Heuristic-driven uav multi-objective mission planning and path planning system

Abstract

The present application relates to the technical field of unmanned aerial vehicles, in particular to a heuristic-driven unmanned aerial vehicle multi-target task planning and path planning system. The present application decouples tasks into unmanned aerial vehicle, task and battlefield environment models through a model module, and is connected with an optimization algorithm module. The optimization algorithm module receives the decoupled model data, determines the planning target and generates the optimization result, and feeds back to the model module. The path planning module generates the unmanned aerial vehicle flight path according to the optimization result. The target identification module receives real-time image data, and detects and identifies the target by using a pre-trained model. The target tracking module receives the identified target information, and calculates the predicted trajectory. The algorithm optimization module iterates and mutates the algorithm based on historical data and dynamic weight optimization. The system effectively improves the task completion efficiency, path optimization, safety and adaptability, and opens up new possibilities for the intelligent application of unmanned aerial vehicles in complex dynamic environments.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a heuristic-driven UAV multi-objective mission planning and trajectory planning system. Background Technology

[0002] With the rapid development of UAV technology, multi-target UAV mission planning systems are increasingly being used in fields such as military reconnaissance, disaster relief, and environmental monitoring. However, facing increasingly complex mission environments and diverse mission requirements, existing UAV mission planning systems still face many challenges and shortcomings.

[0003] Traditional UAV mission planning methods typically employ a single optimization algorithm, such as genetic algorithms or particle swarm optimization, to handle multi-objective planning problems. While these methods can solve simple mission planning problems to some extent, they often perform poorly in complex multi-objective scenarios. First, a single algorithm struggles to effectively balance multiple conflicting objectives, such as mission completion rate, flight distance, execution time, and path safety. Second, these methods are computationally inefficient when facing large-scale missions and complex environments, making them unsuitable for real-time planning. Furthermore, traditional methods lack adaptability to dynamic environments and cannot effectively handle unexpected situations and environmental changes that occur during missions.

[0004] Existing technologies include some improved multi-objective optimization methods, such as the weighted method or the Pareto optimal method, to handle multi-objective problems. However, these methods still have some inherent limitations. The weighted method requires pre-setting the weights of each objective, which is often difficult to determine accurately in practical applications. Although the Pareto optimal method can obtain a set of non-dominated solutions, it lacks an effective mechanism for solution selection and decision-making. Furthermore, when dealing with high-dimensional, large-scale task planning problems, the computational complexity of these methods increases dramatically, making it difficult to meet the needs of practical applications.

[0005] In path planning, existing technologies mostly employ traditional A* algorithms or Rapid Expanding Random Tree (RRT) algorithms. These algorithms perform reasonably well in simple environments, but often struggle to generate optimal paths in complex 3D spaces and dynamic environments. Especially when multiple threat zones and obstacles exist, these algorithms find it difficult to balance path length and safety requirements.

[0006] Furthermore, existing UAV mission planning systems typically separate functions such as mission allocation, path planning, and target identification, lacking effective integration and coordination mechanisms. This makes it difficult for the system to fully consider the interplay of various factors, thus affecting the overall planning effectiveness and efficiency. Summary of the Invention

[0007] This invention aims to solve the aforementioned technical problems by proposing a heuristic-driven UAV multi-objective mission planning and trajectory planning system. Through innovative modular design and the synergy of advanced algorithms, this system achieves efficient planning and optimization for complex multi-objective UAV missions.

[0008] This invention proposes a heuristic-driven UAV multi-objective mission planning and trajectory planning system, comprising:

[0009] The model module is used for:

[0010] The mission model is decoupled into a drone model, a mission model, and a battlefield environment model;

[0011] Transmit the decoupled model data to the optimization algorithm module;

[0012] The optimization algorithm module, which is data-connected to the model module, is used for:

[0013] Receive the decoupled model data sent by the model module;

[0014] Based on the decoupled model data, the planning objective is determined and the optimization result is generated;

[0015] The optimization results are then fed back to the model module.

