Emergency rescue intelligent scheduling system and method based on Internet of Things

Abstract

The invention discloses an emergency rescue intelligent scheduling system and method based on the Internet of Things, and relates to the technical field of rescue scheduling analysis, and the method comprises the steps: scanning terrain point cloud through a three-dimensional laser radar carried by a scout car, and obtaining an aerial image, road condition sensor data and tent hospital coordinates through an unmanned aerial vehicle to build a space-time aligned initial disaster situation map; generating a multi-modal fusion data set with a timestamp; analyzing feature vectors based on the multi-modal fusion data set; analyzing a risk entropy based on the feature vector in combination with the confidence weight of each device, and analyzing the blocking probability of each grid unit based on the risk entropy; splitting a transportation planning path and an actual transportation path of the material supply truck, analyzing the similarity degree between a transportation planning path section and an actual transportation path section, and marking road sections according to the similarity degree; and updating the attribute data of the contained grid units based on each road segment mark. The accuracy of subsequent transportation path planning is improved, and the overall transportation efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rescue dispatch analysis, and specifically relates to an emergency rescue intelligent dispatch system and method based on the Internet of Things. BACKGROUND

[0002] In a public health emergency, natural disaster or major accident, medical emergency rescue often needs to race against time. After the tent hospital is deployed, the Internet of Things sensors in the tent and the vehicle dispatch system data interwork, real-time monitoring of medical resource consumption and triggering intelligent replenishment. When the material supply vehicle carries out regular road transportation and the transportation path is damaged due to the accident, the material supply vehicle cannot pass through, and the unmanned aerial vehicle carried is used for subsequent accurate delivery.

[0003] The current rescue dispatch relies on the initial data collection of the reconnaissance vehicle, so that the related parameters are fixed in the initial setting. However, the specific disaster situation changes in real time; if a mechanical mode of only planning without verification is adopted, it is easy to make the dispatch strategy difficult to adapt to the rapidly evolving road conditions in the disaster site, resulting in misallocation of rescue resources and time delay.

[0004] Therefore, the present application discloses an emergency rescue intelligent dispatch system and method based on the Internet of Things to solve the above problems. SUMMARY

[0005] The present application aims to provide an emergency rescue intelligent dispatch system and method based on the Internet of Things to solve the problems in the prior art.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: an emergency rescue intelligent dispatch method based on the Internet of Things, which comprises the following steps: S1: scanning the terrain point cloud by the three-dimensional laser radar carried by the reconnaissance vehicle, obtaining the aerial image by the unmanned aerial vehicle, the road condition sensor data and the tent hospital coordinate, building the initial disaster map of space-time alignment; generating a multi-modal fusion data set with a time stamp; S2: analyzing the feature vector based on the multi-modal fusion data set; analyzing the risk entropy based on the feature vector combined with the confidence weight of each device; analyzing the blocking probability of each grid unit based on the risk entropy; S3: splitting the transportation planning path and the transportation actual path of the material supply vehicle, analyzing the similarity between the transportation planning path segment and the transportation actual path segment, and marking the road segment according to the similarity; S4: updating the attribute data of the grid unit contained in each road segment mark.

[0007] According to the above scheme, in S1, the following contents are included: Point cloud data generated by the 3D LiDAR scanned by the reconnaissance vehicle and aerial imagery acquired by the reconnaissance drone are mapped to a unified geographic coordinate system through rigid transformation. Soil moisture and crack width data collected by road condition sensors and coordinates of the tent hospital are integrated simultaneously to construct an initial disaster base map with spatiotemporal alignment. The registered point cloud data, aerial imagery, and road condition sensor data are fused, and a Bayesian filtering algorithm is used to estimate the probability of traffic status in missing areas to generate a rasterized traffic map with confidence. A unified timestamp is added to the traffic map, point cloud data, aerial imagery, and road condition sensor data to output a multimodal fusion dataset.

[0008] This invention solves the spatial alignment problem of heterogeneous data from multiple sources, such as reconnaissance vehicles, drones, and road condition sensors, by unifying the coordinate system through rigid transformation, thus eliminating data silos. Adding a unified timestamp ensures the temporal consistency of disaster evolution, providing a structured multimodal dataset for dynamic decision-making; proactively estimating the probability of passability in missing areas improves the integrity of maps in disaster environments; and the confidence-based raster map intuitively quantifies the uncertainty of path risks, providing a quantitative basis for subsequent intelligent scheduling.

