Quantum behavior driven multi-target forest fire source positioning method and device

Abstract

The application discloses a kind of multi-objective forest fire source positioning method and device based on quantum behavior driving, it is related to disaster event analysis technical field, including: establishing forest fire multi-objective optimization model, using heat-resistant reconnaissance robot colony is iterated to form discrete optimal path based on LBPQI algorithm, and according to path total length, LBPQI algorithm is used again to obtain optimal speed strategy;Make heavy fire vehicle travel along discrete optimal path, according to optimal speed strategy to execute fire extinguishing task.Solve the multi-objective collaborative optimization problem of fire source positioning precision, search efficiency and heavy fire vehicle delivery endurance under the complex rescue scene of forest fire, realize the complete process and whole-process collaborative optimization from fire source search to vehicle execution, improve the reliability and success rate of overall rescue task, LBPQI algorithm effectively guides heat-resistant robot colony to avoid being trapped in local high temperature point, ensure the rapid, accurate positioning of real main fire source.

Description

Technical Field

[0001] This invention relates to the field of disaster event analysis technology, and in particular to a method and apparatus for locating multi-target forest fire sources based on quantum behavior. Background Technology

[0002] Rapid and accurate location of the fire source in the early stages of a forest fire is crucial for firefighting efforts. Due to the complex and dangerous conditions of high temperatures and dense smoke at fire sites, employing swarms of heat-resistant reconnaissance robots to replace firefighters in initial fire source reconnaissance is an important means of ensuring the personal safety of firefighters. Particle Swarm Optimization (PSO) algorithms are commonly used to solve multi-agent fire source localization problems. However, because the thermal radiation field at a fire site is uneven, there may be multiple local high-temperature points. Using PSO algorithms can easily lead to local optima, meaning that the robot swarm may prematurely cluster at a particular local high-temperature point, misidentifying it as the fire core and delaying the location of the true main fire source (the globally hottest point). Furthermore, in actual forest fire rescue missions, it is not only necessary to find the most likely fire source but also to consider paths that allow heavy firefighting and rescue vehicles to efficiently and safely enter the fire core area. Existing algorithms therefore face another serious challenge: the actions of robots and vehicles must strictly adhere to physical and safety constraints (such as movement range and speed limits), and the range constraints of heavy firefighting and rescue vehicles cannot be ignored while focusing solely on search accuracy. Especially in forest fire scenarios with complex terrain, if heavy fire-fighting and rescue vehicles consume excessive fuel or even malfunction and become unable to move before reaching the fire source due to violent acceleration and deceleration, the entire rescue mission will fail. Therefore, it is necessary to coordinate and optimize the energy consumption management of heavy fire-fighting and rescue vehicles during their journey and to find a balance between quickly locating the fire source. Summary of the Invention

[0003] This invention provides a method and apparatus for locating multi-target forest fire sources based on quantum behavior, in order to solve the technical problem of coordinating and optimizing energy consumption management and rapid fire source location during the movement of heavy fire fighting and rescue vehicles in forest fire scenarios, thereby quickly and accurately locating the fire source and planning feasible routes for heavy fire fighting and rescue vehicles.

[0004] In a first aspect, embodiments of the present invention provide a multi-target forest fire source localization method based on quantum behavior-driven methods, including:

[0005] S101. Establish a multi-objective optimization model for forest fires and conduct collaborative optimization of three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles.

[0006] S102 utilizes a swarm of heat-resistant reconnaissance robots to iterate based on a quantum imitation algorithm with learning behavior promotion, in order to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source. The Pareto front algorithm is used to determine the globally optimal individual in each iteration, and the position sequence of the globally optimal individual in each iteration is recorded to form a discrete optimal path.

[0007] S103, based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed of the heavy fire extinguishing vehicle to obtain the optimal speed strategy.

[0008] S104 enables the heavy firefighting vehicle to travel along a discrete optimal path and at an optimal speed, eventually reaching the predicted location of the forest fire source to perform firefighting tasks. The predicted location of the forest fire source is the location of the globally optimal individual determined by the heat-resistant reconnaissance robot swarm in the last iteration.

[0009] Secondly, embodiments of the present invention provide a multi-target forest fire source location device based on quantum behavior-driven methods, comprising:

[0010] The multi-objective optimization model construction module is used to establish a multi-objective optimization model for forest fires, and to perform collaborative optimization on three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles.

[0011] The optimal path iteration module is used to utilize a swarm of heat-resistant reconnaissance robots to iterate based on a quantum imitation algorithm with learning behavior promotion, in order to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source to form a discrete optimal path;

[0012] The optimal speed iteration module is used to perform iterative calculations again using a quantum imitation algorithm with learning-enhanced behavior based on the total path length of the discrete optimal path, in order to obtain the optimal speed strategy.

[0013] The task execution module is used to enable heavy firefighting vehicles to travel along discrete optimal paths and at optimal speeds, ultimately reaching the predicted location of the forest fire source to carry out firefighting tasks.

[0014] Thirdly, embodiments of the present invention provide an electronic device, comprising:

[0015] One or more processors;

[0016] Storage device for storing one or more programs.

[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the above-described quantum behavior-driven multi-target forest fire source localization method.

[0018] Fourthly, embodiments of the present invention provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the above-described quantum behavior-driven multi-target forest fire source localization method.

[0019] This invention provides a quantum behavior-driven multi-objective forest fire source localization method and apparatus. The method establishes a constrained multi-objective collaborative optimization model that simultaneously optimizes the fire field temperature detection intensity, robot search distance, and vehicle energy consumption. L1 penalty and smoothing functions are used to process the three objective functions. A two-stage collaborative optimization is employed. A swarm of heat-resistant reconnaissance robots iteratively optimizes the fire field Gaussian temperature detection intensity and the distance between the heat-resistant reconnaissance robots and the predicted fire source using a quantum imitation algorithm with learning-enhanced behavior. The Pareto front algorithm is used to determine the globally optimal individual in each iteration, and the position sequence of the globally optimal individual in each iteration is recorded to form a discrete optimal path. Based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning-enhanced behavior iteratively calculates the speed of the heavy fire-fighting vehicle to obtain the optimal speed strategy. Finally, the heavy fire-fighting vehicle travels along the discrete optimal path at the optimal speed strategy to reach the predicted forest fire source location and perform the fire-fighting task. This research addresses the multi-objective collaborative optimization problem of fire source location accuracy, search efficiency, and the delivery endurance of heavy firefighting vehicles in complex forest fire rescue scenarios. It achieves complete process and end-to-end collaborative optimization from intelligent perception of fire source search and vehicle path planning to energy-saving execution. Through two-stage optimization, the tasks of heat-resistant robots and heavy firefighting vehicles are seamlessly connected, improving the reliability and success rate of the overall rescue mission. The LBPQI algorithm effectively guides the heat-resistant robot swarm to avoid getting stuck in local high-temperature points, ensuring rapid and accurate location of the real main fire source, and providing a high-quality path foundation for the subsequent movement of heavy firefighting vehicles, thereby improving the accuracy and robustness of forest fire source location. Attached Figure Description