[0016] The path planning module, which is data-connected to the optimization algorithm module, is used for:

[0017] Receive the optimization results generated by the optimization algorithm module;

[0018] Based on the optimization results, a drone flight path is generated;

[0019] The target recognition module, which is data-connected to the path planning module, is used for:

[0020] Receive real-time image data;

[0021] Based on a pre-trained target recognition model, targets are detected and recognized from the real-time image data;

[0022] The target tracking module, which is data-connected to the target recognition module, is used for:

[0023] Receive the target information identified by the target recognition module;

[0024] Calculate the target's predicted trajectory based on the trajectory prediction model;

[0025] The algorithm optimization module, which is data-connected to the optimization algorithm module, is used for:

[0026] Based on historical data analysis results and an adaptive mutation operator with dynamic weights, the algorithm's iteration count and mutation operator are optimized.

[0027] Preferably, the model module includes:

[0028] The drone model unit is used to store and update the drone's constraints and capabilities data;

[0029] The task model unit is used to store and update task location, region, timing, and priority data;

[0030] Battlefield environment model unit, used to store and update mission information, threat information, electromagnetic environment information and terrain data within the battlefield area;

[0031] The model decoupling unit is connected to the UAV model unit, mission model unit, and battlefield environment model unit for data decoupling of the received data from each unit, generating decoupled model data.

[0032] Preferably, the optimization algorithm module includes:

[0033] Multi-objective optimization unit, used to construct multi-objective optimization problems based on the MOEA / D (Multi-Objective Evolutionary Algorithm based on Decomposition) framework;

[0034] The decomposition and optimization unit, which is data-connected to the multi-objective optimization unit, is used to decompose the multi-objective optimization problem into a trajectory optimization sub-problem and a task sequence optimization sub-problem.

[0035] The genetic algorithm unit, which is data-connected to the decomposition and optimization unit, is used to solve the trajectory optimization subproblem and the task sequence optimization subproblem.

[0036] The non-dominated sorting unit is data-connected to the genetic algorithm unit and is used to perform non-dominated sorting on individuals in the population.

[0037] The elite strategy unit, which is data-connected to the non-dominated sorting unit, is used to retain and update the elite solution set.

[0038] Preferably, the path planning module includes:

[0039] Local path planning network is used to perform local path planning for a target point and calculate the flight path of the UAV to the target point;

[0040] A global path planning network is used to establish flight paths between various target points for UAVs.

[0041] The path optimization unit is connected to the local path planning network and the global path planning network for optimizing the planning results of the global path planning network based on the output results of the local path planning network.

[0042] Preferably, the target recognition module includes:

[0043] The image preprocessing unit is used to preprocess the received real-time image data;

[0044] An attention model unit, which is data-connected to the image preprocessing unit, is used to extract features from the preprocessed image;

[0045] The target classification unit is data-connected to the attention model unit and is used to classify targets based on the extracted features.

[0046] The target localization unit is data-connected to the target classification unit and is used to determine the position of the classified target in the image.

[0047] Preferably, the target tracking module includes:

[0048] Trajectory prediction networks are used to predict the future trajectory of a target based on historical target location data.

[0049] A trajectory smoothing unit, connected to the trajectory prediction network data, is used to smooth the predicted trajectory;

[0050] The target guidance unit is data-connected to the trajectory smoothing unit and is used to generate UAV guidance commands based on the smoothed predicted trajectory.

[0051] Preferably, the algorithm optimization module includes:

[0052] The iterative optimization unit is used to determine the number of iterations based on the difference between the distance between the UAV and the target calculated each time and the current distance between the UAV and the target.

[0053] The mutation operator optimization unit is used to calculate the differences between individuals in the population and individuals in the previous generation, as well as the stability of the individuals in the population themselves, and to generate adaptive mutation operators.

[0054] The parameter adaptive unit is data-connected to the iterative optimization unit and the mutation operator optimization unit, and is used to dynamically adjust the parameters in the MOEA / D framework.