[0009] According to the above scheme, S2 includes the following: S201: The lidar point cloud is divided into grid cells using the voxel grid method; the elevation variance of each grid cell is analyzed based on the elevation data within each grid cell, and the elevation variance is recorded as the undulation of the grid cell; the aerial imagery from the reconnaissance UAV is input into a lightweight convolutional neural network to classify road and obstacle areas by pixels; the obstacle coverage rate of the grid cell is analyzed based on the pixel classification results; the obstacle coverage rate is equal to the ratio of the number of pixels corresponding to the obstacle area in the grid cell to the total number of pixels in the grid cell; the road condition deterioration coefficient is analyzed by weighted summation of soil moisture and crack width in the road condition sensing data; S202: Confidence weight w is calculated based on the signal-to-noise ratio and operating temperature of data collected from 3D lidar, reconnaissance drones, and road condition sensors, where w = SNR / [1 + exp(α1(temp-α2))]; where SNR represents the signal-to-noise ratio during data collection, temp represents the operating temperature during data collection, α1 and α2 represent confidence coefficients, and the confidence coefficients are preset constants; The undulation, obstacle coverage, road condition deterioration coefficient, and confidence weight are normalized. Based on the normalized undulation, obstacle coverage, road condition deterioration coefficient, and confidence weight, the risk entropy of each grid cell is analyzed: the risk entropy of the nth grid cell is denoted as H. r (V) n ); ; Among them, V n V represents the eigenvector of the nth grid cell.n =[delta n , eta n , lambda n ]; delta n represents the relief degree of the nth grid cell, eta n represents the roadblock coverage of the nth grid cell, lambda n represents the road condition deterioration coefficient of the nth grid cell; V n m represents the mth feature in the feature vector; w m represents the normalized value of the confidence weight of the mth feature; epsilon represents a zero prevention truncation constant; the risk entropy analysis is used to analyze the blocking probability P n of each grid cell; P n =exp[-gamma x H r (V n )]; wherein exp represents an exponential function with a natural number as a base, and gamma represents a first risk sensitivity adjustment factor; the risk sensitivity adjustment factor is a system preset constant.

[0010] The application comprehensively considers three indexes of terrain relief degree (laser radar), roadblock coverage (unmanned aerial vehicle image) and road condition deterioration coefficient (soil sensor), and comprehensively covers physical passing obstacles and environmental risks; a temperature-adaptive confidence weight is introduced: the data reliability is dynamically adjusted through a signal-to-noise ratio and an operating temperature (such as automatically reducing the weight when a sensor drifts caused by high temperature), the anti-interference ability of risk assessment is improved, and the accuracy of analysis is improved; the blocking probability model based on risk entropy conforms to the physical law of the high-risk path in a disaster environment.

[0011] According to the above scheme, in S3, the following content is included: S301: Obtain the transportation planning path and the transportation actual path of the material supply vehicle, extract the shortest straight line path distance in the transportation planning path and the transportation actual path, split the transportation planning path and the transportation actual path according to the shortest straight line path distance, sort based on the driving sequence, and generate a transportation planning path segment set and a transportation actual path segment set; S302: Based on the arrangement order of the transportation actual path segment, analyze the path similarity of the transportation actual path segment according to the arrangement order of the transportation planning path segment and the transportation planning path segment; if the path similarity of the transportation planning path segment and the transportation actual path segment is greater than the path similarity threshold, mark the transportation planning path segment as a path similar segment, and remove the transportation planning path segment marked as the path similar segment and the corresponding transportation actual path segment from the transportation planning path segment set and the transportation actual path segment set, respectively; traverse all the transportation actual path segments; mark the remaining transportation planning path segments in the transportation planning path segment set as path abnormal segments; mark the remaining transportation actual path segments in the transportation actual path segment set as path optimization segments; S303: Extract the predicted transport time of the transport planning path segment marked as the path similar road segment, extract the actual transport time of the transport actual path segment corresponding to the path similarity greater than the path similarity threshold, calculate the ratio of the predicted transport time and the actual transport time, and mark the transport planning path segment as the first abnormal time road segment if the time similarity is less than the first time similarity threshold, and mark the transport planning path segment as the second abnormal time road segment if the time similarity is greater than the second time similarity threshold; the first time similarity threshold is less than the second time similarity threshold.

[0012] The application aligns the driving sequence according to the shortest straight line distance, solves the trajectory distortion problem caused by the curved road and detour, focuses on the path anomaly by eliminating the normal road segment, and improves the analysis efficiency; the output "path abnormal road segment" and "path optimization road segment" labels provide targeted input for subsequent dynamic correction.

[0013] According to the above scheme, in S4, the following contents are included: S401: Extract all transport planning path segments marked as path abnormal road segments, extract all grid cells in the transport planning path segments marked as path abnormal road segments, and mark them as first abnormal grid cells; update the first risk sensitivity adjustment factor of the first abnormal grid cells to the second risk sensitivity adjustment factor; the second risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the first adjustment coefficient; Extract all transport planning path segments marked as the first abnormal time road segment, extract all grid cells in the transport planning path segments marked as the first abnormal time road segment, and mark them as the second abnormal grid cell; update the first risk sensitivity adjustment factor of the second abnormal grid cell to the third risk sensitivity adjustment factor; the third risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the second adjustment coefficient; Extract all transport planning path segments marked as the second abnormal time road segment, extract all grid cells in the transport planning path segments marked as the second abnormal time road segment, and mark them as the third abnormal grid cell; update the first risk sensitivity adjustment factor of the third abnormal grid cell to the fourth risk sensitivity adjustment factor; the fourth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the third adjustment coefficient; Wherein, the first adjustment coefficient is less than the second adjustment coefficient, which is less than one, which is less than the third adjustment coefficient; S402: Extract all the transport actual path segments marked as path optimization road segments, extract all the grid cells in the transport actual path segments marked as path optimization road segments, and mark them as first optimization grid cells, and update the first risk sensitivity adjustment factor of the first optimization grid cells to a fifth risk sensitivity adjustment factor; the fifth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and a fourth adjustment coefficient; the fourth adjustment coefficient is smaller than the third adjustment coefficient but larger than one.

