Action management identification method based on earthquake scene disaster investigation

Abstract

The invention discloses an action management identification method based on earthquake site disaster investigation, and relates to the technical field of data processing, and the method comprises the steps: determining a to-be-acted earthquake site region, and planning a disaster search task corresponding to a plurality of unmanned aerial vehicles in the to-be-acted earthquake site region; issuing the disaster situation search task to multiple unmanned aerial vehicles, controlling the unmanned aerial vehicles to execute the disaster situation search task, and obtaining an actual disaster situation investigation result corresponding to the disaster situation search task; according to an actual disaster situation investigation result, generating disaster situation points on the electronic map corresponding to the earthquake field area to be moved, and obtaining a disaster situation investigation map; and finally, the disaster situation investigation map is fed back to equipment specified by a commander to guide disaster relief actions of the commander, so that the earthquake site disaster situation investigation efficiency is improved, comprehensive disaster situation data can be formed, the commander can comprehensively understand the disaster situation, the commander is assisted to quickly form a global and optimal action scheme, and the disaster relief efficiency of the commander is improved. And the disaster disposal efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to an action management and identification method based on earthquake site disaster investigation. Background Art

[0002] Earthquakes are extremely destructive natural disasters, often causing severe consequences such as casualties, building collapses, road closures, and frequent secondary disasters in a short period of time. After an earthquake strikes, quickly and accurately assessing the damage and formulating effective emergency response management strategies based on this information are crucial for saving lives, minimizing losses, and maintaining social order. Traditional earthquake emergency response management relies heavily on experienced commanders making decisions based on limited and delayed information. Existing technologies generally rely on disaster relief personnel conducting manual surveys at the earthquake site. After the survey is completed, rescue efforts are carried out on-site or feedback is provided to command personnel for centralized processing. Command personnel face the task of processing a large amount of fragmented information, making it difficult to quickly formulate a comprehensive and optimal action plan. This can lead to delays and misalignments in the deployment of emergency resources (such as rescue teams, medical supplies, and construction machinery), missing the optimal rescue opportunity, and compromising overall rescue efficiency. Summary of the Invention

[0003] The present application aims to provide an action management and identification method based on earthquake site disaster investigation, aiming to solve the problem of low rescue efficiency caused by manual investigation of earthquake site disaster in the existing technology.

[0004] This application provides an action management and identification method based on earthquake site disaster investigation, including: Determine the earthquake scene area to be acted upon, and plan disaster search missions for multiple UAVs corresponding to the earthquake scene area to be acted upon; Sending the disaster search mission to multiple drones, controlling the drones to execute the disaster search mission, and obtaining actual disaster investigation results corresponding to the disaster search mission; Generating disaster points on an electronic map corresponding to the earthquake site area to be acted upon based on actual disaster investigation results corresponding to the disaster search task, thereby obtaining a disaster investigation map; The disaster investigation map is fed back to a device designated by a commander to guide the commander's disaster relief actions and complete action management based on the earthquake site disaster investigation.

[0005] In a possible implementation, determining the earthquake site area to be operated and planning the disaster search mission corresponding to the earthquake site area to be operated by multiple drones include: Obtain the closed area determined by the disaster relief personnel on the electronic map through human-computer interaction to obtain the earthquake scene area to be acted upon; Acquiring power transmission line information and road information in the area of ​​the earthquake site to be acted upon; generating a first disaster search task for conducting a disaster investigation along the transmission lines based on the transmission line information of the earthquake scene area to be acted upon; generating a second disaster search task for disaster investigation along the road based on the road information; With the goal of traversing the earthquake scene area to be acted upon, a third disaster search task is generated for conducting a disaster investigation on trapped persons and collapsed houses in the earthquake scene area to be acted upon.

[0006] In a possible implementation, the dispatching of the disaster search mission to multiple drones, controlling the drones to execute the disaster search mission, and obtaining actual disaster investigation results corresponding to the disaster search mission include: Sending the first disaster search mission to a first drone equipped with a power transmission line detection model, and controlling the first drone to perform target detection along the power transmission line to obtain a power transmission line survey result; Sending the second disaster search mission to a second drone equipped with a road detection model, and controlling the second drone to perform target detection along the road to obtain road survey results; The third disaster search mission is sent to a third drone equipped with a personnel and housing detection model, and the third drone is controlled to traverse the earthquake site area to be operated to obtain personnel and housing investigation results; The personnel and house survey results, road survey results, and power line survey results are taken together as actual disaster survey results corresponding to the disaster search task.

[0007] In a possible implementation, the power transmission line detection model, the road detection model, and the person and house detection model are all configured as YOLO models.