[0020] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0021] Figure 1 This is a flowchart of a multi-target forest fire source localization method based on quantum behavior driven according to Embodiment 1 of the present invention;

[0022] Figure 2This is a flowchart of a multi-target forest fire source localization method based on quantum behavior driven according to Embodiment 2 of the present invention;

[0023] Figure 3 This is a schematic diagram of the two-dimensional iterative process for locating forest fire sources as described in Embodiment 2 of the present invention;

[0024] Figure 4 This is a schematic diagram of the iterative process for finding the optimal driving linear velocity of a heavy fire-fighting vehicle as described in Embodiment 2 of the present invention;

[0025] Figure 5 This is a schematic diagram of the path for finding the main fire source as described in Embodiment 2 of the present invention;

[0026] Figure 6 This is a schematic diagram of the structure of a multi-target forest fire source locator based on quantum behavior driven according to Embodiment 3 of the present invention;

[0027] Figure 7 This is a structural diagram of the electronic device described in Embodiment 4 of the present invention. Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0029] Example 1

[0030] Figure 1 The flowchart of a multi-target forest fire source localization method based on quantum behavior driven according to Embodiment 1 of the present invention specifically includes the following steps:

[0031] S101. Establish a multi-objective optimization model for forest fires, and conduct collaborative optimization of three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles.

[0032] First, the forest fire rescue mission is modeled, with three optimization objectives: the intensity of the Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles. These objectives are modeled as objective functions for multi-objective collaborative optimization. The goal is to maximize the intensity of the Gaussian temperature field detection at the fire site to locate the main fire source or the source of the forest fire with the highest temperature. The goal is to minimize the distance between the heat-resistant reconnaissance robot and the online predicted fire source to guide the search and rescue process to efficiently and directionally approach the fire source and plan the optimal route. The goal is to minimize the energy consumption of heavy firefighting vehicles to ensure that they have sufficient endurance and rescue capabilities to complete the search and firefighting mission. For different optimization objectives, corresponding constraints can be set according to actual needs.

[0033] S102 utilizes a swarm of heat-resistant reconnaissance robots to iterate based on a quantum imitation algorithm with learning-enhanced behavior, in order to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source. The Pareto front algorithm is used to determine the globally optimal individual in each iteration, and the position sequence of the globally optimal individual in each iteration is recorded to form a discrete optimal path.

[0034] To achieve multi-objective collaborative optimization, a swarm of heat-resistant robots was used to conduct a collaborative search in a forest fire. Based on the position and temperature / heat source detection of each individual heat-resistant robot, an iterative algorithm based on learning-enhanced quantum imitation (LBPQI) was employed. In each iteration, the movement of each heat-resistant robot in the LBPQI algorithm differs from traditional particle swarm optimization (PSO) algorithms, which rely solely on individual experience and global optima. Instead, it first learns through role model learning and peer learning, enabling the swarm to converge quickly and collaboratively towards high-temperature regions while maintaining diverse information exchange among individuals, avoiding premature convergence due to localized high-temperature points. Combined with quantum imitation, a neutralizing attractor was constructed for each heat-resistant robot to dynamically integrate its historical best position, the current global best position of the swarm, and a reference position combining global and local neighbor information. Then, centered on this neutralizing attractor, probabilistic quantum jump updates were performed within a range that automatically shrank with each iteration, simulating the quantum tunneling effect. This allows the heat-resistant robots to explore a larger area in the early stages of the search to escape local optima, and to perform a more refined search near the optimal solution in later stages. In each iteration, the Pareto front algorithm is applied to select a Pareto non-dominated solution set from the current heat-resistant robot population, considering the potentially conflicting objectives of highest temperature and closest distance. The heat-resistant robot with the highest overall evaluation index is then identified as the globally optimal individual for this iteration, representing the best compromise estimate of the main fire source location for that iteration. The individual positions of the globally optimal individuals from each iteration are concatenated into a sequence, forming a discrete optimal path from the starting point to the final located fire source location.

[0035] S103. Based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed of the heavy fire extinguishing vehicle to obtain the optimal speed strategy.

[0036] After determining the discrete optimal path using the globally optimal individual in each iteration, the total length of the path is calculated. From the perspective of single-objective optimization, the driving speed of the heavy fire extinguishing vehicle is iteratively calculated with the goal of minimizing energy consumption. The quantum imitation (LBPQI) algorithm with learning behavior promotion is used again. At this time, the optimization object becomes driving speed and the optimization variable becomes a single dimension. The optimal speed curve of the heavy fire extinguishing vehicle is solved to find the optimal speed strategy that minimizes the energy consumption objective function of the heavy fire extinguishing vehicle.

[0037] S104 enables the heavy firefighting vehicle to travel along a discrete optimal path and at an optimal speed, eventually reaching the predicted location of the forest fire source to perform firefighting tasks. The predicted location of the forest fire source is the location of the globally optimal individual determined by the heat-resistant reconnaissance robot swarm in the last iteration.

[0038] Heavy firefighting vehicles travel along the discrete optimal paths obtained above, controlling their speed according to the optimal speed strategy described above, to reach the predicted forest fire source with the lowest energy consumption and highest efficiency to carry out firefighting tasks. The predicted forest fire source is the location of the globally optimal individual obtained in the last iteration of the iterative solution for the discrete optimal path; that is, the location with the strongest global fire intensity and closest to the fire source found through iteration can be regarded as the forest fire source. Even if there are deviations, it is still within a controllable range that firefighters can quickly determine based on the on-site situation. This completes the multi-objective collaborative optimization task of locating the forest fire source from the reconnaissance of the heat-resistant robot swarm to the deployment of heavy firefighting vehicles for rescue.

[0039] This embodiment establishes a multi-objective collaborative optimization model that simultaneously optimizes the fire field temperature detection intensity, robot search distance, and vehicle energy consumption. A two-stage collaborative optimization approach is employed. A swarm of heat-resistant reconnaissance robots iteratively optimizes the fire field Gaussian temperature detection intensity and the distance between the heat-resistant reconnaissance robots and the predicted fire source using a quantum imitation algorithm with learning-enhanced behavior. The Pareto front algorithm is used to determine the globally optimal individual in each iteration, and the position sequence of the globally optimal individuals in each iteration is recorded to form a discrete optimal path. Based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning-enhanced behavior iteratively calculates the speed of the heavy firefighting vehicle to obtain the optimal speed strategy. Finally, the heavy firefighting vehicle travels along the discrete optimal path at the optimal speed strategy to reach the predicted forest fire source location and perform the firefighting mission. This research addresses the multi-objective collaborative optimization problem of fire source location accuracy, search efficiency, and the delivery endurance of heavy firefighting vehicles in complex forest fire rescue scenarios. It achieves complete process and end-to-end collaborative optimization from intelligent perception of fire source search and vehicle path planning to energy-saving execution. Through two-stage optimization, the tasks of heat-resistant robots and heavy firefighting vehicles are seamlessly connected, improving the reliability and success rate of the overall rescue mission. The LBPQI algorithm effectively guides the heat-resistant robot swarm to avoid getting stuck in local high-temperature points, ensuring rapid and accurate location of the real main fire source, and providing a high-quality path foundation for the subsequent movement of heavy firefighting vehicles, thereby improving the accuracy and robustness of forest fire source location.