[0014] The application extracts all the transport planning path segments marked as path abnormal road segments, first abnormal time road segments and second abnormal time road segments, wherein the path abnormal road segment represents that the transport planning path segment is not selected by the transport driver, there may be a real-time abnormal congestion situation, the corresponding congestion probability needs to be increased, and the selection of the road segment by the subsequent planning path needs to be reduced; the first abnormal time road segment is a transport time length increasing road segment, there may be a real-time abnormal congestion situation, the corresponding congestion probability also needs to be increased, and the selection of the road segment by the subsequent planning path needs to be reduced, and the situation should be better than that of the path abnormal road segment; the first abnormal time road segment is a transport time length reducing road segment, there may be a real-time traffic situation, the corresponding congestion probability needs to be reduced, and the selection of the road segment by the subsequent planning path needs to be increased; and the path optimization road segment is a road segment that is not selected at the beginning, and the corresponding congestion probability can be reduced based on the real-time situation of the transport driver; The application improves the accuracy of subsequent transport path planning and the overall transport efficiency by dynamically adjusting the risk sensitivity adjustment factor.

[0015] In another aspect of the application, an emergency rescue intelligent scheduling system based on the Internet of Things is provided, which is applied to the above-mentioned emergency rescue intelligent scheduling method based on the Internet of Things, and comprises a disaster data fusion module, a disaster situation feature analysis module, a transport path comparison module and a path attribute data updating module. The disaster data fusion module is used to scan the terrain point cloud by the three-dimensional laser radar carried by the reconnaissance vehicle, obtain aerial images by the unmanned aerial vehicle, obtain road condition sensor data, and construct an initial disaster map with space-time alignment based on the tent hospital coordinates; and a multi-modal fusion data set with a time stamp is generated. The disaster situation feature analysis module is used to analyze the feature vector based on the multi-modal fusion data set; analyze the risk entropy based on the feature vector and the confidence weight of each device; and analyze the congestion probability of each grid cell based on the risk entropy. The transport path comparison module is used to split the transport planning path and the transport actual path of the material supply vehicle, analyze the similarity between the transport planning path segments and the transport actual path segments, and mark the road segments according to the similarity. The path attribute data updating module is used to update the attribute data of the grid cells contained in each road segment based on the road segment marking.

[0016] According to the above scheme, the disaster situation data fusion module comprises a disaster situation data acquisition unit and a multi-modal data fusion unit; The disaster situation data acquisition unit is configured to generate point cloud data based on the three-dimensional laser radar carried by the reconnaissance vehicle, collect aerial images based on the reconnaissance unmanned aerial vehicle, and collect soil humidity and crack width data based on the road condition sensor; The multi-modal data fusion unit is configured to map the point cloud data generated by the three-dimensional laser radar carried by the reconnaissance vehicle and the aerial images obtained by the reconnaissance unmanned aerial vehicle to a unified geographic coordinate system through rigid transformation; synchronously integrate the soil humidity and crack width data collected by the road condition sensor and the tent hospital coordinates to construct an initial disaster situation base map with spatio-temporal alignment; fuse the registered point cloud data, aerial images and road condition sensing data, estimate the passing state probability of the missing area by using the Bayesian filtering algorithm, and generate a rasterized passing map with confidence; and attach a unified timestamp to the passing map, point cloud data, aerial images and road condition sensing data, and output a multi-modal fusion data set.

[0017] According to the above scheme, the disaster situation road condition feature analysis module comprises a disaster situation feature analysis unit and a blockage probability analysis unit; The disaster situation feature analysis unit is configured to analyze the relief degree, roadblock coverage rate and road condition deterioration coefficient of the grid unit based on the multi-modal fusion data set; The blockage probability analysis unit is configured to analyze the confidence weight corresponding to the signal-to-noise ratio and operating temperature of the data collected by the three-dimensional laser radar, the reconnaissance unmanned aerial vehicle and the road condition sensor; normalize the relief degree, roadblock coverage rate, road condition deterioration coefficient and confidence weight; analyze the risk entropy of each grid unit based on the normalized relief degree, roadblock coverage rate, road condition deterioration coefficient and confidence weight; and analyze the blockage probability of each grid unit based on the risk entropy.

[0018] According to the above scheme, the transport path comparison module comprises a path similarity analysis unit and a time similarity analysis unit; The path similarity analysis unit is configured to analyze the path similarity between the transport actual path segments and the transport planning path segments according to the arrangement order of the transport actual path segments; traverse all the transport actual path segments; mark the remaining transport planning path segments in the transport planning path segment set as path abnormal segments; and mark the remaining transport actual path segments in the transport actual path segment set as path optimization segments; The time similarity analysis unit is configured to extract the expected transport time of the transport planning path segments marked as path similar segments, and extract the actual transport time of the transport actual path segments with a path similarity greater than a path similarity threshold; calculate the ratio of the expected transport time to the actual transport time, denoted as time similarity; and mark abnormal time segments according to the time similarity.

[0019] According to the above scheme, the path attribute data updating module comprises a transportation planning path segment updating unit and a transportation actual path segment updating unit; The transportation planning path segment updating unit is configured to update the risk sensitivity adjustment factor of the grid cell in the transportation planning path segment marked as the path abnormal segment, the first abnormal time segment and the second abnormal time segment; The transportation actual path segment updating unit is configured to update the risk sensitivity adjustment factor of the grid cell in the transportation actual path segment marked as the path optimization segment.