[0055] Preferably, the genetic algorithm unit includes:

[0056] Encoding units are used to encode UAV tracks and mission sequences into chromosomes;

[0057] Crossover units are used to perform SBX (Simulated Binary Crossover) operations on selected parent chromosomes.

[0058] The mutation unit is used to perform PM (Polynomial Mutation) mutation operations on the offspring chromosomes after crossover;

[0059] Fitness calculation unit, used to calculate the multi-target fitness value of chromosomes;

[0060] The selection unit is used to select superior individuals to enter the next generation based on their fitness values.

[0061] Preferably, a constraint processing module is also included, which is data-connected to the optimization algorithm module, and is used for:

[0062] Receive the candidate solutions generated by the optimization algorithm module;

[0063] The feasibility of the candidate solutions is evaluated based on preset constraints.

[0064] The evaluation results are fed back to the optimization algorithm module to guide the next round of optimization.

[0065] Preferably, the method includes the following steps:

[0066] S1, obtain the constraints and capabilities model of the UAV, as well as the location, region, timing and priority model of each task;

[0067] S2 establishes a multi-objective optimization function with the optimization objectives of minimizing the drone's flight path distance and minimizing the drone's mission execution time, and with the drone's flight constraints as the constraints.

[0068] S3. The MOEA / D framework is used to construct a task optimization model, and the multi-objective optimization function is decomposed into two sub-problems: trajectory optimization and task sequence optimization.

[0069] S4, use a genetic algorithm to solve the course optimization and task sequence optimization subproblems, where:

[0070] Initialize the population and calculate the multi-objective fitness value of each individual;

[0071] Perform a non-dominated sort to obtain the non-dominated set, the outer set, and the remaining individuals;

[0072] For the remaining individuals, an elite selection strategy, SBX crossover, and PM mutation are used to generate new individuals;

[0073] Update the non-dominated set and the outer set;

[0074] S5, determine whether the termination condition is met. If not, return to step S4. Otherwise, output the optimal solution in the non-dominated set as the planning result of the UAV trajectory and mission sequence.

[0075] S6. Based on the planning results, a detailed UAV flight path is generated using a local path planning network and a global path planning network;

[0076] During mission execution, the S7 uses a target recognition module to detect and identify targets in real time, and a target tracking module to predict target trajectories and guide the drone.

[0077] S8 uses an algorithm optimization module to dynamically adjust the number of iterations and mutation operators, improving algorithm efficiency and adaptability;

[0078] S9. Repeat steps S2 to S8 until all tasks are completed or the preset termination condition is met.

[0079] The beneficial effects of this invention are:

[0080] From a macro perspective, the system of this invention achieves comprehensive optimization of multi-objective UAV mission planning. It can not only effectively balance multiple objectives such as mission completion rate, flight length, execution time, and path safety, but also adapt to complex dynamic environments, providing strong technical support for the efficient operation of UAVs in various application scenarios.

[0081] In terms of system architecture, this invention adopts a modular design, including a model module, an optimization algorithm module, a path planning module, a target recognition module, a target tracking module, and an algorithm optimization module. This design not only improves the system's flexibility and scalability but also enables effective collaboration between the functional modules. For example, the task plan generated by the optimization algorithm module can directly guide the work of the path planning module, while real-time information from the target recognition and tracking modules can be fed back to the optimization algorithm module for dynamic adjustment. This collaborative mechanism significantly improves the overall performance and adaptability of the system.

[0082] At the algorithmic level, this invention cleverly combines several advanced optimization techniques. The MOEA / D framework is employed to handle multi-objective optimization problems, effectively addressing the shortcomings of traditional methods in terms of target balancing and computational efficiency. Improved path planning algorithms, such as the optimized RRT* algorithm, can generate safer and more efficient flight paths in complex 3D environments. Deep learning-based target recognition technology and Kalman filter-based target tracking algorithms significantly enhance the system's ability to handle dynamic targets.

[0083] Of particular note is that the algorithm optimization module of this invention enables dynamic adjustment of the optimization process. By analyzing historical data and using an adaptive mutation operator, the system can automatically adjust the number of iterations and algorithm parameters according to changes in the task environment and optimization process. This not only improves the convergence speed of the algorithm but also enhances the system's adaptability to different task types.