[0040] Example 2

[0041] Figure 2 This is a flowchart of a multi-target forest fire source localization method based on quantum behavior driven by Embodiment 2 of the present invention. This embodiment is an optimization based on the above embodiment. In this embodiment, S102 is specifically optimized as follows:

[0042] The Pareto dominance relation is used to identify the non-dominated solution set in the heat-resistant reconnaissance robot group and calculate the comprehensive evaluation index of each individual in the solution set. Based on the comprehensive evaluation index, the current global best individual is determined. At the same time, a position state vector is constructed for each heat-resistant reconnaissance robot to represent the individual position, historical individual best information, and global best individual information of each heat-resistant reconnaissance robot.

[0043] The position-state vector of each heat-resistant reconnaissance robot is mapped using a quantum imitation algorithm with learning behavior promotion, the position of each heat-resistant reconnaissance robot at the next moment is calculated, and the corresponding heat-resistant reconnaissance robot is driven to move to that position.

[0044] The movement process of the heat-resistant reconnaissance robot group is iterated until the preset maximum number of iterations is reached. The position of the globally optimal individual in each iteration is recorded as the globally optimal individual position sequence, and a discrete optimal path is formed.

[0045] Accordingly, the quantum behavior-driven multi-target forest fire source localization method provided in this embodiment specifically includes:

[0046] S201. Establish a multi-objective optimization model for forest fires, and conduct collaborative optimization of three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles.

[0047] Specifically, a multi-objective optimization model for forest fires is established, including a first objective function characterizing the detection intensity of the Gaussian temperature field at the fire site, a second objective function characterizing the distance between the heat-resistant reconnaissance robot and the predicted fire source, and a third objective function characterizing the energy consumption of heavy firefighting vehicles.

[0048] First, we establish the objectives to be optimized to form a multi-objective optimization model for forest fires, expressed by the following formula:

[0049] ;

[0050] Wherein, the first objective function Second objective function The third objective function Establish the first objective function. The Gaussian temperature field detection intensity is used to characterize forest fires. Multiple Gaussian functions are superimposed to simulate the complex temperature distribution caused by multiple heat sources (including primary and secondary fire sources) in the fire area. The formula is as follows:

[0051] ;

[0052] in, Indicates the spatial location of the field x Temperature at that location; This represents the center of the k-th basis function, i.e., the core coordinates of the k-th combustion point or heat source; The k-th mixing coefficient directly represents the maximum thermal radiation intensity at the k-th combustion point; the larger the value, the higher the temperature. The width of the k-th basis function determines the range of influence of the thermal radiation at the k-th combustion point; a larger value indicates a wider heat diffusion at that point. Since there is only one primary ignition source in this temperature field, and the rest are secondary ignition sources, the goal is to find the location with the highest temperature using the first objective function, i.e., to maximize the first objective function. A second objective function is then established. The Euclidean distance used to characterize the heat-resistant robot and the source of the forest fire to be predicted is given by the following formula:

[0053] ;

[0054] in, Indicates spatial location x Online fire source prediction x ' R 2 The Euclidean distance between them, where R 2 Representing a two-dimensional real number space, online prediction of fire sources. x This is the best estimate from the previous generation and can also be seen as the most likely primary fire source at present. The target is designed to provide a spatial anchor point during the search process. x By minimizing the distance from the best estimate of the previous generation, the heat-resistant robot is prevented from over-searching in the exploration of the new generation, thus stabilizing the entire search process. x In each subsequent iteration of fire source localization, the individual position is dynamically updated to the globally optimal position from the previous iteration to ensure search consistency. A third objective function is established. The energy consumption of heavy-duty firefighting vehicles, based on vehicle dynamics, is represented by the following formula:

[0055] ;

[0056] in, Indicates that the heavy firefighting vehicle is moving at a linear velocity drive a distance The total energy consumption is calculated by comprehensively considering various factors such as vehicle weight, speed, gradient, wind resistance, and engine efficiency to ensure accurate energy consumption calculations. The formula above is divided into weighting modules. ,in , , , Represents the distance the vehicle has traveled. Represents the total weight of the vehicle. Represents vehicle acceleration. Represents the gravitational constant. Represents the road angle. Represents the rolling resistance coefficient of vehicle tires. This represents the vehicle's fuel-air mass ratio. The heat of typical diesel fuel. Represents the vehicle fuel conversion factor. This represents the vehicle's transmission efficiency. Represents the efficiency parameters of a vehicle's diesel engine; speed module ,in, , Represents the aerodynamic drag coefficient. Represents air density, Represents the frontal surface area of ​​the vehicle. Represents vehicle speed; and engine module ,in, This represents the friction coefficient of the vehicle's engine. Represents the vehicle's engine speed. This represents the engine displacement of a vehicle.

[0057] Initialize the parameters of each objective function separately, and randomly initialize the initial search population of the heat-resistant reconnaissance robot swarm.

[0058] A swarm of heat-resistant reconnaissance robots serves as the actuator in the main fire source search phase of a multi-target forest fire source localization problem. The swarm is initially randomly distributed in a two-dimensional space within the forest fire area. The robot swarm at the next iteration is defined as ,in ,for , Indicates the first The robot in the first Spatial position at the next iteration and Representing the number of robots and the total number of iterations, respectively, the th The dynamic model formula of a robot can be expressed as follows: , indicating the first The robot from the first The iteration to the... The position update rule for the next iteration, where Indicates the first The robot from the first The iteration to the... The driving strategy for this iteration. Since the first and second objective functions share the same decision variable, namely the position x of the heat-resistant robot, they are processed collaboratively. The first objective function is maximized to seek the maximum value of the Gaussian temperature field, while the second objective function is minimized to approximate the distance to the online predicted fire source. To unify the optimization direction, the second objective function is negatively evaluated, transforming it into a maximization objective. The formula for their collaborative optimization is as follows:

[0059] ;

[0060] In the first stage of execution, the LBPQI algorithm and the Pareto front algorithm can be used to solve for the path. Then, in the second stage, based on the path obtained in the first stage, the LBPQI algorithm can be used again to solve for the optimal speed strategy for the heavy fire-fighting vehicle. To maintain consistency with the overall optimization direction of the algorithm, the minimization of energy consumption in the third objective function can be transformed into a maximization problem. Similarly, the third objective function is represented by a negative value, expressed as follows: .

[0061] Optionally, the first objective function and the second objective function are constrained by the fire range boundary using the L1 penalty function, the third objective function is constrained by the upper and lower limits of the driving speed using the driving speed limit constraint, and the smoothing is performed using the smoothing function, while ensuring the minimum boundary approximation error.