[0020] Compared with the prior art, the beneficial effects of the present application are: the present application solves the spatial alignment problem of multi-source heterogeneous data such as reconnaissance vehicles, unmanned aerial vehicles, road condition sensors, etc. by rigid transformation of a unified coordinate system, and eliminates data islands. The additional unified timestamp ensures the temporal consistency of disaster evolution, provides a structured multi-modal data set for dynamic decision-making; actively estimates the missing area traffic state probability, improving the integrity of the map in the disaster environment; the confidence grid map intuitively quantifies the path risk uncertainty, providing a quantitative basis for subsequent intelligent scheduling; the present application comprehensively covers physical traffic obstacles and environmental risks by comprehensively considering three major indicators: terrain undulation, roadblock coverage rate and road condition deterioration coefficient; the temperature-adaptive confidence weight is introduced: by dynamically adjusting the data reliability through the signal-to-noise ratio and the operating temperature, the anti-interference ability of risk assessment is improved, and the accuracy of analysis is improved; the congestion probability model based on risk entropy conforms to the physical law of high-risk paths in disaster environments; the present application aligns the driving order by the shortest straight-line distance, solving the problem of trajectory distortion caused by curves and detours; the spatial similarity removes normal road segments and focuses on path abnormalities, improving analysis efficiency; the output of "path abnormal segment" and "path optimization segment" labels provides targeted input for subsequent dynamic correction; the present application improves the accuracy of subsequent transportation path planning by dynamically adjusting the risk sensitivity adjustment factor, and improves the overall transportation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate the present application together with the embodiments thereof, and explain the present application, but do not limit the present application. In the drawings: Figure 1 The flowchart of the present application is a flowchart of an intelligent scheduling method for emergency rescue based on the Internet of Things; Figure 2 The structure diagram of the present application is a structure diagram of an intelligent scheduling system for emergency rescue based on the Internet of Things. DETAILED DESCRIPTION

[0022] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0023] With reference to Figure 1 The present application provides a technical solution: an emergency rescue intelligent scheduling method based on Internet of Things, which comprises the following steps: S1: scanning the terrain point cloud by the three-dimensional laser radar carried by the reconnaissance vehicle, acquiring the aerial image by the unmanned aerial vehicle, acquiring the road condition sensor data and the tent hospital coordinate, building the initial disaster situation map in space-time alignment; generating the multi-modal fusion data set with time stamp; In S1, the following contents are included: The point cloud data generated by the three-dimensional laser radar carried by the reconnaissance vehicle and the aerial image acquired by the reconnaissance unmanned aerial vehicle are mapped to the unified geographic coordinate system through rigid transformation; the soil moisture and crack width data collected by the road condition sensor and the tent hospital coordinate are synchronously integrated to build the initial disaster situation base map in space-time alignment; the registered point cloud data, aerial image and road condition sensing data are fused, the Bayesian filtering algorithm is used to estimate the passing state probability of the missing area, and the rasterized passing map with confidence is generated; the unified time stamp is added to the passing map, point cloud data, aerial image and road condition sensing data, and the multi-modal fusion data set is output.

[0024] S2: analyzing the feature vector based on the multi-modal fusion data set; analyzing the risk entropy based on the feature vector and the confidence weight of each device; analyzing the blockage probability of each grid unit based on the risk entropy; In S2, the following contents are included: S201: dividing the laser radar point cloud into grid units by using the voxel grid method; analyzing the elevation variance of the grid unit based on the elevation data in each grid unit, and recording the elevation variance as the relief degree of the grid unit; inputting the aerial image of the reconnaissance unmanned aerial vehicle into a lightweight convolutional neural network to classify the pixels of the road and the obstacle area; analyzing the roadblock coverage rate of the grid unit based on the pixel classification result; the roadblock coverage rate is equal to the ratio of the number of pixels corresponding to the obstacle area in the grid unit to the total number of pixels in the grid unit; analyzing the road condition deterioration coefficient by using the weighted summation method for the soil moisture and crack width in the road condition sensing data; S202: Analyzing the confidence weight w based on the signal-to-noise ratio and operating temperature when collecting data by the three-dimensional laser radar, the reconnaissance unmanned aerial vehicle and the road condition sensor, w = SNR / [1 + exp (α1 (temp-α2))]; wherein SNR represents the signal-to-noise ratio when collecting data, temp represents the operating temperature when collecting data; α1 and α2 represent the confidence coefficient; the confidence coefficient is a preset constant; The relief degree, the roadblock coverage rate, the road condition deterioration coefficient and the confidence weight are normalized; the risk entropy of each grid cell is analyzed based on the normalized relief degree, the roadblock coverage rate, the road condition deterioration coefficient and the confidence weight: the risk entropy of the nth grid cell is denoted as H r (V n ); ; wherein, V n represents the feature vector of the nth grid cell, V n = [δ n , η n , λ n ]; δ n represents the relief degree of the nth grid cell, η n represents the roadblock coverage rate of the nth grid cell, and λ n represents the road condition deterioration coefficient of the nth grid cell; V n m represents the mth feature in the feature vector; w m represents the normalized value of the confidence weight of the mth feature; ε represents the zero-truncation constant; the blocking probability P n of each grid cell is analyzed based on the risk entropy; P n = exp [-γ × H r (V n )]; wherein exp represents the exponential function with natural number as the base, and γ represents the first risk sensitivity adjustment factor; the risk sensitivity adjustment factor is a system preset constant.