[0008] In one possible implementation, the transmission line detection model, the road detection model, and the person / house detection model are installed in the same manner, and all include: A hybrid mapping strategy is used to initialize the model to be carried out to obtain multiple parameter vectors; wherein the model to be carried out is a power transmission line detection model, a road detection model or a person and house detection model; Obtaining the fitness corresponding to each of the parameter vectors, and obtaining the optimal parameter vector according to the fitness of the parameter vectors; Based on the optimal parameter vector, performing an adaptive search based on distribution uniformity on the parameter vector to obtain a parameter vector after adaptive search; Performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain a parameter vector after the volatility spiral chain collaborative search; Performing a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search to obtain a parameter vector after the bidirectional mutation search; Determine whether the current number of training times has reached the maximum number of training times. If so, re-obtain the optimal parameter vector based on the parameter vector after the bidirectional mutation search to obtain the target parameter vector. Otherwise, return to the step of obtaining the optimal parameter vector and enter the next training process. The target parameter vector is used as the final parameter of the model to be mounted, and the model to be mounted is deployed on the UAV according to the final parameter of the model to be mounted.

[0009] In a possible implementation, the hybrid mapping strategy is used to initialize the model to be loaded to obtain multiple parameter vectors, including: Randomly initializing the model parameters of the model to be loaded between an upper limit and a lower limit, and encoding the initialized model parameters into a vector to obtain a basic vector; Based on the basic vector, a plurality of other vectors are obtained by adopting a chaotic mapping sequence; A parameter vector is generated based on the other vectors, the upper limit of the model parameters, and the lower limit of the model parameters, and all other vectors are processed to obtain multiple parameter vectors.

[0010] In a possible implementation, obtaining the fitness corresponding to each parameter vector and obtaining the optimal parameter vector according to the fitness of the parameter vector includes: For any of the parameter vectors, obtain the loss function value corresponding to the parameter vector, and add the loss function value to a preset non-zero constant term and then take the inverse to obtain the fitness corresponding to the parameter vector; The parameter vector with the largest fitness is taken as the optimal parameter vector.

[0011] In a possible implementation, performing an adaptive search on the parameter vector based on distribution uniformity based on the optimal parameter vector to obtain a parameter vector after the adaptive search includes: Arrange the parameter vectors in ascending order of fitness, take the arranged parameter vectors as a basis, and obtain an adaptive balance search factor according to the fitness of the optimal parameter vector, the fitness of the parameter vector, and the current number of training times; Obtaining the Euclidean distance between adjacent parameter vectors and determining the distribution uniformity based on the Euclidean distance between adjacent parameter vectors; When the distribution uniformity is less than a preset uniformity threshold, searching the parameter vector according to the optimal parameter vector and the adaptive balance search factor to obtain a parameter vector after adaptive search; When the distribution uniformity is greater than or equal to a preset uniformity threshold, the parameter vector is searched according to the model parameter upper limit and the model parameter lower limit to obtain a parameter vector after adaptive search.

[0012] In a possible implementation, performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain the parameter vector after the volatility spiral chain collaborative search includes: Determine the waveform adjustment factor based on the current number of training times; When the volatility adjustment factor is greater than or equal to a preset adjustment threshold, performing a spiral chain collaborative search on the parameter vector after the adaptive search according to the optimal parameter vector to obtain a parameter vector after the volatility spiral chain collaborative search; When the volatility adjustment factor is less than the preset adjustment threshold, a random parameter vector is determined in the parameter vector after the adaptive search, and a spiral chain collaborative search is performed based on the parameter vector after the adaptive search of the random parameter vector to obtain the parameter vector after the volatility spiral chain collaborative search.

[0013] In a possible implementation, performing a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search to obtain the parameter vector after the bidirectional mutation search includes: Obtain the number of mutation dimensions, forward mutation factors, and reverse mutation factors based on the current number of training times; Determining, based on the number of mutation dimensions, the dimensions to be mutated of the parameter vector after the volatility spiral chain collaborative search; For any dimension to be mutated, a forward mutation factor is used to perform a forward mutation on the parameter vector after the volatility spiral chain collaborative search, or a reverse mutation factor is used to perform a reverse mutation on the parameter vector after the volatility spiral chain collaborative search, to obtain the parameter vector after the bidirectional mutation search.

[0014] Beneficial effects: The present application provides an action management and identification method based on earthquake site disaster investigation, the method comprising: determining the earthquake site area to be acted upon, and planning the disaster search tasks corresponding to the earthquake site area to be acted upon by multiple drones; then dispatching the disaster search tasks to multiple drones, and controlling the drones to execute the disaster search tasks, and obtaining actual disaster investigation results corresponding to the disaster search tasks; then, based on the actual disaster investigation results, generating disaster points on an electronic map corresponding to the earthquake site area to be acted upon, and obtaining a disaster investigation map; finally, feeding back the disaster investigation map to the equipment designated by the commander to guide the commander's disaster relief actions, which not only improves the efficiency of earthquake site disaster investigations, but also forms comprehensive disaster data, so that the commander can fully understand the disaster situation, assist the commander in quickly forming a global and optimal action plan, and improve the efficiency of disaster handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0016] Figure 1 This is a flowchart of an action management identification method based on earthquake site disaster investigation proposed in one embodiment of the present application.

[0017] Figure 2 This is a structural diagram of an action management and identification device based on earthquake site disaster investigation proposed in one embodiment of the present application.

