypted during the transmission process to ensure the security and integrity of the data transmission. The central processing system receives the environmental parameter data from each marine environment monitoring device, and classifies and organizes the received environmental parameter data according to the timestamp information contained in the data and the device geographic location information obtained in real time through the built-in GPS module. The moving median method is used to remove outliers, and the linear interpolation method is used to handle short-term data loss. The K-means clustering algorithm is used to group the cleaned environmental parameter data, and the environmental data set of the search and rescue site is constructed based on the time and space dimensions. The Kriging interpolation method is used to fill the data in the spatial dimension to form a complete search and rescue site environmental data model. The marine environment monitoring equipment deployed at the search and rescue site adopts a multi-layer sensor structure, including a light intensity sensor on the top layer, a raindrop refractive index sensor in the middle layer, and an absorptivity sensor on the bottom layer. The light intensity sensor collects data every 5 seconds, with a measurement range of 0-100000 lux and an accuracy of ±1 lux. The raindrop refractive index sensor collects data every 30 seconds, with a measurement range of 1.3300-1.3900 and an accuracy of ±0.0001. The absorptivity sensor collects data every 60 seconds, with a measurement range of 0-100% and an accuracy of ±0.1%. The collected data is transmitted through the LoRaWAN network, and the AES-128 encryption algorithm is used to ensure data security. The transmission distance can reach 10 kilometers, and the power consumption is only 10mW. After the central processing system receives the data, it first uses the moving median method to remove outliers, with the window size set to 5, and removes data points that exceed ±3 times the standard deviation of the median. For short-term data missing of no more than 5 minutes, linear interpolation is used to fill in. The location accuracy of the GPS module is ±2 meters, and the location information is updated every 60 seconds. The K-means clustering algorithm is used to group the cleaned data. Five cluster centers are initially set, the number of iterations is 100, and the data points are allocated by calculating the Euclidean distance. Finally, the Kriging interpolation method is used to fill in the data in the spatial dimension. The interpolation grid size is set to 10 meters × 10 meters. Considering the influence of the surrounding 16 nearest neighbor points, the weight is calculated inversely proportional to the distance.Through this series of processing, a three-dimensional environmental data model with a resolution of 10 meters × 10 meters × 1 minute was formed, covering key environmental parameters such as light intensity, raindrop refractive index and absorption rate at the search and rescue site.

[0020] S102. Using the pre-trained rainy day visibility prediction model, the raindrop refractive index, absorptivity and light intensity parameters in the sorted environmental data set are input to calculate the real-time visibility values ​​of each area at the search and rescue site, and the calculation results are verified to remove abnormal values, so as to generate a visibility distribution map of the search and rescue site.

[0021] The raindrop refractive index, absorptivity and light intensity parameters are extracted from the environmental data set, and the parameters are processed by the minimum-maximum normalization method to obtain normalized characteristic parameters; according to the normalized characteristic parameters, the real-time visibility values ​​of each area of ​​the search and rescue site are calculated by a pre-trained rainy day visibility prediction model, and the pre-trained rainy day visibility prediction model is constructed by a random forest algorithm; for the real-time visibility values, the standard deviation of the prediction results of the decision tree in the random forest is calculated to obtain the predicted confidence interval; if there are outliers in the real-time visibility values, the median absolute deviation method is used to detect the outliers of the real-time visibility values, and the values ​​exceeding the threshold are marked as outliers and removed; for the verified visibility values, the Kriging interpolation method is used for spatial interpolation, and the optimal theoretical semivariogram model is determined by the experimental semivariogram to generate a high-resolution visibility distribution map covering the entire search and rescue site.

[0022] Specifically, the raindrop refractive index, absorptivity and light intensity parameters are extracted from the environmental data set, and the minimum-maximum normalization method is used to adjust the value range of each feature to between 0 and 1. The calculation formula is X_norm=(X-X_min) / (X_max-X_min), where X is the original value, X_min and X_max are the minimum and maximum values ​​of the feature, respectively. The pre-trained rainy day visibility prediction model is loaded. The model is constructed using the random forest algorithm, the number of decision trees is set to 100, the maximum tree depth is 10, the minimum number of leaf node samples is 5, and the Gini index is used as the splitting standard. The normalized raindrop refractive index, absorptivity and light intensity parameters are input, and the real-time visibility values ​​of each area of ​​the search and rescue site are calculated by the model. The standard deviation of the prediction results of all decision trees in the random forest is calculated to obtain the confidence interval of the prediction. The calculated real-time visibility values ​​were verified. Considering the possible non-normal distribution of the data at the rescue site, the median absolute deviation method (MAD) was used to detect outliers. The threshold was set to 3 times the MAD, and the values ​​exceeding the threshold were marked as outliers and removed. According to the geographic coordinate information of the rescue site, the Kriging interpolation method was used to spatially interpolate the verified visibility values. First, the best theoretical semivariogram model, such as the spherical model, exponential model or Gaussian model, was determined by experimental semivariogram. Then, the search radius was set to 500 meters, and 12 nearest neighbor points were used for interpolation calculation to generate a high-resolution visibility distribution map covering the entire rescue site, and the distribution map data was stored in raster format. At the rescue site, the environmental data set contained raindrop refractive index ranging from 1.33 to 1.36, absorptivity ranging from 0.1 to 0.9, and light intensity ranging from 100 to 10,000 lux. The minimum-maximum normalization method was used to process these data, such as the raindrop refractive index of 1.345 was normalized to 0.5. The pre-trained random forest model contains 100 decision trees, each with a maximum depth of 10 and a minimum number of leaf node samples of 5. The Gini index is used as the splitting criterion. The model inputs the normalized parameters and outputs the visibility prediction value. For example, if the input is [0.5, 0.4, 0.6], the output visibility prediction value is 200 meters. At the same time, the standard deviation of the prediction results of 100 trees is calculated to obtain the prediction confidence interval, such as ±20 meters. Outlier detection is performed on the prediction results. First, the median, such as 200 meters, and the median absolute deviation, such as 15 meters, are calculated, and the threshold is set to 3 times the MAD, that is, 45 meters. If the predicted value exceeds the range of 155 meters to 245 meters, it is marked as an outlier and removed. The visibility distribution map is generated using the Kriging interpolation method. The best model is determined to be a spherical model through the experimental semivariogram function. The search radius is set to 500 meters, and 12 nearest neighbor points are used for interpolation. For example, at the coordinates 100,100, the interpolation result of the point is 195 meters based on the visibility values ​​of the surrounding 12 points such as 180 meters and 210 meters.Finally, a visibility distribution map with a grid format of 10 m × 10 m resolution is generated, covering the entire search and rescue area of ​​1000 m × 1000 m, and each grid cell contains the interpolated visibility value.