[0025] S3: Splitting the transportation planning path and the transportation actual path of the material supply vehicle, analyzing the similarity degree between the transportation planning path segment and the transportation actual path segment, and marking the path segment according to the similarity degree; In S3, the following contents are included: S301: Obtaining the transportation planning path and the transportation actual path of the material supply vehicle, extracting the shortest straight line path distance in the transportation planning path and the transportation actual path, splitting the transportation planning path and the transportation actual path according to the shortest straight line path distance, sorting based on the driving sequence, and generating the transportation planning path segment set and the transportation actual path segment set; S302: based on the arrangement order of the transportation actual path segments, the transportation actual path segments are analyzed for path similarity according to the arrangement order of the transportation planning path segments; if the path similarity of the transportation planning path segment and the transportation actual path segment is greater than the path similarity threshold value, the transportation planning path segment is marked as a path similar segment, and the transportation planning path segment and the corresponding transportation actual path segment marked as the path similar segment are removed from the transportation planning path segment set and the transportation actual path segment set respectively; all the transportation actual path segments are traversed; the remaining transportation planning path segments in the transportation planning path segment set are marked as path abnormal segments; the remaining transportation actual path segments in the transportation actual path segment set are marked as path optimization segments; S303: the predicted transportation time of the transportation planning path segment marked as the path similar segment is extracted, and the actual transportation time of the transportation actual path segment with the path similarity greater than the path similarity threshold value is extracted; the ratio of the predicted transportation time and the actual transportation time is calculated, which is recorded as the time similarity; if the time similarity is less than the first time similarity threshold value, the transportation planning path segment is marked as the first abnormal time segment; if the time similarity is greater than the second time similarity threshold value, the transportation planning path segment is marked as the second abnormal time segment; the first time similarity threshold value is less than the second time similarity threshold value.

[0026] Embodiment 1: In this embodiment, there is a transportation planning path segment set L={L1, L2, L3}, and a transportation actual path segment set S={S1, S2, S3}; In this embodiment, the path similarity of S1 and L1 is greater than the path similarity threshold value, the path similarity of S2 and L2 is greater than the path similarity threshold value, and the path similarity of S3 and L3 is less than the path similarity threshold value, so the transportation planning path segment L3 is marked as a path abnormal segment; the transportation actual path segment S3 is marked as a path optimization segment; In the transportation planning path, the take-off point of the original transportation unmanned aerial vehicle is located at the end of L3, but as the disaster changes in real time, L3 appears abnormal congestion; the transportation driver is forced to replace the L3 segment with the S3 segment; therefore, it is necessary to improve the congestion probability of the L3 segment in subsequent planning; and reduce the congestion probability of the S3 segment; so as to improve the accuracy of subsequent planning.

[0027] Embodiment 2: In this embodiment, there is a transportation planning path segment set L={L4, L5}, and a transportation actual path segment set S={S4, S5}; In this embodiment, the path similarity of S4 and L4 is greater than the path similarity threshold value, and the path similarity of S5 and L5 is greater than the path similarity threshold value; therefore, there is no path abnormal segment and path optimization segment; In this embodiment, the predicted transportation time of L4 and L5 is 1h, the actual transportation time of S4 is 0.5h, and the actual transportation time of S5 is 2h, so the time similarity of L4 and S4 is 2, and the time similarity of L5 and S5 is 0.5; In this embodiment, the first time similarity threshold is 0.8, and the second time similarity threshold is 1.2; Therefore, in this embodiment, the time similarity of L4 and S4 is greater than the second time similarity threshold, indicating that the transportation planning path segment L4 is the preferred path segment, and the transportation planning path segment L4 is marked as the second abnormal time path segment; The time similarity of L5 and S5 is less than the first time similarity threshold, indicating that the transportation planning path segment L5 is the non-preferred path segment, and the transportation planning path segment L4 is marked as the first abnormal time path segment; Therefore, it is necessary to improve the congestion probability of L5 in subsequent planning; reduce the congestion probability of L4; and improve the accuracy of subsequent planning.

[0028] S4: updating the attribute data of the grid cells contained in each road segment based on the marking.

[0029] In S4, the following contents are included: S401: extracting all transportation planning path segments marked as path abnormal road segments, extracting all grid cells in the transportation planning path segments marked as path abnormal road segments, and marking them as first abnormal grid cells, updating the first risk sensitivity adjustment factor of the first abnormal grid cells to a second risk sensitivity adjustment factor; the second risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and a first adjustment coefficient; extracting all transportation planning path segments marked as first abnormal time path segments, extracting all grid cells in the transportation planning path segments marked as first abnormal time path segments, and marking them as second abnormal grid cells, updating the first risk sensitivity adjustment factor of the second abnormal grid cells to a third risk sensitivity adjustment factor; the third risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and a second adjustment coefficient; extracting all transportation planning path segments marked as second abnormal time path segments, extracting all grid cells in the transportation planning path segments marked as second abnormal time path segments, and marking them as third abnormal grid cells, updating the first risk sensitivity adjustment factor of the third abnormal grid cells to a fourth risk sensitivity adjustment factor; the fourth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and a third adjustment coefficient; Wherein, the first adjustment coefficient is less than the second adjustment coefficient, which is less than one, which is less than the third adjustment coefficient; S402: Extract all the transport actual path segments marked as path optimization road segments, extract all the grid cells in the transport actual path segments marked as path optimization road segments, and mark the grid cells as first optimization grid cells; update the first risk sensitivity adjustment factor of the first optimization grid cells to a fifth risk sensitivity adjustment factor; the fifth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and a fourth adjustment coefficient; the fourth adjustment coefficient is smaller than the third adjustment coefficient but larger than one.