[0018] Figure 3 This is a structural diagram of an electronic device proposed in one embodiment of the present application.

[0019] Explanation of the reference numerals: 201 - task planning module, 202 - disaster investigation module, 203 - disaster summary module, 204 - disaster feedback module, 301 - memory, 302 - processor, 303 - communication bus. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0021] In existing technologies, disaster relief workers typically conduct manual surveys at earthquake sites. After conducting on-site rescue operations, they provide feedback to command personnel for centralized processing. Command personnel must process a large amount of fragmented information, making it difficult to quickly develop a comprehensive and optimal action plan. This can lead to delays and misalignments in the deployment of emergency resources (such as rescue teams, medical supplies, and construction machinery), missing optimal rescue opportunities, and compromising overall rescue efficiency.

[0022] In view of this, the embodiment of the present application proposes an action management and identification method based on earthquake site disaster investigation, which combines deep learning technology with the collaborative execution of tasks by multiple drones to investigate the earthquake site disaster, and finally summarizes the actual disaster investigation results into a disaster investigation map. It not only improves the efficiency of earthquake site disaster investigation, but also can form comprehensive disaster data, so that commanders can have a comprehensive understanding of the disaster situation, assist commanders to quickly form a global and optimal action plan, and improve the efficiency of disaster handling.

[0023] like Figure 1 As shown, the embodiment of the present application provides an action management and identification method based on earthquake site disaster investigation, comprising: S101: Determine an earthquake scene area to be operated, and plan disaster search missions corresponding to multiple UAVs in the earthquake scene area to be operated.

[0024] The earthquake scene area to be operated can be a closed area determined by the commander on the electronic map through human-computer interaction. By planning the investigation routes of multiple drones in the earthquake scene area to be operated, a disaster search mission can be obtained.

[0025] S102: Send the disaster search mission to multiple drones, control the drones to execute the disaster search mission, and obtain actual disaster investigation results corresponding to the disaster search mission.

[0026] Different disaster search tasks can be assigned to different drones, so that multiple drones can collaborate to conduct earthquake site disaster investigations and ultimately determine the actual disaster investigation results corresponding to different disaster search tasks.

[0027] S103 , generating disaster points on an electronic map corresponding to the earthquake site area to be acted upon based on actual disaster investigation results corresponding to the disaster search task, thereby obtaining a disaster investigation map.

[0028] The actual disaster investigation results may include disaster photos, location and disaster classification labels, and generate points of interest on the electronic map. Each point of interest records the disaster photos, location and disaster classification labels, thereby generating disaster points on the electronic map corresponding to the earthquake site area to be acted upon.

[0029] S104: Feedback the disaster investigation map to a device designated by a commander to guide the commander's disaster relief actions and complete action management based on the earthquake site disaster investigation.

[0030] By feeding the disaster investigation map back to the equipment designated by the commander, comprehensive disaster data can be generated, allowing the commander to fully understand the disaster situation, assisting the commander to quickly form a global and optimal action plan, and improving the efficiency of disaster handling.

[0031] In a possible implementation, determining the earthquake site area to be operated and planning the disaster search mission corresponding to the earthquake site area to be operated by multiple drones include: Obtain the closed area determined by the disaster relief personnel on the electronic map through human-computer interaction to obtain the earthquake scene area to be acted upon.

[0032] Acquire power transmission line information and road information of the earthquake scene area to be acted upon.

[0033] A first disaster search task for conducting a disaster investigation along the transmission lines is generated based on the transmission line information of the earthquake scene area to be acted upon.

[0034] The transmission line information of the earthquake scene area to be acted upon can be obtained from a power grid data system or a third-party map, and then a first disaster search task is generated to conduct a disaster investigation along the transmission lines in the earthquake scene area to be acted upon.

[0035] A second disaster search task is generated based on the road information to conduct a disaster investigation along the road.

[0036] Road information can be obtained from a third-party map, and then a second disaster search task is generated to conduct a disaster investigation along the roads within the earthquake site area to be operated.

[0037] With the goal of traversing the earthquake scene area to be acted upon, a third disaster search task is generated for conducting a disaster investigation on trapped persons and collapsed houses in the earthquake scene area to be acted upon.

[0038] Trapped people and collapsed buildings are key areas of earthquake relief efforts. Therefore, it's necessary to generate a search route that traverses the area of ​​the earthquake site to be addressed, creating the third disaster search task. For example, an S-shaped search route can be generated, allowing the drone to capture images of all areas within the earthquake site for search and rescue operations.

[0039] It is worth noting that the embodiment of the present application does not consider the problem of drone endurance. The problems of drone return and endurance can be solved using existing technologies.

[0040] In a possible implementation, the dispatching of the disaster search mission to multiple drones, controlling the drones to execute the disaster search mission, and obtaining actual disaster investigation results corresponding to the disaster search mission include: The first disaster search mission is sent to a first UAV equipped with a transmission line detection model, and the first UAV is controlled to perform target detection along the transmission line to obtain a transmission line survey result.