[0062] To ensure the physical feasibility of the solution, constraints need to be imposed on the forest fire area, the search process of the heat-resistant robot, and the driving process of the heavy fire-fighting vehicle when constructing the multi-objective optimization model. These constraints must be set to ensure that the vehicle does not exceed the forest fire boundary or the physical performance limits of the heavy fire-fighting vehicle, strictly meeting the actual physical constraints in forest fire rescue and minimizing boundary violations to ensure the accuracy of the results and the safety of the verification process. The optimization of the first and second objective functions requires that the detection position and the position of the heat-resistant robot in the Gaussian temperature field of the fire area do not exceed the safety boundary defined by the forest fire area. The optimization of the third objective function requires that the driving speed of the heavy fire-fighting vehicle be within the safety and mechanical limits. For example, to automatically satisfy these constraints in the LBPQI algorithm, the constraint violation is added as a penalty term to the original objective function through an L1 penalty function. This allows the algorithm to spontaneously reduce constraint violations during the search process to pursue the original objective. For each constraint... The degree of its violation can be measured as (Value is 0 when the constraint is satisfied, and positive when it is violated). This is achieved through a sufficiently large penalty parameter. We then perform a weighted summation to construct a new first objective function, expressed as follows:

[0063] ;

[0064] in, The number of constraints representing the first objective function. It is a penalty parameter greater than zero, applied when a candidate solution violates the first... One constraint (i.e.) When ), the corresponding The item will produce a positive penalty value, resulting in The total function value is reduced, thus penalizing the infeasible solution during optimization and guiding the algorithm automatically away from the constraint violation region. When When the value is sufficiently large, the solution to the new function is equivalent to the exact solution to the original constrained problem, i.e., when... When large enough, maximize The solution must satisfy all constraints. .However, If a function is not differentiable at its zeros (non-smooth), direct use of it can lead to convergence difficulties in optimization algorithms. Therefore, it is necessary to simultaneously employ a smooth function to address the convergence difficulties caused by the non-differentiability of the function at its boundaries (function boundaries are often sharp angles). This can be achieved through a smooth function. To approximate it, the formula is as follows:

[0065] ;

[0066] in, It is an adjustable smoothing parameter; the smaller its value, the smoother the function... right The higher the approximation accuracy, the better. The goal of this smooth function is to approximate... It has a smaller approximation error, where the independent variable is used. The specific value of any constraint function is determined by the actual constraints being modeled. For example, if the constraint is that the robot's position coordinates must not exceed the boundary, then z corresponds to the difference between the robot's position and the boundary. Its sign (z>0 indicates an out-of-bounds violation) and magnitude directly quantify the degree of constraint violation. Smaller errors mean more accurate assessment of constraint violations, resulting in a more rigorous satisfaction of the safety boundary in the final solution. This enables the LBPQI algorithm to search smoothly and efficiently within the feasible solution space. It is important to note that the set smoothness function satisfies non-negativity. Boundedness of derivative Uniform convergence Error boundedness Furthermore, its theoretical approximation error upper bound must be strictly smaller than the error upper bound of similar smooth functions, ensuring that the function... The surrounding area can fit more closely. This systematically reduces the overall approximation error. These properties collectively ensure that during the optimization process, It can simulate the original with smaller systematic errors and controllable smoothness. The behavior of the operator, when applied to the penalty term, allows for a more accurate measurement of constraint violations, making the constructed objective function continuously differentiable at the constraint boundaries. Substituting this smooth function into the formula for the first objective function expresses the following:

[0067] ;

[0068] Similarly, the optimization process of the second objective function is mainly limited by the spatial movement range of the heat-resistant robot. Its constraints are related to the spatial dimension and boundary definition. Substituting the smooth function into the formula of the second objective function, it can be expressed as:

[0069] ;

[0070] in, This indicates the total number of constraints involved in the first and second objective functions; therefore, the subscript starts from... arrive .

[0071] Similarly, the optimization process of the third objective function must satisfy vehicle velocity dynamics constraints (i.e., upper and lower speed limits). Substituting the smooth function into the formula of the third objective function, it can be expressed as:

[0072] ;

[0073] in, This indicates the total number of constraints involved in the first, second, and third objective functions; therefore, the subscript starts from... arrive This ensures that the entire process, from path search for the heat-resistant robot to speed planning for the heavy firefighting vehicle, is conducted within the feasible solution space, thus guaranteeing the physical feasibility of the solution at the system level.

[0074] S202, using a quantum imitation algorithm with learning behavior promotion, the position state vector of each heat-resistant reconnaissance robot is mapped, the position of each heat-resistant reconnaissance robot at the next moment is calculated, and the corresponding heat-resistant reconnaissance robot is driven to move to that position.

[0075] A quantum imitation algorithm with learning-enhanced behavior (LBPQI) is used to collaboratively optimize multi-objective models. In locating forest fire sources, iterative optimization is performed on multiple potentially conflicting objectives to find the optimal strategy. The LBPQI algorithm learns from role models and peers, moving towards a better solution. Then, quantum imitation behavior is used to map the position-state vector of each heat-resistant reconnaissance robot, obtaining the position of each robot in the next time step, and driving the corresponding heat-resistant reconnaissance robot to move to that position. For example, in... In 3D space, a [something] in the 1st [dimension] The evolutionary population of the next iteration is defined as ,in This indicates the number of individuals in this population. Representing the Individual, =1,2,..., That is, a person with A population of individuals, where the index of each individual in the population is... Each individual corresponds to a heat-resistant robot.

[0076] S203. Using Pareto dominance relations, identify the non-dominated solution set in the heat-resistant reconnaissance robot group and calculate the comprehensive evaluation index of each individual in the solution set. Based on the comprehensive evaluation index, determine the current global optimal individual. At the same time, construct a position state vector for each heat-resistant reconnaissance robot to represent the individual position, historical individual optimal information, and global optimal individual information of each heat-resistant reconnaissance robot.

[0077] In each iteration, when the heat-resistant reconnaissance robot moves to its position at the next moment, the Pareto front algorithm is used to identify the non-dominated solution set in the heat-resistant reconnaissance robot swarm using Pareto dominance relations. Then, the comprehensive evaluation index of each individual heat-resistant reconnaissance robot is calculated, and the globally optimal individual for the current iteration is determined accordingly. Simultaneously, a position-state vector representing each individual heat-resistant reconnaissance robot is constructed for subsequent calculations. For example, the heat-resistant reconnaissance robot swarm... Considered as an evolutionary population in the LBPQI algorithm, the corresponding parameter Z = n (number of robots), and the optimization variable is two-dimensional position coordinates, i.e., M = 2. Then, the comprehensive evaluation index is defined by the first objective function and the second objective function after handling the constraints, and the formula is expressed as follows:

[0078] ;

[0079] in, And satisfy and indicate the goal and The weights, these values ​​depend on the actual needs. This is achieved by introducing configurable weights. The two sub-objectives are merged into a unified scalar evaluation function, which serves as the comprehensive evaluation index driving the search in the LBPQI algorithm. This allows the algorithm to guide the robot in each iteration to weigh the trade-offs between "finding the highest temperature" and "shortening the search distance" based on this metric. However, relying solely on the comprehensive evaluation metric to drive the algorithm in each iteration makes the quality of the solution overly dependent on the manual setting of weights, and it cannot obtain a set of balanced solutions simultaneously in a single optimization. Therefore, the Pareto front algorithm is introduced: in each iteration, the algorithm first selects the non-dominated solution set based on the Pareto dominance relation. ,Right now , express Then, the position with the highest comprehensive evaluation index is selected as the globally optimal individual. That is, the globally optimal individual index is represented as This individual position represents the current best compromise estimate of the main fire source location. To maintain directional consistency in the search process, the globally optimal position obtained from the previous generation is used... Dynamically set as online prediction source ,Right now This guides the robot swarm to continuously and purposefully approach the fire source in each iteration.