[0030] Referring to Figure 2 The application provides a technical scheme: an emergency rescue intelligent scheduling system based on an Internet of Things, which comprises a disaster data fusion module, a disaster situation feature analysis module, a transport path comparison module and a path attribute data updating module. The disaster data fusion module is used for constructing an initial disaster map in space-time alignment by scanning a terrain point cloud by a three-dimensional laser radar carried by a reconnaissance vehicle, acquiring aerial images by a reconnaissance unmanned aerial vehicle, collecting road condition sensor data and tent hospital coordinates; and generating a multi-modal fusion data set with a time stamp. The disaster situation feature analysis module is used for analyzing a feature vector based on the multi-modal fusion data set; analyzing a risk entropy based on the feature vector and a confidence weight of each device; and analyzing a blockage probability of each grid cell based on the risk entropy. The transport path comparison module is used for splitting a transport planning path and a transport actual path of a material supply vehicle, analyzing the similarity between the transport planning path segments and the transport actual path segments, and marking the road segments according to the similarity. The path attribute data updating module is used for updating attribute data of the grid cells contained in each road segment based on the road segment marking.

[0031] The disaster data fusion module comprises a disaster data acquisition unit and a multi-modal data fusion unit. The disaster data acquisition unit is used for generating point cloud data based on scanning by a three-dimensional laser radar carried by a reconnaissance vehicle, acquiring aerial images based on a reconnaissance unmanned aerial vehicle, and collecting soil humidity and crack width data based on a road condition sensor. The multi-modal data fusion unit is used for mapping the point cloud data generated by scanning by a three-dimensional laser radar carried by a reconnaissance vehicle and the aerial images acquired by a reconnaissance unmanned aerial vehicle to a unified geographic coordinate system through rigid transformation; synchronously integrating the soil humidity and crack width data collected by a road condition sensor and tent hospital coordinates to construct an initial disaster base map in space-time alignment; fusing the registered point cloud data, aerial images and road condition sensing data, estimating the passing state probability of a missing area by using a Bayesian filtering algorithm, and generating a rasterized passing map with a confidence degree; and attaching a unified time stamp to the passing map, the point cloud data, the aerial images and the road condition sensing data, and outputting a multi-modal fusion data set.

[0032] The disaster situation and road condition feature analysis module comprises a disaster situation feature analysis unit and a blockage probability analysis unit. The disaster situation feature analysis unit analyzes the fluctuation degree, roadblock coverage rate and road condition deterioration coefficient of the grid unit based on the multi-modal fusion data set. The blockage probability analysis unit is configured to analyze the corresponding confidence weight based on the signal-to-noise ratio and operating temperature when the three-dimensional laser radar, reconnaissance unmanned aerial vehicle and road condition sensor collect data; normalize the fluctuation degree, roadblock coverage rate, road condition deterioration coefficient and confidence weight; analyze the risk entropy of each grid unit based on the normalized fluctuation degree, roadblock coverage rate, road condition deterioration coefficient and confidence weight; and analyze the blockage probability of each grid unit based on the risk entropy.

[0033] The transport path comparison module comprises a path similarity analysis unit and a time similarity analysis unit. The path similarity analysis unit analyzes the path similarity between the transport actual path segment and the transport planning path segment according to the arrangement order of the transport planning path segment based on the arrangement order of the transport actual path segment; traverses all the transport actual path segments; marks the remaining transport planning path segments in the transport planning path segment set as path abnormal path segments; and marks the remaining transport actual path segments in the transport actual path segment set as path optimization path segments. The time similarity analysis unit is configured to extract the predicted transport time of the transport planning path segment marked as the path similar path segment, extract the actual transport time of the transport actual path segment with the corresponding path similarity greater than the path similarity threshold, calculate the ratio of the predicted transport time to the actual transport time, denoted as the time similarity, and mark the abnormal time path segment according to the time similarity.

[0034] The path attribute data updating module comprises a transport planning path segment updating unit and a transport actual path segment updating unit. The transport planning path segment updating unit is configured to update the risk sensitivity adjustment factor of the grid unit in the transport planning path segment marked as the path abnormal path segment, the first abnormal time path segment and the second abnormal time path segment. The transport actual path segment updating unit is configured to update the risk sensitivity adjustment factor of the grid unit in the transport actual path segment marked as the path optimization path segment.

[0035] It is to be understood that the phrases such as first and second, and the like, can refer to different entities or operations without necessarily implying any actual relationship or order between such entities or operations. Also, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0036] It will be apparent to those skilled in the art that the application is not limited to the details of the foregoing exemplary embodiments and that the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No feature of the application is considered critical unless expressly stated in the claims.

Claims

1. An intelligent dispatching method for emergency rescue based on the Internet of Things, characterized in that, The method includes the following steps: S1: Using a 3D lidar mounted on a reconnaissance vehicle to scan terrain point clouds, drones to acquire aerial images, road condition sensor data, and tent hospital coordinates to construct an initial disaster map with spatiotemporal alignment; generate a multimodal fusion dataset with timestamps; S2: Feature vector analysis based on multimodal fusion dataset; Risk entropy is analyzed based on feature vectors combined with the confidence weights of each device, and the blocking probability of each grid cell is analyzed based on risk entropy. S3: Decompose the planned transportation route and the actual transportation route of the supply vehicle, analyze the similarity between the planned transportation route segment and the actual transportation route segment, and mark the route segment according to the similarity. S4: Update the attribute data of the included grid cells based on the road segment markers.