[0023] The anomaly detection algorithm is used to screen the collected raindrop parameter and environmental parameter data in real time to determine whether the data exceeds the normal range. If abnormal values ​​are found, they will be eliminated.

[0024] Obtain historical data of raindrop parameters and environmental parameters, calculate the mean and standard deviation of the parameters, and determine the normal value range based on the mean and standard deviation. Use the local anomaly factor algorithm to detect anomalies in real-time data. If continuous abnormal data points are detected, an equipment failure alarm is triggered. For the abnormal value data points, select an interpolation method based on the data change characteristics, obtain the interpolation result to replace the abnormal value. Input the corrected data set into the visibility prediction algorithm, and obtain the visibility distribution map based on the prediction algorithm. Set a sliding time window based on the visibility distribution map, and update the visibility distribution map through the sliding time window.

[0025] Specifically, based on the historical data of the past 30 days, the mean and standard deviation of raindrop parameters and environmental parameters are calculated, the normal value range is set to the mean ± 3 times the standard deviation, and a parameter threshold library containing multiple indicators such as raindrop refractive index, absorptivity, and light intensity is established. The local anomaly factor algorithm is used to detect anomalies in the collected real-time data. According to the data density, the k value is set to 1% of the total data, and the minimum is not less than 5. The local density ratio of each data point to its k nearest neighbors is calculated. If the ratio is significantly higher than the preset threshold, it is determined to be an anomaly. If three consecutive data points are determined to be abnormal, the equipment fault alarm is triggered. For data points determined to be abnormal values, the appropriate interpolation method is selected according to the data change characteristics. Linear interpolation is used when the change is gentle, and spline interpolation is used when the change is drastic. The interpolation result is calculated using the normal value of the time period before and after the data point to replace the original abnormal value. The corrected data set is input into the visibility prediction algorithm, and the sliding time window size is set to 10 minutes. The data in the window is updated every 1 minute, and the visibility distribution is recalculated. Parallel computing technology is used to ensure that the data processing time does not exceed 30 seconds, and the visibility distribution map is continuously updated to ensure the real-time and accuracy of the image. In the anomaly detection system at the search and rescue site, the parameter threshold library is first established using historical data from the past 30 days. For example, the mean value of the refractive index of raindrops is 1.33, the standard deviation is 0.01, and the normal range is set to 1.30 to 1.36; the mean value of the light intensity is 5000 lux, the standard deviation is 1000 lux, and the normal range is 2000 to 8000 lux. The real-time data collection frequency is once per second, and the local anomaly factor algorithm is used for anomaly detection. When the total amount of data is 10,000, the k value is set to 100. If the local density ratio of a data point exceeds 2.0, it is judged to be abnormal. For example, the raindrop refractive index measured at a certain moment is 1.40, which is obviously beyond the normal range and is marked as an abnormal value. When the system detects three abnormal values ​​in succession, such as the refractive index measured for three consecutive seconds is 1.40, 1.41, and 1.42, the equipment fault alarm is automatically triggered. For the determined abnormal values, the system selects the interpolation method according to the data change rate. If the data change rate within the previous and next 10 seconds is less than 5%, linear interpolation is used; otherwise, cubic spline interpolation is used. The interpolated data replaces the original abnormal value, such as correcting 1.40 to 1.35. The corrected data set is input into the visibility prediction algorithm, using a 10-minute sliding time window and updated every minute. For example, the data from 9:00 to 9:10 is used to generate the visibility distribution map at 9:10, the data from 9:01 to 9:11 is used to generate the distribution map at 9:11, and so on. Through parallel computing, 6 processors process data from different areas at the same time to ensure that a complete visibility distribution map update is completed within 30 seconds.

[0026] S103. Based on the generated visibility distribution map, a clustering algorithm is used to divide the search and rescue site into regions, regions with similar visibility are clustered into the same search and rescue sub-regions, the average visibility value and environmental characteristic coefficient of each sub-region are calculated, and a corresponding relationship table between the search and rescue regions and the visibility and environmental characteristics is established.