[0041] It is worth noting that the transmission line detection model is mainly used to detect the collapse of facilities on the transmission line, such as towers and transformers. The transmission line detection model can be pre-trained so that it can identify normal facilities and collapsed facilities.

[0042] The second disaster search mission is sent to a second UAV equipped with a road detection model, and the second UAV is controlled to perform target detection along the road to obtain a road survey result.

[0043] It's worth noting that the road detection model primarily detects road breaks. This model can be pre-trained to identify both normal and broken roads. However, both roads and power lines may have bifurcations. This reduces the search effort by limiting the search to the main road or transmission line. Alternatively, a return subtask can be assigned to each bifurcated node. This means that after searching along a branch road, the drone must return to the corresponding bifurcation node and continue investigating the remaining unsearched sections.

[0044] The third disaster search mission is sent to a third drone equipped with a personnel and house detection model, and the third drone is controlled to traverse the earthquake site area to be operated to obtain personnel and house investigation results.

[0045] It is worth noting that the personnel and house detection model is mainly used to detect trapped people and collapsed houses. The personnel and house detection model can be pre-trained so that the personnel and house detection model can identify non-trapped people, trapped people, normal houses and collapsed houses.

[0046] The personnel and house survey results, road survey results, and power line survey results are taken together as actual disaster survey results corresponding to the disaster search task.

[0047] When a drone identifies the target corresponding to its mission, it should record disaster photos, location and disaster classification labels to generate disaster points on the electronic map.

[0048] In a possible implementation, the power transmission line detection model, the road detection model, and the person and house detection model are all configured as YOLO models.

[0049] In one possible implementation, the transmission line detection model, the road detection model, and the person / house detection model are installed in the same manner, and all include: A hybrid mapping strategy is used to initialize the model to be loaded, thereby obtaining multiple parameter vectors, wherein the model to be loaded is a power transmission line detection model, a road detection model, or a person and house detection model.

[0050] Optionally, the trainable parameters of the YOLO model are typically the connection weights between each network layer. These connection weights can be initialized to determine multiple parameter vectors. For example, the model parameters can be randomly initialized between upper and lower bounds, and the randomly initialized model parameters can be encoded into a vector to obtain a parameter vector. This can be repeated multiple times to obtain multiple parameter vectors.

[0051] Obtain the fitness corresponding to each of the parameter vectors, and obtain the optimal parameter vector according to the fitness of the parameter vectors.

[0052] Based on the optimal parameter vector, an adaptive search based on distribution uniformity is performed on the parameter vector to obtain a parameter vector after the adaptive search.

[0053] Performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain a parameter vector after the volatility spiral chain collaborative search.

[0054] A bidirectional mutation search is performed on the parameter vector after the volatility spiral chain collaborative search to obtain a parameter vector after the bidirectional mutation search.

[0055] Determine whether the current number of training times has reached the maximum number of training times. If so, re-obtain the optimal parameter vector based on the parameter vector after the bidirectional mutation search to obtain the target parameter vector. Otherwise, return to the step of obtaining the optimal parameter vector and enter the next training process.

[0056] The target parameter vector is used as the final parameter of the model to be mounted, and the model to be mounted is deployed on the UAV according to the final parameter of the model to be mounted.

[0057] In the prior art, the model parameters of the YOLO model are usually trained using a gradient descent intelligent optimization algorithm or a particle swarm optimization algorithm. However, regardless of which algorithm is used, it will fall into a local optimum during the training process, resulting in the trained YOLO model being unable to accurately complete the recognition task. Therefore, this application provides a new training algorithm to improve the global search capability while ensuring the local fine search capability, thereby improving the target recognition accuracy of the YOLO model and ultimately improving the accuracy of disaster investigation.

[0058] In a possible implementation, the hybrid mapping strategy is used to initialize the model to be loaded to obtain multiple parameter vectors, including: Randomly initializing the model parameters of the model to be loaded between an upper limit and a lower limit, and encoding the initialized model parameters into a vector to obtain a basic vector; Based on the basic vector, a plurality of other vectors are obtained by using a chaotic mapping sequence: in, Indicates the i The other vectors j dimensional model parameters, and i =1, represents the j-th dimension model parameter of the basis vector, j =1,2,…,M, where M represents the total dimension of model parameters. Indicates the i +1 other vector's j dimensional model parameters, represents the chaotic mapping parameter, which is set to 0.2. For the YOLO model, its model parameters can be the connection weights between different network layers, and these connection weights are between [0, 1]. Therefore, based on this, a chaotic mapping sequence can be used to obtain multiple other vectors.

[0059] Based on the other vectors, the upper limit of the model parameters and the lower limit of the model parameters, the parameter vector is generated as follows: in, Indicates the i The parameter vector j dimensional model parameters, Indicates the i The intermediate vector j dimensional model parameters, Indicates the i The reverse learning vector corresponding to the intermediate vector j dimensional model parameters, Represents the fitness corresponding to the intermediate vector, Represents the fitness corresponding to the reverse learning vector, Indicates the j The upper limit of the model parameters corresponding to the dimension model parameters, Indicates the j The lower limit of the model parameters corresponding to the dimension model parameters; All other vectors are processed to obtain multiple parameter vectors.