[0080] S204, iterate the movement of the heat-resistant reconnaissance robot group until the preset maximum number of iterations is reached, record the position of the globally optimal individual in each iteration as the globally optimal individual position sequence, and form a discrete optimal path.

[0081] Since a globally optimal individual is obtained based on the LBPQI algorithm and the Pareto front algorithm in each iteration, the positions of the globally optimal individuals from each iteration are concatenated in temporal order to form a sequence of globally optimal individual positions. This sequence, containing the optimal positions from each iteration, forms a discrete optimal path based on the positions and temporal order. Heavy firefighting vehicles can then travel along this discrete optimal path to ultimately reach the location of the forest fire source. All key state information is mapped to a driving strategy, allowing... You can get ,in Indicates from arrive The mapping. Therefore, the definition. So the robot's driving strategy is... This allows the robot commander to move. Therefore, as the iteration progresses, the suspected primary fire source obtained in each round... This forms a path that leads to the final, found main fire source, i.e., the discrete optimal path. This path not only recorded the robot's reconnaissance trajectory but also served as the basic passage for subsequent firefighting vehicles.

[0082] An optional implementation of this embodiment is as follows: the quantum imitation algorithm with learning behavior promotion (LBPQI algorithm) includes: comparing the current individual position of each heat-resistant reconnaissance robot with the historical individual position of the previous iteration, selecting the best one to update itself according to the comprehensive evaluation index, determining the individual optimal, and using the best one among all individual optimals to determine the new global optimal individual, while using the mean of all individual optimals to determine the average optimal individual.

[0083] The rules for updating individual and group memories in the LBPQI algorithm are as follows: Each individual's current position is compared to its historical best position; the better position is taken as the update value for this iteration, i.e., the individual best position is selected. Then, the globally optimal individual for this iteration is determined from all individual best positions. Simultaneously, the mean of all individual best positions is calculated to determine the average optimal individual (vector) for subsequent calculations. For example, the individual best position... This is the first Each individual robot stores its optimal position discovered in the search history, forming an individual best record, based on the current position of the heat-resistant robot in the current iteration. If the objective function value is better than the historical record, then update the record. Conversely, maintaining the original optimal state is expressed by the following formula:

[0084] ;

[0085] in, The evaluation metric representing maximization is used to evaluate the search direction (it can be one of the objective functions in a single-objective optimization task, and the weighted sum of multiple objective functions in a multi-objective optimization task). For a switching operator (or selection function), according to The comparison result selects the better one from the two inputs as the output. To ensure algorithm consistency and indirectness, this operator is uniformly used in all processes that require a binary choice based on the evaluation value. Global optimal individual. ,in This is the index of the individual, and the current globally optimal individual of the population is selected by comparing the optimality of each individual. Average optimal individual. Optimal for all individuals The average is obtained, where , It reflects the distribution center of the population and is used to make the average level of the entire population move closer to the globally optimal individual during the learning process and maintain the diversity of the population, providing a balance in updates to prevent over-convergence.

[0086] Each heat-resistant reconnaissance robot calculates its learning direction based on the individual positions of the globally optimal individual and the average optimal individual, forming an intermediate variable for role model learning.

[0087] In the LBPQI algorithm, the first step is role model learning. For each individual heat-resistant reconnaissance robot, the globally optimal individual is used as a role model for learning. The globally optimal individual and the average optimal individual jointly regulate the learning direction, preventing the group from excessively converging towards the global optimum. Then, a selection function retains the better individuals, forming intermediate variables for role model learning. For example, each individual updates itself by learning from the globally optimal individual in the group, referencing the globally optimal individual in the group. The direction guides the population to explore the current optimal region in a coordinated and efficient manner, avoiding blind exploration, and using the average optimal individual... Reflecting the overall distribution of the population, it adjusts the guidance direction for global optimization, effectively preventing the population from becoming too homogenous prematurely. The formula for role model learning is expressed as follows:

[0088] ;

[0089] in, Indicates role model learning factor, and It is a random number. This represents the guiding coefficient. This strategy will... Decomposing the dimensional vector into independent components for updating not only simplifies the solution process in high-dimensional space but also significantly enhances population diversity by introducing independent randomness in each dimension. This provides crucial diversity assurance for the algorithm to enhance its global exploration capabilities and avoid prematurely getting trapped in local optima. The intermediate variables obtained from the role model learning are used to temporarily store the new positions generated after role model learning, thus providing comparison objects for subsequent selections without overwriting the original positions. The selection logic is the same as above. To ensure that a better location is retained for evaluation As an intermediate variable in role model learning.

[0090] Each heat-resistant reconnaissance robot is compared with another random individual, and adjustments are made based on the comparison results to form a peer-learning position.

[0091] In the learning phase, after role model learning, peer learning follows. For each individual heat-resistant robot, another individual is randomly selected for comparison. Adjustments are made based on the comparison results, and the superior robot is selected as the learning target, thus obtaining the peer learning position. For example, the first... Individual basis Select the individual As a learning object, among them From The formula is randomly selected from the data and updated as follows:

[0092] ;

[0093] Among them, random numbers It is a peer learning factor. It indicates that if the current... If an individual's state is better, it continues exploring in its own direction; otherwise, it adjusts towards the other's state. The selection logic is the same as above: Given , and This selection process allows for the acquisition of learning positions from peers. This random pairing and bidirectional adjustment mechanism effectively promotes information exchange within the population, maintaining consistency in the group's search direction while allowing for the timely absorption of better information. This helps prevent individual agents from deviating from the overall trend or becoming trapped in localized areas.

[0094] Based on the peer learning position, the individual optimal, the global optimal, and the average optimal individuals are updated respectively to obtain the updated individual optimal, the updated global optimal, and the updated average optimal individuals.