[0027] The generated visibility distribution map is gridded, and the search and rescue site is divided into square grid units, and the side length of the square grid unit is a preset value. According to the visibility data after gridding, the K-means clustering algorithm is used for clustering analysis. The K-means clustering algorithm clusters grid units with similar visibility into the same search and rescue sub-area by calculating the Euclidean distance between grid units. If there is a boundary area in the clustering result, the fuzzy C-means algorithm is used to optimize the boundary area. The fuzzy C-means algorithm allows the grid unit to belong to multiple sub-areas at the same time. For each search and rescue sub-area obtained after clustering, the average visibility value of all grid units in the sub-area is calculated, and the environmental characteristic vector is constructed in combination with the environmental parameters in the sub-area. The environmental characteristic vector includes the mean and standard deviation of the refractive index of raindrops, the mean and standard deviation of the absorptivity, and the mean and standard deviation of the light intensity. According to the average visibility value and the environmental feature vector, a correspondence table between the search and rescue area and visibility and environmental features is established, wherein each row in the correspondence table contains a sub-area ID, an average visibility value, an environmental feature vector and area boundary coordinates.

[0028] Specifically, the generated visibility distribution map is gridded, and the search and rescue site is divided into 10m×10m square grid cells. Each grid cell records its corresponding visibility value and geographic coordinate information. The K-means clustering algorithm is used to perform cluster analysis on the gridded visibility data, and the silhouette coefficient method is used to determine the optimal K value. The initial K value range is set to 3 to 10. By calculating the Euclidean distance between grid cells, grid cells with similar visibility are clustered into the same search and rescue sub-area, and preliminary regional division results are obtained. Specifically, by calculating the Euclidean distance between grid cells, the K-means algorithm will cluster grid cells with close distances, that is, with similar visibility values, into the same search and rescue sub-area. For the cluster boundary area, the fuzzy C-means algorithm is used for optimization, allowing grid cells to belong to multiple sub-areas at the same time. For each search and rescue sub-area after clustering, the average visibility value of all grid cells in the sub-area is calculated, and the environmental parameters such as raindrop refractive index, absorption rate, and light intensity in the sub-area are combined to construct the environmental feature vector, including the mean and standard deviation of raindrop refractive index, the mean and standard deviation of absorption rate, and the mean and standard deviation of light intensity, a total of 6 feature values. Based on the calculated average visibility value and environmental feature vector, a corresponding relationship table between the search and rescue area and visibility and environmental characteristics is established. Each row in the table represents a sub-area, including columns of sub-area ID, average visibility value, 6 values ​​of environmental feature vector, and minimum longitude and latitude and maximum longitude and latitude of the area boundary coordinates. In the process of regional division of the search and rescue site, the 1000m × 1000m search and rescue area is first divided into 10,000 10m × 10m grid cells, and each cell records its visibility value and center point coordinates. For example, the visibility value of the grid cell at coordinates 500,500 is 200 meters. Subsequently, the K-means clustering algorithm was used to classify these grid cells, and the optimal K value was determined by the silhouette coefficient method. Assuming that the initial K value is set to 3 to 10, it is calculated that the maximum silhouette coefficient is 0.68 when K=5, so the search and rescue area is divided into 5 sub-areas. During the clustering process, for the grid cells in the boundary area, such as the cells at coordinates 495,505, the fuzzy C-means algorithm is used for optimization, allowing them to belong to two adjacent sub-areas at the same time, with membership degrees of 0.6 and 0.4, respectively. The average visibility value is calculated for each sub-area, such as the average visibility of sub-area 1 is 180 meters. At the same time, the environmental feature vector is constructed, including the mean refractive index of raindrops of 1.33 and standard deviation of 0.01, the mean absorptivity of 0.5 and standard deviation of 0.05, and the mean light intensity of 5000 lux and standard deviation of 1000 lux. Finally, a corresponding relationship table is generated. For example, the record of sub-area 1 is ID=1, average visibility=180 meters, environmental feature vector=[1.33, 0.01, 0.5, 0.05, 5000, 1000], and boundary coordinates=(minimum longitude 100.5, minimum latitude 30.2, maximum longitude 100.7, maximum latitude 30.4).

[0029] A three-dimensional grid model is constructed according to the maritime characteristics and environmental factors of the search and rescue area. An adaptive grid division algorithm is used to refine the complex wave area and coarsen the calm wave area. The complexity of the sea is reflected by the grid node density, and the maritime complexity distribution map of the search and rescue area is obtained.

[0030] An initial three-dimensional grid model is constructed based on the maritime characteristics and environmental factor data of the search and rescue area, and the search and rescue area is divided into cubic grid units, which record the corresponding environmental parameters of wave height, wind speed and current speed; an octree algorithm is used to adaptively divide the initial grid, and refine the area where the wave height changes dramatically. If the wave height difference between adjacent grid units exceeds a preset threshold, the grid unit is recursively divided into smaller sub-units until the minimum grid size is reached or the refinement threshold condition is met; grid coarsening is performed on the area with calm waves, and if the wave height change of multiple consecutive grid units is less than a preset threshold, these units are merged into a large grid unit; the grid node density of each area in the processed grid model is calculated, and the density value is mapped to a complexity level range, and the complexity level range is determined using the quartile method based on the overall data distribution; a maritime complexity distribution map of the search and rescue area is generated based on the complexity level, and different colors are used to represent different complexity levels.