[0060] This application adopts a hybrid mapping strategy to initialize the model, which can make the initial solution more evenly distributed in the solution space, and uses reverse learning to select the best one, further improving the training speed of the algorithm.

[0061] In a possible implementation, obtaining the fitness corresponding to each parameter vector and obtaining the optimal parameter vector according to the fitness of the parameter vector includes: For any of the parameter vectors, obtain the loss function value corresponding to the parameter vector, and add the loss function value to a preset non-zero constant term (such as 0.0001) and take the inverse to obtain the fitness corresponding to the parameter vector; The parameter vector with the largest fitness is taken as the optimal parameter vector.

[0062] The smaller the loss function value, the better the position of the parameter vector in the solution space. Therefore, it can be determined that the greater the fitness, the better the position of the parameter vector in the solution space. Therefore, the parameter vector with the largest fitness can be determined as the optimal parameter vector.

[0063] During the training process of the transmission line detection model, the road detection model, and the personnel and house detection model, the corresponding normal data and the target data to be detected can be used for acquisition. For example, in the process of training the transmission line detection model, image data with normal towers and / or transformers (i.e., normal data) and image data with collapsed towers and / or transformers (i.e., target data to be detected) can be used for training. Obtaining the loss function of the YOLO model is a relatively conventional technology and will not be repeated in the embodiments of this application.

[0064] In a possible implementation, performing an adaptive search on the parameter vector based on distribution uniformity based on the optimal parameter vector to obtain a parameter vector after the adaptive search includes: Arrange the parameter vectors in order of fitness from small to large, and based on the arranged parameter vectors, and according to the fitness of the optimal parameter vector, the fitness of the parameter vector, and the current number of training times, obtain the adaptive balance search factor: in, Indicates the t During the training m The adaptive balance search factor corresponding to the parameter vector after permutation, represents the preset maximum value of the adaptive balance search factor and is set to 0.9; represents a natural constant, t Indicates the current number of training times, T indicates the preset maximum number of training times, Indicates thet During the training m parameter vector, represents the optimal parameter vector, represents the worst parameter vector, Represents the parameter vector The fitness of represents the optimal parameter vector The fitness of Represents the worst parameter vector The fitness of represents the total number of parameter vectors, lg represents the logarithmic function; Get the Euclidean distance between adjacent parameter vectors, and determine the distribution uniformity based on the Euclidean distance between adjacent parameter vectors: in, Indicates the uniformity of distribution, Represents the qth Euclidean distance, that is, the q The parameter vector after the permutation is the same as the q +1 Euclidean distance between parameter vectors after permutation, represents the mean Euclidean distance, represents the Euclidean distance between the parameter vector and the solution center after the first permutation, It represents the Euclidean distance between the parameter vector after the NP-th permutation and the solution center. The solution center = (UB + LB) / 2. UB represents the upper limit vector, that is, the parameter of each dimension is the upper limit mean of the model parameters of all parameter vectors in the same dimension; LB represents the lower limit vector, that is, the parameter of each dimension is the lower limit mean of the model parameters of all parameter vectors in the same dimension. A smaller distribution uniformity indicates that the parameter vectors are more evenly distributed in the solution space, with a wider coverage and higher overall quality. In this case, a faster search can be performed around the optimal parameter vector. A larger distribution uniformity indicates that the parameter vectors are more unevenly distributed in the solution space, and a larger search range is required.

[0065] When the distribution uniformity is less than a preset uniformity threshold, the parameter vector is searched according to the optimal parameter vector and the adaptive balance search factor, and the parameter vector obtained after the adaptive search is: in, Indicates the m The parameter vector after adaptive search, represents the first random parameter vector selected from the permuted parameter vector, represents the first random number between (-b, b), b represents the oscillation factor, and , represents the inverse hyperbolic tangent function; alternatively, b can be set to a constant term, such as 0.4, 0.5, 0.6, etc.; When the distribution uniformity is greater than or equal to the preset uniformity threshold, the parameter vector is searched according to the upper limit and the lower limit of the model parameter, and the parameter vector obtained after the adaptive search is: in, represents the second random number between (0,1), represents the inertia weight, and , Represents pi.

[0066] The embodiment of the present application can adaptively control the population distribution by performing an adaptive search on the parameter vector based on distribution uniformity, and perform search control through an adaptive balanced search factor and an inertia weight. On the one hand, it improves the collaborative search capability and search accuracy under uniform distribution conditions, and on the other hand, it improves the global search capability under uneven distribution conditions, and also avoids excessive global search in the later stage, thereby comprehensively improving the training effect of the algorithm.