2. The intelligent dispatching method for emergency rescue based on the Internet of Things according to claim 1, characterized in that: S1 includes the following: Point cloud data generated by the 3D LiDAR scanned by the reconnaissance vehicle and aerial imagery acquired by the reconnaissance drone are mapped to a unified geographic coordinate system through rigid transformation. Soil moisture and crack width data collected by road condition sensors and coordinates of the tent hospital are integrated simultaneously to construct an initial disaster base map with spatiotemporal alignment. The registered point cloud data, aerial imagery, and road condition sensor data are fused, and a Bayesian filtering algorithm is used to estimate the probability of traffic status in missing areas to generate a rasterized traffic map with confidence. A unified timestamp is added to the traffic map, point cloud data, aerial imagery, and road condition sensor data to output a multimodal fusion dataset.

3. The intelligent dispatching method for emergency rescue based on the Internet of Things according to claim 2, characterized in that: S2 includes the following: S201: The point cloud of the lidar is divided into grid cells using the voxel grid method; the elevation variance of the grid cell is analyzed based on the elevation data in each grid cell, and the elevation variance is recorded as the undulation of the grid cell; the aerial image of the reconnaissance UAV is input into a lightweight convolutional neural network to perform pixel classification of road and obstacle areas. The obstacle coverage rate of the grid cell is analyzed based on the pixel classification results; the obstacle coverage rate is equal to the ratio of the number of pixels corresponding to the obstacle area in the grid cell to the total number of pixels in the grid cell. The road condition deterioration coefficient was analyzed by weighted summation of soil moisture and crack width in road condition sensor data. S202: Confidence weight w is calculated based on the signal-to-noise ratio and operating temperature of data collected from 3D lidar, reconnaissance drones, and road condition sensors, where w = SNR / [1 + exp(α1(temp-α2))]; where SNR represents the signal-to-noise ratio during data collection, temp represents the operating temperature during data collection, α1 and α2 represent confidence coefficients, and the confidence coefficients are preset constants; The undulation, obstacle coverage, road condition deterioration coefficient, and confidence weight are normalized. Based on the normalized undulation, obstacle coverage, road condition deterioration coefficient, and confidence weight, the risk entropy of each grid cell is analyzed: the risk entropy of the nth grid cell is denoted as H. r (V) n Based on risk entropy analysis, the blocking probability P of each grid cell is analyzed. n ;P n =exp[-γ×H r (V) n ]; where exp represents an exponential function with a base of natural numbers, γ represents the first risk sensitivity adjustment factor; the risk sensitivity adjustment factor is a system preset constant.

4. The intelligent dispatching method for emergency rescue based on the Internet of Things according to claim 3, characterized in that: S3 includes the following: S301: Obtain the planned transportation route and actual transportation route of the supply vehicle, extract the shortest straight path distance between the planned transportation route and the actual transportation route, split the planned transportation route and the actual transportation route according to the shortest straight path distance, sort them based on the driving order, and generate a set of planned transportation route segments and a set of actual transportation route segments. S302: Based on the arrangement order of the actual transportation route segments, perform a route similarity analysis between the actual transportation route segments and the planned transportation route segments according to their arrangement order. If the path similarity between the planned transportation route segment and the actual transportation route segment is greater than the path similarity threshold, the planned transportation route segment is marked as a path similarity segment, and the planned transportation route segment marked as a path similarity segment and the corresponding actual transportation route segment are removed from the set of planned transportation route segments and the set of actual transportation route segments, respectively; all actual transportation route segments are traversed; the remaining planned transportation route segments in the set of planned transportation route segments are marked as path abnormality segments; Mark the remaining actual transportation route segments in the actual transportation route segment set as route optimization segments; S303: Extract the estimated transportation time of the transportation planning route segments marked as similar road segments, and extract the actual transportation time of the corresponding actual transportation route segments with a path similarity greater than the path similarity threshold; Calculate the ratio of the estimated transportation time to the actual transportation time, and record it as the time similarity. If the time similarity is less than the first time similarity threshold, the transportation planning route segment is marked as the first abnormal time segment. If the time similarity is greater than the second time similarity threshold, the transportation planning route segment is marked as the second abnormal time segment. The first time similarity threshold is less than the second time similarity threshold.