[0095] After completing the two learning processes in the learning phase, the optimal record after learning is updated. Based on the peer learning position, the same method is used to re-determine the individual optimal, the globally optimal, and the average optimal individuals, and the corresponding values ​​are updated accordingly. For example, to determine the peer learning position based on the new state... To update various optimal records while avoiding symbol confusion, new candidate optimal representations are introduced for each: the updated individual optimal is denoted as... The globally optimal individual after the update is denoted as The updated average best individual is denoted as , represent the historical best record of each individual in the group, the historical best record of the group, and the average position of the group distribution, respectively. Their update method is the same as the selection rule described above. By comparison Optimal with the original individual ,in accordance with Choose the better one. Optimal from all new individuals Select The one with the highest value, Optimal from all new candidate individuals The arithmetic mean of each dimension is taken, which reflects the center of the updated population distribution. The first two aim to search for the location of the maximum value of the objective function or fitness function, while the average optimality reflects the center of the population distribution. The role of these three optimalities is to construct a neutralizing attractor in quantum mimicry, ensuring their own advantage. And then towards the global optimum Closer, and through To prevent being too inclined convergence.

[0096] By randomly weighting and combining the updated individual optimal, the updated global optimal, and the updated average optimal individuals, a neutral attractor is constructed and quantum mimicry is performed.

[0097] After obtaining the updated optimal individual, the updated globally optimal individual, and the updated average optimal individual, a random weighted combination is performed. The weights of the three are randomly assigned and summed to ensure the total weight is 1, constructing a neutral attractor for forming a potential well in the quantum mimicry stage. A potential well is a region where the potential energy is lower than the surrounding environment; particles within it experience a restoring force pointing towards the bottom of the potential well. Therefore, the lowest point of the potential well is the stable attractor of the system. The gradient of the potential energy distribution determines the direction of particle motion. Drawing on this attraction trend, a neutral attractor is constructed as the position guidance center for the individual, and a quantum mimicry-like approach is used... The probability distribution derived from the potential well bound states is used to update the new position of the individual. The scale parameter of this probability distribution corresponds to the range of the potential well; adjusting it controls the balance between the algorithm's exploration capability and convergence speed.

[0098] Specifically, constructing neutralizing attractors and performing quantum mimicry includes:

[0099] By utilizing the updated individual optimal, the updated global optimal, and the updated average optimal individuals, and by introducing a comprehensive reference position that randomly combines the mean of the individual optimal in the group with the mean of the local neighbors, a neutral attractor is formed through random weighted combination.

[0100] Suppose each individual has a center as a neutralizing attractor. of Within the potential well, its size gradually decreases with iteration. In the early stages, the potential well is large, allowing individuals to freely explore a wide range; in the later stages, the potential well is small, and individuals only perform detailed searches near the attractor. This allows for finding more possibilities early on and converging to the optimal solution faster and more stably in the later stages. The formula for constructing the neutralizing attractor is as follows:

[0101] ;

[0102] in, It is a random number and satisfies ; Defined the first The comprehensive average of individuals, the first individual Each component includes global average information. , is the th of the average optimal individual Each component represents the average of the historical best position of each individual in the population to reflect the global trend; it also includes local neighbor information. Defined as an individual Centered on, with radius The neighbor information within the neighborhood, i.e., the set All neighboring individuals (the first) The position of the individual) The average of the components, i.e. . Through random weights This mechanism integrates global distribution trends with local environmental states to generate a comprehensive reference position for each individual. This allows the neutralizing attractor to simultaneously consider both global orientation and local refinement, enhancing the algorithm's ability to balance large-scale and fine-grained searches, thus improving algorithm performance.

[0103] Based on quantum mimicry behavior, probabilistic exploration is carried out with neutralizing attractors as the center, and the position of each heat-resistant reconnaissance robot is updated in a probabilistic manner to form the position of the quantum mimicry individual.

[0104] Based on quantum mimicry behavior, probabilistic exploration is performed centered on the constructed neutralizing attractor to update the population's [number]th [unit] during this iteration. The evolution of each component is expressed by the following formula:

[0105] ;

[0106] in, For random parameters in quantum sampling, after The transformation yields an offset that follows an exponential distribution; It is the length of the feature vector in this dimension, used to control the individual's orbit around the attractor. The search scope; The coefficient of compression-expansion increases with the number of iterations. The increase gradually decreases, causing the search to transition from a "large range, high randomness" to a "smaller range, precise convergence." Intuitively, this update method reflects quantum mimicry behavior; particles do not move directly towards the neutralizing attractor, but rather sample near the attractor according to a quantum probability distribution. Initially, due to... Larger size, wider feature length, allowing particles to freely explore a larger area; later stage As the particle size gradually decreases, it will only perform a fine-grained search near the neutralizing attractor, thus achieving a natural transition from "global exploration" to "local convergence." Therefore, from the... The iteration to the... Next, the first of the population The formula for updating the expression of an individual is as follows:

[0107] ;

[0108] For example, the execution process of the LBPQI algorithm is as follows:

[0109] Input: Parameters (Population size) (Total number of iterations) (Difficulty of the problem) and (Neighbor distance threshold)

[0110] Initialization: Generate population ,in ,make and ,

[0111] while do

[0112] Learning behaviors:

[0113] for from to do

[0114] / / If the current individual is better than its own historical record, update its own optimal state.

[0115] calculate

[0116] end for

[0117] / / Find the globally optimal individual in the current population and calculate the average optimal individual of the population.

[0118] choose And calculate

[0119] for from to do

[0120] Generate random numbers

[0121] for from to do

[0122] / / Learn from the globally optimal individual in the group (role model learning)

[0123] generate

[0124] end for

[0125] / / Preserve better results

[0126] get

[0127] end for

[0128] for from to do

[0129] Generate random numbers And obtain the set

[0130] for from to do

[0131] / / Randomly select a peer in the population to learn from, increasing exploration diversity.

[0132] calculate

[0133] end for

[0134] get

[0135] end for

[0136] for from to do

[0137] Calculate the new optimal individual

[0138] end for

[0139] / / After learning from role models and peers, we need to update the globally optimal individual and the average optimal individual.

[0140] Select a new globally optimal individual And calculate the new average optimal individual.

[0141] Quantum mimicry behavior:

[0142] for from to do

[0143] Generate random numbers

[0144] for from to do

[0145] Obtain a neutral attractor

[0146] Generate random numbers

[0147] / / By using neutral attractors and probabilistic exploration, we can improve the situation of getting stuck in local optima and enhance global search capabilities.

[0148] if then

[0149]

[0150] else

[0151]

[0152] end for

[0153] / / Preserve the optimal position after quantum mimicry behavior to complete this iteration's update.

[0154] renew

[0155] end for

[0156] i i +1

[0157] end while

[0158] Output: Globally optimal individual .

[0159] S205. Based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed of the heavy fire extinguishing vehicle to obtain the optimal speed strategy.

[0160] Specifically, heavy fire-fighting vehicles are used to optimize and convert their travel speeds based on discrete optimal paths.

[0161] After using the Pareto front algorithm and the LBPQI algorithm to solve the problem in a coordinated manner, the location of the main fire source with the highest temperature was found and the discrete optimal path was determined. The heavy fire-fighting vehicle then needs to follow the determined path. To reduce computational complexity, the speed optimization problem for heavy firefighting vehicles is decoupled from path-speed co-optimization into a single-variable speed optimization problem. The optimization variable is simplified from two-dimensional position coordinates to a single linear velocity; therefore, the algorithm's dimension parameter is set to M=1. First, the total length of the discrete optimal path is calculated, which can be determined by the sum of the distances between all points along the path.