[0067] In a possible implementation, performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain the parameter vector after the volatility spiral chain collaborative search includes: According to the current number of training times, the waveform adjustment factor is determined as: in, represents the waveform regulation factor, represents the nonlinear control parameter, represents pi; When the volatility adjustment factor is greater than or equal to the preset adjustment threshold, a spiral chain collaborative search is performed on the parameter vector after the adaptive search according to the optimal parameter vector, and the parameter vector after the volatility spiral chain collaborative search is obtained as: in, Indicates the t During the training n The parameter vector after adaptive search, Indicates the n The parameter vector after the collaborative search of the volatility spiral chain, Indicates the t During the training n+1 parameter vector after adaptive search; when n is equal to NP, represents the parameter vector after random adaptive search; Represents the third random number between (0,1); represents the spiral collaborative search coefficient, ; Represents the fourth random number between (0,1); When the volatility adjustment factor is less than the preset adjustment threshold, a random parameter vector is determined in the parameter vector after the adaptive search, and a spiral chain collaborative search is performed based on the parameter vector after the adaptive search of the random parameter vector to obtain the parameter vector after the volatility spiral chain collaborative search.

[0068] in, represents the randomly generated second random parameter vector.

[0069] The present invention performs a volatility spiral chain collaborative search on the parameter vector after the adaptive search, and selects different spiral chain collaborative searches through a volatility adjustment factor. The waveform adjustment factor can increase with the increase in the number of training times, and there will be ups and down fluctuations in the process of increase, so that different spiral chain collaborative searches are used for training, which can effectively explore the space between different solutions, improve the ability to jump out of the local optimum, and increase the ability to find a better solution, thereby improving the global search capability. It can be seen that in the early stage of the algorithm, random parameter vectors are mainly used for spiral chain collaborative search, which provides a stronger global search capability while searching the unknown space. In the middle stage of the calculation, random parameter vectors or optimal parameter vectors are used for spiral chain collaborative search, which can improve the local search capability to a certain extent while maintaining the diversity of solutions. In the later stage of the algorithm, only the optimal parameter vector is used for spiral chain collaborative search to ensure the quality of the solution.

[0070] In a possible implementation, performing a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search to obtain the parameter vector after the bidirectional mutation search includes: According to the current number of training times, the number of mutation dimensions, forward mutation factor, and reverse mutation factor are obtained as follows: in, Indicates the number of mutation dimensions, represents the ceiling function, Represents the total dimension of model parameters, represents the fifth random number between (0,1), Indicates the t During the training k The parameter vector after the volatility spiral chain collaborative search j Dimensional model parameters The corresponding positive mutation factor, Indicates the t During the training k The parameter vector after the volatility spiral chain collaborative search j Dimensional model parameters The corresponding reverse mutation factor, Indicates the total mutation order, which can be set to 2, 3 or 4; Determining, based on the number of mutation dimensions, the dimensions to be mutated of the parameter vector after the volatility spiral chain collaborative search; For example, based on the number of mutation dimensions , can be randomly selected from the parameter vector after the volatility spiral chain collaborative search dimensional model parameters, and obtain the dimension of the parameter vector to be mutated after the volatility spiral chain collaborative search.

[0071] For any dimension to be mutated, a forward mutation factor is used to perform a forward mutation on the parameter vector after the volatility spiral chain collaborative search, or a reverse mutation factor is used to perform a reverse mutation on the parameter vector after the volatility spiral chain collaborative search. The parameter vector after the bidirectional mutation search is: or in, Indicates the t During the training k The dimension of the parameter vector to be mutated after the collaborative search of the volatility spiral chain, Indicates the k The dimensions of the parameter vector to be mutated after the bidirectional mutation search. For the dimensions that do not need to be mutated, the model parameters remain unchanged. represents the mutation control coefficient and is set to 0.4; represents the mutation order, The optimal parameter vector j dimensional model parameters, represents the positive mutation factor, Represents a reverse mutation factor.

[0072] Optionally, a simulated annealing algorithm can be used to control the parameter vector after the bidirectional mutation search. This can increase the diversity of the parameter vector and make the parameter vector evolve in a more favorable direction, helping to accelerate the convergence of the algorithm and improve the stability of the solution. It is also possible to perform out-of-bounds processing on the parameter vector after each change, and use the out-of-bounds processing parameter vector to maintain the optimal parameter vector. For example, the out-of-bounds model parameters can be randomly generated within the corresponding upper and lower limits or set to the closest limit value.

[0073] The embodiment of the present application performs a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search, and has a powerful global search capability to help the algorithm escape from the local optimal solution.

[0074] The present application provides an action management and identification method based on earthquake site disaster investigation, the method comprising: determining the earthquake site area to be acted upon, and planning the disaster search tasks corresponding to the earthquake site area to be acted upon by multiple drones; then dispatching the disaster search tasks to multiple drones, and controlling the drones to execute the disaster search tasks, and obtaining actual disaster investigation results corresponding to the disaster search tasks; then, based on the actual disaster investigation results, generating disaster points on an electronic map corresponding to the earthquake site area to be acted upon, and obtaining a disaster investigation map; finally, feeding back the disaster investigation map to the equipment designated by the commander to guide the commander's disaster relief actions, which not only improves the efficiency of earthquake site disaster investigations, but also forms comprehensive disaster data, so that the commander can fully understand the disaster situation, assist the commander in quickly forming a global and optimal action plan, and improve the efficiency of disaster handling.