5. The intelligent dispatching method for emergency rescue based on the Internet of Things according to claim 4, characterized in that: S4 includes the following: S401: Extract all transportation planning route segments marked as abnormal routes, extract all grid cells in the transportation planning route segments marked as abnormal routes, and mark them as first abnormal grid cells. Update the first risk sensitivity adjustment factor of the first abnormal grid cell to a second risk sensitivity adjustment factor; the second risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the first adjustment coefficient. Extract all transportation planning route segments marked as first abnormal time segments, extract all grid cells in all transportation planning route segments marked as first abnormal time segments, and mark them as second abnormal grid cells. Update the first risk sensitivity adjustment factor of the second abnormal grid cells to a third risk sensitivity adjustment factor; the third risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the second adjustment coefficient. Extract all transportation planning route segments marked as second abnormal time segments, extract all grid cells in all transportation planning route segments marked as second abnormal time segments, and mark them as third abnormal grid cells. Update the first risk sensitivity adjustment factor of the third abnormal grid cells to a fourth risk sensitivity adjustment factor; the fourth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the third adjustment coefficient. Among them, the first adjustment coefficient is less than the second adjustment coefficient, which is less than one, which is less than the third adjustment coefficient; S402: Extract all actual transportation route segments marked as route optimization segments, extract the grid cells in all actual transportation route segments marked as route optimization segments, and mark them as first optimized grid cells. Update the first risk sensitivity adjustment factor of the first optimized grid cell to the fifth risk sensitivity adjustment factor; the fifth risk sensitivity adjustment factor is equal to the product of the first risk sensitivity adjustment factor and the fourth adjustment coefficient; the fourth adjustment coefficient is less than the third adjustment coefficient but greater than one.

6. An Internet of Things (IoT)-based intelligent emergency rescue dispatch system, wherein the system is applied to the IoT-based intelligent emergency rescue dispatch method described in any one of claims 1-5, characterized in that, The system includes a disaster data fusion module, a disaster road condition feature analysis module, a transportation route comparison module, and a route attribute data update module; The disaster data fusion module is used to construct an initial disaster map with spatiotemporal alignment by scanning terrain point clouds with a 3D lidar mounted on a reconnaissance vehicle, acquiring aerial images from drones, using road condition sensor data and the coordinates of the tent hospital; and to generate a multimodal fusion dataset with timestamps. The disaster and road condition feature analysis module is used to analyze feature vectors based on a multimodal fusion dataset; based on The risk entropy is analyzed by combining the feature vectors with the confidence weights of each device, and the blocking probability of each grid cell is analyzed based on the risk entropy. The transportation route comparison module is used to separate the planned transportation route and the actual transportation route of the material supply vehicle, analyze the similarity between the planned transportation route segment and the actual transportation route segment, and mark the route segment according to the similarity. The path attribute data update module is used to update the attribute data of the included grid cells based on the road segment markers.

7. The IoT-based intelligent dispatch system for emergency rescue as described in claim 6, characterized in that: The disaster data fusion module includes a disaster data acquisition unit and a multimodal data fusion unit; The disaster data acquisition unit is used to collect point cloud data based on the three-dimensional lidar scanning generated by the reconnaissance vehicle, aerial images collected by the reconnaissance drone, and soil moisture and crack width data collected by the road condition sensor. The multimodal data fusion unit is used to map point cloud data generated by the 3D LiDAR scanned by the reconnaissance vehicle and aerial images acquired by the reconnaissance UAV to a unified geographic coordinate system through rigid transformation; simultaneously integrate soil moisture and crack width data collected by road condition sensors and the coordinates of the tent hospital to construct an initial disaster base map with spatiotemporal alignment; fuse the registered point cloud data, aerial images and road condition sensor data, use Bayesian filtering algorithm to estimate the probability of traffic status in missing areas, and generate a rasterized traffic map with confidence; add a unified timestamp to the traffic map, point cloud data, aerial images and road condition sensor data, and output a multimodal fusion dataset.

8. The IoT-based intelligent dispatch system for emergency rescue as described in claim 6, characterized in that: The disaster and road condition feature analysis module includes a disaster feature analysis unit and a congestion probability analysis unit; The disaster characteristic analysis unit analyzes the undulation, road obstruction coverage, and road condition deterioration coefficient of the grid unit based on the multimodal fusion dataset; The blockage probability analysis unit is used to analyze the confidence weights corresponding to the signal-to-noise ratio and operating temperature when data is collected from 3D lidar, reconnaissance drones and road condition sensors; normalize the undulation, road obstacle coverage, road condition deterioration coefficient and confidence weights; and analyze the risk entropy of each grid cell based on the normalized undulation, road obstacle coverage, road condition deterioration coefficient and confidence weights. The blocking probability of each grid cell is analyzed based on risk entropy.

9. The IoT-based intelligent dispatch system for emergency rescue as described in claim 6, characterized in that: The transportation route comparison module includes a route similarity analysis unit and a time similarity analysis unit; The path similarity analysis unit performs path similarity analysis between the actual transportation path segments and the planned transportation path segments, based on the arrangement order of the actual transportation path segments; it traverses all actual transportation path segments; and marks the remaining planned transportation path segments in the set of planned transportation path segments as abnormal path segments. Mark the remaining actual transportation route segments in the actual transportation route segment set as route optimization segments; The time similarity analysis unit is used to extract the estimated transportation time of transportation planning route segments marked as path similarity segments, and to extract the actual transportation time of the corresponding actual transportation route segments whose path similarity is greater than the path similarity threshold. Calculate the ratio of the estimated transportation time to the actual transportation time, and record it as the time similarity; mark abnormal time segments based on the time similarity.

10. The IoT-based intelligent dispatch system for emergency rescue according to claim 6, characterized in that: The path attribute data update module includes a transportation planning path segment update unit and a transportation actual path segment update unit; The transportation planning route segment update unit is used to update the risk sensitivity adjustment factor of the grid cells in the transportation planning route segment marked as abnormal route segment, first abnormal time segment and second abnormal time segment. The actual transport route segment update unit is used to update the risk sensitivity adjustment factor of the grid cells in the actual transport route segment marked as a route optimization segment.

Images