[0162] The quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed based on the total path length of the discrete optimal path, and solve for the optimal speed strategy that minimizes the third objective function.

[0163] To minimize energy consumption and ensure mission endurance, after obtaining the total path length, the LBPQI algorithm is used again for iterative optimization of vehicle speed. Leveraging the global search and balancing capabilities of the LBPQI algorithm in the speed dimension, each updated individual in the iteration is the driving speed of the heavy fire extinguishing vehicle. The algorithm eventually converges to the optimal speed value (specifically a linear velocity value) that minimizes the third objective function, thus obtaining the optimal speed strategy that minimizes the energy consumption objective.

[0164] S206, the heavy firefighting vehicle travels along a discrete optimal path and at an optimal speed strategy, ultimately reaching the predicted location of the forest fire source to perform firefighting tasks. The predicted forest fire source location is the position of the globally optimal individual determined in the last iteration of the heat-resistant reconnaissance robot swarm. Finally, on the path explored by the heat-resistant robot swarm, the heavy firefighting vehicle travels to the predicted main fire source location (i.e., the globally optimal individual position from the last iteration) at the optimal speed strategy planned by the algorithm. This significantly improves the endurance and mission reliability of the heavy vehicle in complex fire environments while ensuring positioning accuracy, thus fully realizing how to solve the multi-target source localization problem in forest fires.

[0165] For example, in this embodiment, a simulated fire scene containing three heat sources is defined by setting three sets of parameters: the center coordinates of each heat source are... Set the heat source influence range as Given three heat sources with intensities of This indicates that the second heat source has the highest intensity, therefore the location of the main heat source is... To assess algorithm effectiveness, typical parameter settings for heavy-duty firefighting vehicles are shown in Table 1.

[0166] Table 1 Typical parameters of heavy-duty fire-fighting vehicles

[0167]

[0168] At the same time, the boundary constraints of the fire area are set to , , , ,(Right now , The speed upper and lower bound constraints for heavy firefighting vehicles are set as follows: , (Right now To ensure the strictness of constraint handling and the stability of numerical calculation, this embodiment sets a penalty parameter. This value is large enough to ensure the penalty for constraint violations, thereby driving the optimization solution to strictly satisfy all physical constraints; at the same time, the smoothing parameter r is set to 0.01, which is small enough to ensure the smoothness of the function. right The operator has high approximation accuracy, thus enabling precise measurement of constraint violations. The number of heat-resistant robots can be set. Z=n=30 Maximum number of iterations I= 35 Distance threshold to neighbors And set the population size for the heavy fire-fighting vehicle speed optimization strategy. Z=n=20 Maximum number of iterations I=20 and keep the same Set the target weight coefficients respectively. =0.9, =0.1. The two-dimensional iterative process for finding forest fire sources is obtained, as follows: Figure 3 As shown in the figure Represents the x-axis. The vertical axis represents the coordinate. The iterative process for finding the optimal linear velocity of heavy firefighting vehicles is obtained, as follows: Figure 4 As shown in the figure Representing the The second iteration Each individual speed is optimized. The final path to find the main fire source is as follows: Figure 5 As shown.

[0169] This embodiment establishes a multi-target source localization problem using three objective functions: a first objective function with fire area boundary constraints and upper and lower speed limits, respectively. L1 penalty and smoothing functions are used to constrain and smooth each objective function with constraints. The LBPQI algorithm sequentially executes role model learning, peer learning, and quantum mimicry behavior centered on neutralizing attractors to drive group co-evolution. Combined with Pareto dominance, the globally optimal individual is dynamically determined from the non-dominated solution set, effectively coordinating the trade-offs between conflicting objectives and enhancing the algorithm's optimization under strict physical constraints. The feasibility of the capability and solution reduces the constraint processing error and ensures that the generated path and speed strategies strictly comply with safety specifications. It improves the convergence performance and global exploration capability of the multi-objective collaborative optimization algorithm. The learning phase accelerates the collaborative convergence of the group to the high-quality region. The quantum imitation phase uses the neutralizing attractor to fuse global and local information for dynamic and probabilistic global exploration, which can effectively overcome the problems of premature convergence and getting trapped in local optima. It enhances the practicality and adaptability of the method and realizes multi-objective collaborative optimization with strict constraint satisfaction, global search and local fine balance, and significant energy consumption reduction in the multi-objective source localization of forest fires.

[0170] Example 3

[0171] Figure 6 This is a schematic diagram of a quantum behavior-driven multi-target forest fire source locator according to Embodiment 3 of the present invention. In this embodiment, the quantum behavior-driven multi-target forest fire source locator includes:

[0172] The multi-objective optimization model construction module 810 is used to establish a multi-objective optimization model for forest fires, and to perform collaborative optimization on three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles.

[0173] The optimal path iteration module 820 is used to utilize a swarm of heat-resistant reconnaissance robots to iterate based on a quantum imitation algorithm with learning behavior promotion, in order to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source to form a discrete optimal path;

[0174] The optimal speed iteration module 830 is used to perform iterative calculations again using a quantum imitation algorithm with learning behavior promotion based on the total path length of the discrete optimal path to obtain the optimal speed strategy.

[0175] The task execution module 840 is used to enable heavy fire-fighting vehicles to travel along discrete optimal paths and at optimal speeds, ultimately reaching the predicted location of the forest fire source to carry out fire-fighting tasks.

[0176] This embodiment establishes a multi-objective optimization model for forest fires through a multi-objective optimization model construction module. An optimal path iteration module utilizes a swarm of heat-resistant reconnaissance robots to determine discrete optimal paths. An optimal speed iteration module determines the optimal speed strategy based on the total length of the discrete optimal paths. A task execution module enables heavy fire-fighting vehicles to perform fire-fighting tasks along the discrete optimal paths and at the optimal speed strategy. This solves the multi-objective collaborative optimization problem of fire source location accuracy, search efficiency, and the delivery endurance of heavy fire-fighting vehicles in complex forest fire rescue scenarios. It achieves complete process and end-to-end collaborative optimization from intelligent fire source search and vehicle path planning to energy-saving execution. The two-stage optimization seamlessly connects the tasks of heat-resistant robots and heavy fire-fighting vehicles, improving the reliability and success rate of the overall rescue mission. The LBPQI algorithm effectively guides the heat-resistant robot swarm to avoid getting stuck in local high-temperature points, ensuring rapid and accurate location of the true main fire source and providing a high-quality path foundation for the subsequent movement of heavy fire-fighting vehicles, thus improving the accuracy and robustness of forest fire source location.