[0031] Specifically, according to the maritime characteristics and environmental factors data of the search and rescue area, an initial three-dimensional grid model is constructed, and the search and rescue area is divided into 100m×100m×10m cubic grid units. Each grid unit records its corresponding environmental parameters such as wave height, wind speed, and current speed. The octree algorithm is used to adaptively divide the initial grid, and refine the area where the wave height changes dramatically. When the wave height difference between adjacent grid units exceeds 1 meter, the grid unit is recursively divided into smaller sub-units until the minimum grid size of 10m×10m×1m is reached or the refinement threshold condition is met. The grid is coarsened in the calm wave area. When the wave height change of 10 consecutive grid units is less than 0.1 meter, these units are merged into a large grid unit to reduce the number of grid nodes and reduce the computational complexity, while retaining sufficient accuracy to describe the characteristics of the calm sea area. The density of grid nodes in each area of ​​the grid model after calculation is mapped to a complexity level range of 1-10, with 1 representing the calmest and 10 representing the most complex. The density range corresponding to each level is determined using the quartile method based on the overall data distribution to generate a maritime complexity distribution map of the search and rescue area, with different colors representing different complexity levels. In the process of three-dimensional grid modeling of the search and rescue area, the search and rescue space of 10 km × 10 km × 100 m is first divided into 10,000 initial grid cells, each with a size of 100 m × 100 m × 10 m. Each cell records environmental parameters such as wave height, wind speed, and current speed. For example, the grid cell at coordinates 5000, 5000, 50 records a wave height of 2.5 m, a wind speed of 10 m / s, and a current speed of 1.2 m / s. Subsequently, the octree algorithm is used for adaptive partitioning. Assume that the difference in wave height between adjacent cells at coordinates 5000,5000,50 and 5100,5000,50 is 1.2 meters, exceeding the threshold of 1 meter, and the two cells are each divided into 8 sub-cells of 10 meters × 10 meters × 1 meters. In calm areas, such as 10 consecutive cells from coordinates 1000,1000,0 to 2000,1000,0, the wave height change is less than 0.1 meters, and these cells are merged into a large cell of 1000 meters × 100 meters × 10 meters. After processing, the number of grid nodes in each 100 meters × 100 meters area is calculated to obtain the node density distribution. The density value is mapped to a complexity level of 1-10 using the quartile method, such as areas with a density less than 10 are mapped to level 1, and areas with a density between 90-100 are mapped to level 10. In the final complexity distribution map, for example, the 100m×100m area at coordinates 5000,5000 is displayed in red, indicating a complexity level of 10, while the area at coordinates 1500,1000 is displayed in blue, indicating a complexity level of 1.

[0032] S104. Obtain available search and rescue resource information from the preset search and rescue force database, including the type, quantity, performance parameters and current status of search and rescue vessels, helicopters and UAV equipment, and formulate a preliminary search and rescue force allocation plan based on the correspondence table between the search and rescue area and visibility and environmental characteristics.

[0033] Obtain available search and rescue resource information in the preset search and rescue force database, the available search and rescue resource information includes the type, quantity, performance parameters and real-time status of search and rescue vessels, helicopters and unmanned aerial vehicle equipment. According to the available search and rescue resource information, the search and rescue area is divided into 1 km × 1 km grids, and the environmental complexity index of each grid unit is calculated. The KM algorithm is used to optimally match the available search and rescue resources with the grid units to obtain a preliminary search and rescue force allocation plan. The preliminary allocation plan is optimized using a genetic algorithm, and the population size, crossover rate and mutation rate are set. The fitness function is defined as the weighted sum of the search and rescue coverage rate and the resource utilization rate. Through multiple generations of iterative calculations, the spatial distribution of search and rescue resources is optimized to obtain the final search and rescue force allocation plan. If the final search and rescue force allocation plan meets the preset coverage rate threshold, the final search and rescue force allocation plan is output; if the final search and rescue force allocation plan does not meet the preset coverage rate threshold, the genetic algorithm optimization link is re-executed.

[0034] Specifically, the SQL statement SELECT * FROM resources WHERE status = 'available' is used to extract the currently available search and rescue resource information from the preset search and rescue force database, including the type, quantity, performance parameters and real-time status of search and rescue ships, helicopters and drone equipment, and generate a list of available resources. According to the corresponding relationship table between the search and rescue area and visibility and environmental characteristics, the search and rescue area is divided into 1 km × 1 km grids, and the environmental complexity index of each grid unit is calculated. The index comprehensively considers multiple environmental factors through the formula environmental complexity index = 0.4*(1 / visibility)+0.3*sea state level+0.3*wind speed level. The KM algorithm is used to optimally match the available search and rescue resources with the grid units, and the adaptability of the performance parameters of the search and rescue equipment and the environmental complexity index is considered to obtain a preliminary search and rescue force allocation plan. The preliminary allocation scheme was optimized using a genetic algorithm. The population size was set to 100, the crossover rate was 0.8, the mutation rate was 0.1, and the fitness function was defined as the weighted sum of the search and rescue coverage rate and the resource utilization rate. Through multiple generations of iterative calculations, the spatial distribution of search and rescue resources was optimized to generate the final search and rescue force allocation scheme. In the search and rescue force allocation process, the available resources were first extracted from the database through SQL queries, such as obtaining detailed information on 5 search and rescue ships, 3 helicopters, and 10 drones. The 100 square kilometers of search and rescue area was then divided into 100 1 km × 1 km grid cells. The environmental complexity index was calculated for each cell. For example, the cell at coordinate (5,5) had a visibility of 2 km, a sea state level of 4, and a wind speed level of 5. The environmental complexity index was 0.4*(1 / 2)+0.34+0.35=2.9. The KM algorithm is used to match 18 search and rescue resources with 100 grid cells, considering the adaptability of resource performance and environmental complexity, such as allocating large search and rescue ships to cells with high environmental complexity. After obtaining the preliminary plan, the genetic algorithm is used for optimization, the population size is set to 100, and 1000 generations are iterated. Each individual represents an allocation plan, and new plans are generated through crossover and mutation. The fitness function is 0.6 search and rescue coverage + 0.4 resource utilization. For example, if a plan covers 80% of the area and the resource utilization rate is 90%, the fitness is 0.680+0.490=84. After optimization, the search and rescue force allocation plan with the highest fitness is finally obtained, such as allocating 2 search and rescue ships to coordinates (3,3) and (7,7), 1 helicopter to (5,5), and 3 drones to (2,8), (6,4) and (9,1).