[0075] like Figure 2 As shown, based on the same inventive concept, another embodiment of the present application provides an action management and identification device based on earthquake site disaster investigation, comprising: The mission planning module 201 is used to determine the earthquake scene area to be operated and plan the disaster search mission corresponding to the earthquake scene area to be operated by multiple drones.

[0076] The disaster investigation module 202 is used to send the disaster search task to multiple drones, control the drones to execute the disaster search task, and obtain actual disaster investigation results corresponding to the disaster search task.

[0077] The disaster summary module 203 is used to generate disaster points on the electronic map corresponding to the earthquake site area to be acted upon according to the actual disaster investigation results corresponding to the disaster search task, and obtain a disaster investigation map.

[0078] The disaster situation feedback module 204 is used to feed back the disaster situation survey map to the device designated by the commander to guide the commander's disaster relief actions and complete action management based on the earthquake site disaster situation survey.

[0079] An action management and identification device based on earthquake site disaster investigation provided in an embodiment of the present application can execute the above-mentioned method and technical solution. Its principles and beneficial effects are similar and will not be repeated here.

[0080] like Figure 3 As shown, based on the same inventive concept, another embodiment of the present application provides an electronic device, which includes a memory 301 and a processor 302. The memory 301 and the processor 302 communicate with each other via a communication bus 303.

[0081] The memory 301 is used to store code instructions.

[0082] The processor 302 is configured to execute the code instructions so that the electronic device executes the action management and identification method based on earthquake site disaster investigation provided by the embodiment of the present application.

[0083] The communication bus 303 mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 303 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in the figure, but this does not mean that there is only one bus or one type of bus. The communication interface is used for communication between the above-mentioned terminal and other devices. The memory 301 can include random access memory (RAM) or non-volatile memory, such as at least one disk storage. Optionally, the memory 301 can also be at least one storage device located away from the aforementioned processor 302.

[0084] The processor 302 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0085] In addition, to achieve the above-mentioned purpose, an embodiment of the present application also proposes a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the action management and identification method based on earthquake site disaster investigation provided by the embodiment of the present application is implemented.

[0086] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0087] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, apparatuses, electronic devices, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0088] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0090] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0091] Finally, it should be noted that, in the embodiments of the present application, relational terms such as first and second, etc., are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise", or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or terminal device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or terminal device. In the absence of further restrictions, an element defined by the statement "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or terminal device comprising the element.

[0092] The principles and implementation methods of the present application are described in detail using specific examples in the embodiments of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those skilled in the art, according to the idea of ​​the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present application.

Claims

1. An action management identification method based on earthquake site disaster investigation, characterized in that: include: Determine the earthquake scene area to be acted upon, and plan disaster search missions for multiple UAVs corresponding to the earthquake scene area to be acted upon; Sending the disaster search mission to multiple drones, controlling the drones to execute the disaster search mission, and obtaining actual disaster investigation results corresponding to the disaster search mission; Generating disaster points on an electronic map corresponding to the earthquake site area to be acted upon based on actual disaster investigation results corresponding to the disaster search task, thereby obtaining a disaster investigation map; The disaster investigation map is fed back to a device designated by a commander to guide the commander's disaster relief actions and complete action management based on the earthquake site disaster investigation.

2. The action management identification method based on earthquake site disaster investigation according to claim 1 is characterized in that: The step of determining the earthquake scene area to be acted upon and planning the disaster search mission corresponding to the earthquake scene area to be acted upon by multiple UAVs includes: Obtain the closed area determined by the disaster relief personnel on the electronic map through human-computer interaction to obtain the earthquake scene area to be acted upon; Acquiring power transmission line information and road information in the area of ​​the earthquake site to be acted upon; generating a first disaster search task for conducting a disaster investigation along the transmission lines based on the transmission line information of the earthquake scene area to be acted upon; generating a second disaster search task for disaster investigation along the road based on the road information; With the goal of traversing the earthquake scene area to be acted upon, a third disaster search task is generated for conducting a disaster investigation on trapped persons and collapsed houses in the earthquake scene area to be acted upon.

3. The action management identification method based on earthquake site disaster investigation according to claim 2 is characterized in that: The step of sending the disaster search mission to multiple drones, controlling the drones to execute the disaster search mission, and obtaining actual disaster investigation results corresponding to the disaster search mission includes: Sending the first disaster search mission to a first drone equipped with a power transmission line detection model, and controlling the first drone to perform target detection along the power transmission line to obtain a power transmission line survey result; Sending the second disaster search mission to a second drone equipped with a road detection model, and controlling the second drone to perform target detection along the road to obtain road survey results; The third disaster search mission is sent to a third drone equipped with a personnel and housing detection model, and the third drone is controlled to traverse the earthquake site area to be operated to obtain personnel and housing investigation results; The personnel and house survey results, road survey results, and power line survey results are taken together as actual disaster survey results corresponding to the disaster search task.