[0177] The quantum behavior-driven multi-target forest fire source localization device provided in the embodiments of the present invention can execute the quantum behavior-driven multi-target forest fire source localization method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0178] Example 4

[0179] Figure 7 This is a structural diagram of an electronic device according to Embodiment 4 of the present invention. Figure 7 A block diagram is shown of an exemplary electronic device 12 suitable for implementing embodiments of the present invention. Figure 7 The electronic device 12 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0180] like Figure 7 As shown, the electronic device 12 is represented in the form of a general-purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0181] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0182] Electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0183] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (… Figure 7 Not shown; usually referred to as a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0184] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0185] Electronic device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with the electronic device 12 / server / computer, and / or with any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through input / output (I / O) interface 22. Furthermore, electronic device 12 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 20. Figure 7 As shown, network adapter 20 communicates with other modules of electronic device 12 via bus 18. It should be understood that, although... Figure 7As not shown, other hardware and / or software modules may be used in conjunction with electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0186] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the quantum behavior-driven multi-target forest fire source location method provided in the embodiments of the present invention.

[0187] Example 5

[0188] Embodiment 5 of the present invention also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the quantum behavior-driven multi-target forest fire source localization method provided in the above embodiments.

[0189] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0190] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0191] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0192] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0193] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A multi-target forest fire source localization method based on quantum behavior, characterized in that, include: S101. Establish a multi-objective optimization model for forest fires and conduct collaborative optimization of three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles. S102, a group of heat-resistant reconnaissance robots is used to iterate based on a quantum imitation algorithm with learning behavior promotion to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source. The Pareto front algorithm is used to determine the globally optimal individual in each iteration, and the position sequence of the globally optimal individual in each iteration is recorded to form a discrete optimal path. The quantum imitation algorithm with learning behavior promotion is based on the quantum particle swarm optimization (QPSO) algorithm, introduces peer learning and role model learning processes, and uses local averaging and global averaging to construct a neutralizing attractor. S103, based on the total path length of the discrete optimal path, the quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed of the heavy fire extinguishing vehicle to obtain the optimal speed strategy. S104 enables the heavy firefighting vehicle to travel along a discrete optimal path and at an optimal speed, eventually reaching the predicted location of the forest fire source to perform firefighting tasks. The predicted location of the forest fire source is the location of the globally optimal individual determined by the heat-resistant reconnaissance robot swarm in the last iteration.

2. The method according to claim 1, characterized in that, S101 includes: A multi-objective optimization model for forest fires was established, including a first objective function characterizing the detection intensity of the Gaussian temperature field at the fire site, a second objective function characterizing the distance between the heat-resistant reconnaissance robot and the predicted fire source, and a third objective function characterizing the energy consumption of heavy firefighting vehicles. Initialize the parameters of each objective function separately, and randomly initialize the initial search population of the heat-resistant reconnaissance robot swarm.

3. The method according to claim 2, characterized in that, S101 further includes: The first and second objective functions are constrained by the fire area boundary using the L1 penalty function, the third objective function is constrained by the upper and lower limits of the driving speed using the driving speed limit constraint, and the smoothing is performed using the smoothing function, while ensuring the minimum approximation error.

4. The method according to claim 1, characterized in that, S102 includes: The position-state vector of each heat-resistant reconnaissance robot is mapped using a quantum imitation algorithm with learning behavior promotion, the position of each heat-resistant reconnaissance robot at the next moment is calculated, and the corresponding heat-resistant reconnaissance robot is driven to move to that position. The Pareto dominance relation is used to identify the non-dominated solution set in the heat-resistant reconnaissance robot group and calculate the comprehensive evaluation index of each individual in the solution set. Based on the comprehensive evaluation index, the current global best individual is determined. At the same time, a position state vector is constructed for each heat-resistant reconnaissance robot to represent the individual position, historical individual best information, and global best individual information of each heat-resistant reconnaissance robot. The movement process of the heat-resistant reconnaissance robot group is iterated until the preset maximum number of iterations is reached. The position of the globally optimal individual in each iteration is recorded as the globally optimal individual position sequence, and a discrete optimal path is formed.

5. The method according to claim 1, characterized in that, S103 includes: The driving speed of heavy firefighting vehicles is optimized and converted based on discrete optimal paths. The quantum imitation algorithm with learning behavior promotion is used again to iteratively calculate the driving speed based on the total path length of the discrete optimal path, and solve for the optimal speed strategy that minimizes the third objective function.

6. The method according to claim 4, characterized in that, The quantum imitation algorithm with learning behavior enhancement includes: The current position of each heat-resistant reconnaissance robot is compared with the historical position of each robot in the previous iteration. The best one is selected based on the comprehensive evaluation index to update itself, thus determining the individual best. The best one among all individual bests is used to determine the new global best individual, and the average best individual is used to determine the average best individual. Each heat-resistant reconnaissance robot calculates its learning direction based on the individual position of the globally optimal individual and the individual position of the average optimal individual, forming an intermediate variable for role model learning; Each heat-resistant reconnaissance robot is compared with another random individual, and adjustments are made based on the comparison results to form a peer-learning position; Based on the peer learning position, the individual optimal, the global optimal, and the average optimal individuals are updated respectively to obtain the updated individual optimal, the updated global optimal, and the updated average optimal individuals; By randomly weighting and combining the updated individual optimal, the updated global optimal, and the updated average optimal individuals, a neutral attractor is constructed and quantum mimicry is performed.

7. The method according to claim 6, characterized in that, The construction of the neutralizing attractor and the quantum mimicry include: By utilizing the updated individual optimal, the updated global optimal, and the updated average optimal, and by introducing a comprehensive reference position that randomly merges the mean of the individual optimal in the group with the mean of the local neighbors, a neutral attractor is formed through random weighted combination. Based on quantum mimicry behavior, probabilistic exploration is carried out with neutralizing attractors as the center, and the position of each heat-resistant reconnaissance robot is updated in a probabilistic manner to form the position of the quantum mimicry individual.

8. A quantum behavior-driven multi-target forest fire source localization device, used to implement the quantum behavior-driven multi-target forest fire source localization method as described in any one of claims 1-7, characterized in that, include: The multi-objective optimization model construction module is used to establish a multi-objective optimization model for forest fires, and to perform collaborative optimization on three objectives: the intensity of Gaussian temperature field detection at the fire site, the distance between the heat-resistant reconnaissance robot and the online predicted fire source, and the energy consumption of heavy firefighting vehicles. The optimal path iteration module is used to utilize a swarm of heat-resistant reconnaissance robots to iterate based on a quantum imitation algorithm with learning behavior promotion, in order to collaboratively optimize the detection intensity of the Gaussian temperature field in the fire and the distance between the heat-resistant reconnaissance robots and the online predicted fire source to form a discrete optimal path; The optimal speed iteration module is used to perform iterative calculations again using a quantum imitation algorithm with learning-enhanced behavior based on the total path length of the discrete optimal path, in order to obtain the optimal speed strategy. The task execution module is used to enable heavy firefighting vehicles to travel along discrete optimal paths and at optimal speeds, ultimately reaching the predicted location of the forest fire source to carry out firefighting tasks.

9. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the quantum behavior-driven multi-target forest fire source localization method as described in any one of claims 1-7.

10. A storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the quantum behavior-driven multi-target forest fire source localization method as described in any one of claims 1-7.

Images