[0035] S105. Establish a search and rescue force dispatch priority assessment model, input the average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area, calculate the search and rescue difficulty coefficient and priority of each sub-area, sort the sub-areas according to the priority, determine the priority of search and rescue force input in combination with the preliminary search and rescue force allocation plan, and determine the final search and rescue plan.

[0036] The average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area are obtained as input layer data of the neural network; a multilayer perceptron neural network is constructed according to the input layer data, and the multilayer perceptron neural network includes two hidden layers; the historical search and rescue data are standardized, and the performance of the neural network model is evaluated by a cross-validation method; if the model performance meets the preset threshold, the neural network parameters are optimized by a back propagation algorithm; a trained priority evaluation model is obtained by minimizing the mean square error loss function; the characteristic data of the sub-area is input into the trained neural network model, and the search and rescue difficulty coefficient and its uncertainty of each sub-area are calculated by a Monte Carlo Dropout method; the sub-areas are sorted in descending order according to the search and rescue difficulty coefficient to obtain a search and rescue priority sorting list; in combination with the search and rescue force allocation plan and the sub-area priority sorting list, the sub-areas are traversed in priority order, and the minimum required resources are allocated to each area until the resources are exhausted or all areas are covered, and the final search and rescue plan is determined.

[0037] Specifically, a search and rescue force dispatch priority evaluation model is constructed, and the optimal network structure is determined through grid search. A multi-layer perceptron neural network is used. The input layer contains four neurons: the average visibility value of the sub-region, the environmental characteristic coefficient, the area size, and the geographical location factor. The hidden layer is set to two layers, each containing 8 neurons, and the output layer is a neuron representing the search and rescue difficulty coefficient. The historical search and rescue data is standardized, and the model performance is evaluated using cross-validation. The network parameters are optimized using the back propagation algorithm. The learning rate is set to 0.01, the number of iterations is 1000, and the mean square error loss function is minimized to obtain a trained priority evaluation model. The characteristic data of each sub-region is input into the trained neural network model, and the confidence interval predicted by the model is estimated using the Monte Carlo Dropout method. The search and rescue difficulty coefficient and its uncertainty of each sub-region are calculated, and the sub-regions are sorted in descending order according to the difficulty coefficient to obtain a search and rescue priority ranking list. Combined with the preliminary search and rescue force allocation plan and the sub-region priority ranking list, the sub-regions are traversed in priority order, and the minimum required resources are allocated to each region until the resources are exhausted or all regions are covered, thereby determining the final search and rescue plan. In the process of search and rescue force dispatch priority assessment, the neural network structure was first determined through grid search, and 1 to 3 hidden layers, 4 to 16 neurons per layer were tried. Finally, a structure with 2 hidden layers and 8 neurons per layer was selected. 1000 historical search and rescue data were standardized, and visibility, environmental characteristic coefficients, area and geographical location factors were mapped to the range of 0-1. Using 5-fold cross validation, the average validation set accuracy reached 85%. The Adam optimizer with a learning rate of 0.01 was used to train the model after 1000 iterations. For a new search and rescue task, the feature data of 10 sub-areas were input. For example, the normalized data of sub-area 1 was [0.3, 0.7, 0.5, 0.6]. The search and rescue difficulty coefficient was 0.75 and the confidence interval was [0.70, 0.80] after 50 Monte Carlo Dropout calculations. The 10 sub-areas were sorted according to the difficulty coefficient to obtain a priority list [1, 3, 7, 2, 9, 5, 8, 4, 6, 10]. Combined with the initially allocated 20 drones, 5 search and rescue ships and 3 helicopters, the sub-areas are traversed according to priority, and 2 drones and 1 search and rescue ship are allocated to area 1 with the highest difficulty coefficient, 2 drones and 1 helicopter are allocated to area 3, and so on until the resources are exhausted. Finally, a search and rescue plan covering 8 sub-areas is formed.

[0038] S106. Build a real-time communication network to connect the search and rescue command center with each search and rescue unit. Use an encrypted transmission protocol to send the final search and rescue plan and dispatch instructions. After receiving the instructions, the search and rescue units send confirmation information to the command center. The system records the time when the instructions are issued and confirmed, and monitors the response speed of the search and rescue forces.