4. The action management identification method based on earthquake site disaster investigation according to claim 3 is characterized in that: The transmission line detection model, road detection model, and personnel and house detection model are all set as YOLO models.

5. The action management identification method based on earthquake site disaster investigation according to claim 4 is characterized in that: The transmission line detection model, the road detection model, and the personnel and house detection model are installed in the same manner and all include: A hybrid mapping strategy is used to initialize the model to be carried out to obtain multiple parameter vectors; wherein the model to be carried out is a power transmission line detection model, a road detection model or a person and house detection model; Obtaining the fitness corresponding to each of the parameter vectors, and obtaining the optimal parameter vector according to the fitness of the parameter vectors; Based on the optimal parameter vector, performing an adaptive search based on distribution uniformity on the parameter vector to obtain a parameter vector after adaptive search; Performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain a parameter vector after the volatility spiral chain collaborative search; Performing a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search to obtain a parameter vector after the bidirectional mutation search; Determine whether the current number of training times has reached the maximum number of training times. If so, re-obtain the optimal parameter vector based on the parameter vector after the bidirectional mutation search to obtain the target parameter vector. Otherwise, return to the step of obtaining the optimal parameter vector and enter the next training process. The target parameter vector is used as the final parameter of the model to be mounted, and the model to be mounted is deployed on the UAV according to the final parameter of the model to be mounted.

6. The action management and identification method based on earthquake site disaster investigation according to claim 5 is characterized in that: The hybrid mapping strategy is used to initialize the model to be loaded, and multiple parameter vectors are obtained, including: Randomly initializing the model parameters of the model to be loaded between an upper limit and a lower limit, and encoding the initialized model parameters into a vector to obtain a basic vector; Based on the basic vector, a plurality of other vectors are obtained by adopting a chaotic mapping sequence; A parameter vector is generated based on the other vectors, the upper limit of the model parameters, and the lower limit of the model parameters, and all other vectors are processed to obtain multiple parameter vectors.

7. The action management and identification method based on earthquake site disaster investigation according to claim 5 is characterized in that: The obtaining of the fitness corresponding to each parameter vector and obtaining the optimal parameter vector according to the fitness of the parameter vector includes: For any of the parameter vectors, obtain the loss function value corresponding to the parameter vector, and add the loss function value to a preset non-zero constant term and then take the inverse to obtain the fitness corresponding to the parameter vector; The parameter vector with the largest fitness is taken as the optimal parameter vector.

8. The action management and identification method based on earthquake site disaster investigation according to claim 5 is characterized in that: The step of performing an adaptive search on the parameter vector based on distribution uniformity based on the optimal parameter vector to obtain a parameter vector after the adaptive search includes: Arrange the parameter vectors in ascending order of fitness, take the arranged parameter vectors as a basis, and obtain an adaptive balance search factor according to the fitness of the optimal parameter vector, the fitness of the parameter vector, and the current number of training times; Obtaining the Euclidean distance between adjacent parameter vectors and determining the distribution uniformity based on the Euclidean distance between adjacent parameter vectors; When the distribution uniformity is less than a preset uniformity threshold, searching the parameter vector according to the optimal parameter vector and the adaptive balance search factor to obtain a parameter vector after adaptive search; When the distribution uniformity is greater than or equal to a preset uniformity threshold, the parameter vector is searched according to the model parameter upper limit and the model parameter lower limit to obtain a parameter vector after adaptive search.

9. The action management and identification method based on earthquake site disaster investigation according to claim 8, characterized in that: The step of performing a volatility spiral chain collaborative search on the parameter vector after the adaptive search to obtain the parameter vector after the volatility spiral chain collaborative search comprises: Determine the waveform adjustment factor based on the current number of training times; When the volatility adjustment factor is greater than or equal to a preset adjustment threshold, performing a spiral chain collaborative search on the parameter vector after the adaptive search according to the optimal parameter vector to obtain a parameter vector after the volatility spiral chain collaborative search; When the volatility adjustment factor is less than the preset adjustment threshold, a random parameter vector is determined in the parameter vector after the adaptive search, and a spiral chain collaborative search is performed based on the parameter vector after the adaptive search of the random parameter vector to obtain the parameter vector after the volatility spiral chain collaborative search.

10. The action management and identification method based on earthquake site disaster investigation according to claim 9, characterized in that: The performing a bidirectional mutation search on the parameter vector after the volatility spiral chain collaborative search to obtain the parameter vector after the bidirectional mutation search includes: Obtain the number of mutation dimensions, forward mutation factors, and reverse mutation factors based on the current number of training times; Determining, based on the number of mutation dimensions, the dimensions to be mutated of the parameter vector after the volatility spiral chain collaborative search; For any dimension to be mutated, a forward mutation factor is used to perform a forward mutation on the parameter vector after the volatility spiral chain collaborative search, or a reverse mutation factor is used to perform a reverse mutation on the parameter vector after the volatility spiral chain collaborative search, to obtain the parameter vector after the bidirectional mutation search.

Images