[0039] A real-time communication system is deployed according to the 4G / 5G network. The real-time communication system establishes a mesh topology structure, uses the OSPF protocol for dynamic routing, and obtains an optimized data transmission path; uses the AES-256 encryption algorithm to encrypt the search and rescue plan and dispatch instructions, and uses a 2048-bit RSA key for key exchange to obtain the encrypted search and rescue plan and dispatch instructions; develops a real-time message push mechanism based on the WebSocket protocol, and the search and rescue command center sends the encrypted search and rescue plan and dispatch instructions to each search and rescue unit through the real-time message push mechanism. If the sending fails, it will automatically retry three times; constructs a distributed time synchronization system, which uses the NTP protocol and the two-way time synchronization algorithm, combined with the round-trip transmission delay, to determine whether the clocks of each node are consistent; uses a time series database to store the instruction issuance time and confirmation time, and calculates the time difference between the instruction issuance time and the confirmation time to obtain the response speed of the search and rescue force.

[0040] Specifically, a real-time communication system based on 4G / 5G network is deployed, a mesh topology is established between the search and rescue command center and each search and rescue unit, and the OSPF open shortest path first protocol is used for dynamic routing to optimize the data transmission path to ensure the stability and reliability of the communication network. At the same time, a satellite communication system is deployed as a backup in areas where 4G / 5G network coverage is insufficient. The AES-256 encryption algorithm is used to encrypt the search and rescue plan and dispatch instructions, and a 2048-bit RSA key is used for key exchange to achieve end-to-end secure communication and prevent data from being intercepted or tampered with during transmission. A real-time message push mechanism based on the WebSocket protocol is developed, through which the search and rescue command center sends encrypted search and rescue plans and dispatch instructions to each search and rescue unit, realizing the message queue and retransmission mechanism. When the message fails to be sent, it will automatically retry 3 times. If it still fails, an alarm will be triggered. After receiving the instruction, the search and rescue unit will automatically decrypt and send a confirmation message to the command center. Construct a distributed time synchronization system, use the NTP protocol and two-way time synchronization algorithm, consider the round-trip transmission delay, ensure the consistency of the clock of each node, accurately record the time of command issuance and confirmation, use the time series database to store these time information, monitor the response speed of the search and rescue force by calculating the time difference, and use the moving average algorithm to calculate the average response time of each search and rescue unit, and update the response speed ranking in real time. In the implementation of the search and rescue communication system, first deploy a 4G / 5G network covering a search and rescue area of ​​100 square kilometers, set up routers at 10 key nodes, and update the routing table every 30 seconds through the OSPF protocol to achieve dynamic optimization. At the same time, satellite communication terminals are installed in 3 remote locations as backup for 4G / 5G signal weak areas. The AES encryption algorithm with a 256-bit key is used to process the search and rescue instructions, and the key exchange is performed once an hour using a 2048-bit RSA key. After the WebSocket server is established, push messages to 20 search and rescue units, set a retransmission interval of 100ms, and retry up to 3 times. If all three times fail, a red alert is sent to the command center console. The NTP server synchronizes time to all units at 10-second intervals, using a two-way synchronization algorithm to calculate the round-trip delay and keep the error within 5ms. The time of command issuance and confirmation is recorded in the InfluxDB time series database, and the average response time is calculated every minute. For example, the last 10 response times of unit A are [15s, 18s, 20s, 17s, 16s, 19s, 21s, 18s, 17s, 20s], and the average response time is 18.1 seconds using a 5-minute moving average window. The system updates the response speed rankings in real time based on these data and displays them on the large screen of the command center to help commanders quickly identify search and rescue units that are slow to respond.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for dispatching and commanding a maritime search and rescue force, characterized in that: The method comprises: First, deploy marine environment monitoring equipment and sensor networks in various areas of the search and rescue site to collect multiple environmental parameter data including raindrop refractive index, absorption rate, and light intensity. The collected real-time data is transmitted to the central processing system through wireless transmission. The data is classified and sorted according to the data collection timestamp and geographic location information to form an environmental data set for the search and rescue site. Using the pre-trained rain visibility prediction model, the raindrop refractive index, absorptivity and light intensity parameters in the sorted environmental data set are input to calculate the real-time visibility values ​​of each area at the search and rescue site. The calculated results are verified, outliers are removed, and the visibility distribution map of the search and rescue site is generated. According to the generated visibility distribution map, a clustering algorithm is used to divide the search and rescue site into regions, and regions with similar visibility are clustered into the same search and rescue sub-regions. The average visibility value and environmental characteristic coefficient of each sub-region are calculated, and a corresponding relationship table between the search and rescue region and visibility and environmental characteristics is established. Obtain available search and rescue resource information from the preset search and rescue force database, including the type, quantity, performance parameters and current status of search and rescue vessels, helicopters and drones, and formulate a preliminary search and rescue force allocation plan based on the corresponding relationship table between the search and rescue area and visibility and environmental characteristics; Establish a search and rescue force dispatch priority assessment model, input the average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area, calculate the search and rescue difficulty coefficient and priority of each sub-area, sort the sub-areas according to the priority, determine the priority of search and rescue force input in combination with the preliminary search and rescue force allocation plan, and determine the final search and rescue plan; Build a real-time communication network to connect the search and rescue command center with various search and rescue units. Use encrypted transmission protocol to send the final search and rescue plan and dispatch instructions. After receiving the instructions, the search and rescue units send confirmation information to the command center. The system records the time when the instructions are issued and confirmed, and monitors the response speed of the search and rescue forces.

2. The method according to claim 1, wherein: The marine environment monitoring equipment and sensor network are first deployed in various areas of the search and rescue site to collect multiple environmental parameter data including raindrop refractive index, absorption rate, and light intensity. The collected real-time data is transmitted to the central processing system by wireless transmission. The data is classified and sorted according to the data collection timestamp and geographic location information to form an environmental data set for the search and rescue site, including: Multiple marine environment monitoring devices are set up in different areas of the search and rescue site, and the marine environment monitoring devices obtain multiple environmental parameters such as raindrop refractive index, absorption rate and light intensity; The environmental parameter data collected by the marine environment monitoring equipment is transmitted to a central processing system using LoRaWAN communication, and the central processing system classifies and organizes the received environmental parameter data; Performing outlier processing on the environmental parameter data according to the timestamp information contained in the environmental parameter data and the device geographic location information acquired in real time by the built-in GPS module; The processed environmental parameter data are grouped using the K-means clustering algorithm to obtain grouped environmental parameter data; The grouped environmental parameter data are filled in the spatial dimension using the Kriging interpolation method to form a search and rescue scene environmental data model.

3. The method according to claim 1, wherein: The method uses the pre-trained rain visibility prediction model, inputs the raindrop refractive index, absorptivity and light intensity parameters in the sorted environmental data set, calculates the real-time visibility values ​​of each area of ​​the search and rescue site, performs data verification on the calculation results, removes abnormal values, and generates a visibility distribution map of the search and rescue site, including: Extracting the raindrop refractive index, absorptivity and light intensity parameters from the environmental data set, processing the parameters using a minimum-maximum normalization method to obtain normalized characteristic parameters; According to the normalized characteristic parameters, the real-time visibility values ​​of each area of ​​the search and rescue site are calculated by using a pre-trained rainy day visibility prediction model, where the pre-trained rainy day visibility prediction model is constructed using a random forest algorithm; For the real-time visibility value, calculate the standard deviation of the prediction results of the decision tree in the random forest to obtain the confidence interval of the prediction; If there are outliers in the real-time visibility values, the median absolute deviation method is used to detect outliers in the real-time visibility values, and the values ​​exceeding the threshold are marked as outliers and removed; For the verified visibility values, the Kriging interpolation method is used for spatial interpolation. The optimal theoretical semivariogram model is determined through the experimental semivariogram to generate a high-resolution visibility distribution map covering the entire search and rescue site. It also includes: using anomaly detection algorithms to screen the collected raindrop parameter and environmental parameter data in real time to determine whether the data exceeds the normal range. If abnormal values ​​are found, they will be eliminated.

4. The method according to claim 3, wherein: The method uses an anomaly detection algorithm to screen the collected raindrop parameter and environmental parameter data in real time to determine whether the data exceeds the normal range. If an abnormal value is found, it is removed, including: Obtain historical data of raindrop parameters and environmental parameters, calculate the mean and standard deviation of the parameters, and determine the normal value range based on the mean and standard deviation; The local anomaly factor algorithm is used to detect anomalies in real-time data. If continuous abnormal data points are detected, an equipment failure alarm is triggered; For the outlier data point, an interpolation method is selected according to the data variation characteristics, and an interpolation result is obtained to replace the outlier; Inputting the corrected data set into a visibility prediction algorithm, and obtaining a visibility distribution map according to the prediction algorithm; A sliding time window is set according to the visibility distribution map, and the visibility distribution map is updated through the sliding time window.

5. The method according to claim 1, wherein: The generated visibility distribution map is used to divide the search and rescue site into regions using a clustering algorithm, clustering regions with similar visibility into the same search and rescue sub-region, calculating the average visibility value and environmental characteristic coefficient of each sub-region, and establishing a corresponding relationship table between the search and rescue region and the visibility and environmental characteristics, including: Performing grid processing on the generated visibility distribution map, dividing the search and rescue site into square grid units, where the side length of the square grid unit is a preset value; According to the visibility data after gridding, the K-means clustering algorithm is used for clustering analysis. The K-means clustering algorithm calculates the Euclidean distance between grid cells and clusters grid cells with similar visibility into the same search and rescue sub-area. If there is a boundary area in the clustering result, the fuzzy C-means algorithm is used to optimize the boundary area, and the fuzzy C-means algorithm allows the grid unit to belong to multiple sub-areas at the same time; For each search and rescue sub-area obtained after clustering, the average visibility value of all grid cells in the sub-area is calculated, and an environmental feature vector is constructed in combination with the environmental parameters in the sub-area. The environmental feature vector includes the mean and standard deviation of the refractive index of raindrops, the mean and standard deviation of the absorptivity, and the mean and standard deviation of the light intensity; According to the average visibility value and the environmental feature vector, a corresponding relationship table between the search and rescue area and visibility and environmental features is established, wherein each row in the corresponding relationship table contains a sub-area ID, an average visibility value, an environmental feature vector and area boundary coordinates; It also includes: constructing a three-dimensional grid model based on the maritime characteristics and environmental factors of the search and rescue area, using an adaptive grid division algorithm to refine the complex wave area, coarsening the calm wave area, reflecting the complexity of the sea through the grid node density, and obtaining a maritime complexity distribution map of the search and rescue area.

6. The method according to claim 5, wherein: The three-dimensional grid model is constructed according to the maritime characteristics and environmental factors of the search and rescue area, and the complex wave area is refined by an adaptive grid division algorithm, and the calm wave area is coarsened. The complexity of the sea is reflected by the density of grid nodes, and the maritime complexity distribution map of the search and rescue area is obtained, including: Constructing an initial three-dimensional grid model based on the maritime characteristics and environmental factor data of the search and rescue area, dividing the search and rescue area into cubic grid units, and recording corresponding environmental parameters of wave height, wind speed and current speed in the cubic grid units; The octree algorithm is used to adaptively divide the initial grid, and the area with drastic changes in wave height is refined. If the difference in wave height between adjacent grid cells exceeds the preset threshold, the grid cells are recursively divided into smaller sub-cells until the minimum grid size is reached or the refinement threshold condition is met. Grid coarsening is performed on the calm wave area. If the wave height change of multiple consecutive grid cells is less than the preset threshold, these cells are merged into a large grid cell. Calculating the density of grid nodes in each area of ​​the processed grid model, and mapping the density value to a complexity level range, wherein the complexity level range is determined using the quartile method based on the overall data distribution; A maritime complexity distribution map of the search and rescue area is generated according to the complexity level, with different colors representing different complexity levels.

7. The method according to claim 1, wherein: The available search and rescue resource information is obtained from the preset search and rescue force database, including the type, quantity, performance parameters and current status of search and rescue vessels, helicopters and drone equipment, and a preliminary search and rescue force allocation plan is formulated in combination with the corresponding relationship table between the search and rescue area and visibility and environmental characteristics, including: Obtaining available search and rescue resource information in a preset search and rescue force database, wherein the available search and rescue resource information includes the type, quantity, performance parameters and real-time status of search and rescue vessels, helicopters and unmanned aerial vehicle equipment; According to the available search and rescue resource information, the search and rescue area is divided into 1 km × 1 km grids, and the environmental complexity index of each grid unit is calculated; Using the KM algorithm to optimally match the available search and rescue resources with the grid units, and obtaining a preliminary search and rescue force allocation plan; The preliminary search and rescue force allocation scheme is optimized by using a genetic algorithm, the population size, crossover rate and mutation rate are set, and the fitness function is defined as the weighted sum of the search and rescue coverage rate and the resource utilization rate; Through multiple generations of iterative calculations, the spatial distribution of search and rescue resources is optimized to obtain the final search and rescue force allocation plan; If the final search and rescue force allocation plan meets the preset coverage rate threshold, output the final search and rescue force allocation plan; If the final search and rescue force allocation plan does not meet the preset coverage threshold, the genetic algorithm optimization step is re-executed.

8. The method according to claim 1, wherein: The search and rescue force dispatch priority evaluation model is established, and the average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area are input, and the search and rescue difficulty coefficient and priority of each sub-area are calculated, and the sub-areas are sorted according to the priority, and the priority of the search and rescue force input is determined in combination with the preliminary search and rescue force allocation plan, and the final search and rescue plan is determined, including: The average visibility value, environmental characteristic coefficient, area size and geographical location factors of the sub-area are obtained as the input layer data of the neural network; Constructing a multilayer perceptron neural network according to the input layer data, wherein the multilayer perceptron neural network includes two hidden layers; Standardizing historical search and rescue data and using cross-validation methods to evaluate the performance of the neural network model; If the model performance meets the preset threshold, the back propagation algorithm is used to optimize the neural network parameters; The trained priority evaluation model is obtained by minimizing the mean square error loss function; Inputting the characteristic data of the sub-areas into the trained priority evaluation model, and using the Monte Carlo Dropout method to calculate the search and rescue difficulty coefficient and its uncertainty of each sub-area; Sort the sub-areas in descending order according to the search and rescue difficulty coefficient to obtain a search and rescue priority sorting list; Combine the search and rescue force allocation plan and the sub-area priority list, traverse the sub-areas in order of priority, allocate the minimum required resources to each area, until the resources are exhausted or all areas are covered, and determine the final search and rescue plan.

9. The method according to claim 1, wherein: The real-time communication network is constructed to connect the search and rescue command center with each search and rescue unit, and the final search and rescue plan and dispatch instructions are sent using an encrypted transmission protocol. After receiving the instructions, the search and rescue units send confirmation information to the command center. The system records the time when the instructions are issued and confirmed, and monitors the response speed of the search and rescue forces, including: Deploy a real-time communication system according to a 4G / 5G network, wherein the real-time communication system establishes a mesh topology structure and uses an OSPF protocol for dynamic routing to obtain an optimized data transmission path; The AES-256 encryption algorithm is used to encrypt the search and rescue plan and dispatch instructions, and the key exchange is performed through a 2048-bit RSA key to obtain the encrypted search and rescue plan and dispatch instructions; Develop a real-time message push mechanism based on the WebSocket protocol. The search and rescue command center sends the encrypted search and rescue plan and dispatch instructions to each search and rescue unit through the real-time message push mechanism. If the sending fails, it will automatically retry three times. Construct a distributed time synchronization system, which uses the NTP protocol and a two-way time synchronization algorithm, combined with round-trip transmission delay, to determine whether the clocks of each node are consistent; A time series database is used to store the instruction issuance time and confirmation time, and the response speed of the search and rescue force is obtained by calculating the time difference between the instruction issuance time and the confirmation time.

Images