[0084] In terms of practical application, the system of this invention exhibits significant advantages. Compared to traditional methods, it can accomplish more tasks while significantly reducing the total flight distance and average task execution time. More importantly, the flight paths generated by the system offer higher safety, which is crucial for UAVs performing tasks in complex environments. Simultaneously, the system's rapid convergence characteristics enable it to meet real-time planning requirements, providing possibilities for dynamic task adjustments for UAVs.

[0085] In summary, the multi-target UAV mission planning system of this invention effectively solves several problems existing in the prior art through innovative system design and algorithm integration. It not only achieves comprehensive improvements in mission completion efficiency, path optimization, safety, and adaptability, but also opens up new possibilities for the intelligent application of UAVs in complex and dynamic environments. The successful application of this system will greatly promote the further development and application of UAV technology in military and civilian fields. Attached Figure Description

[0086] Figure 1 This is the top-level structure of the system of the present invention;

[0087] Figure 2 This is an internal structural diagram of the model module of the present invention;

[0088] Figure 3 This is a diagram showing the internal structure of the optimization algorithm module of the present invention;

[0089] Figure 4 This is an internal structural diagram of the path planning module of the present invention;

[0090] Figure 5 This is an internal structural diagram of the target recognition module of the present invention;

[0091] Figure 6 This is an internal structural diagram of the target tracking module of the present invention;

[0092] Figure 7 This is a diagram showing the internal structure of the algorithm optimization module of the present invention; Detailed Implementation

[0093] Please refer to the attached document. Figure 1-7This invention provides a heuristic-driven system for planning and trajectories of multi-objective UAV missions. Through innovative modular design and the synergy of advanced algorithms, this system achieves efficient planning and optimization of complex multi-objective UAV missions.

[0094] The system of this invention includes a model module, an optimization algorithm module, a path planning module, a target recognition module, a target tracking module, and an algorithm optimization module. These modules work together through close data connections and interactions to complete the complex process of multi-target UAV mission planning.

[0095] Model Module 1 is responsible for decoupling the mission model into a UAV model, a mission model, and a battlefield environment model. This decoupling design enhances the system's flexibility and scalability. For example, the UAV model can include parameters such as flight speed, endurance, and payload capacity; the mission model can include information such as target location, priority, and time window; and the battlefield environment model can include data such as terrain, weather conditions, and threat distribution. Model Module 1 transmits these decoupled model data to Optimization Algorithm Module 2, providing the foundation for subsequent optimization calculations.

[0096] The optimization algorithm module 2 is the core of this system. It receives the decoupled model data sent by the model module 1, determines the planning objectives based on this data, and generates optimization results. The optimization algorithm module 2 adopts a multi-objective optimization method, taking the shortest UAV flight path distance and the shortest task execution time as optimization objectives. This multi-objective design is more in line with practical application needs and can achieve a balance between task efficiency and resource utilization.

[0097] Preferably, the optimization algorithm module 2 employs the MOEA / D (Multi-Objective Evolutionary Algorithm based on Decomposition) framework to construct and solve the multi-objective optimization problem. The core idea of ​​the MOEA / D algorithm is to decompose the multi-objective optimization problem into a series of scalar optimization subproblems, and then simultaneously optimize these subproblems. In this invention, the MOEA / D algorithm can be expressed as:

[0098] minF(x)=(f1(x),f2(x),...,f m (x)),

[0099] Where F(x) is a multi-objective optimization function, f i (x) represents the i-th objective function, and m represents the number of objective functions. In this invention, m = 2, corresponding to the track distance and mission time, respectively.

[0100] The path planning module 3 receives the optimization results generated by the optimization algorithm module 2 and generates the specific flight path of the UAV based on these results. In one embodiment of the present invention, the path planning module 3 may use an improved A* algorithm to generate the path. The improved A* algorithm can be expressed as:

[0101] f(n)=g(n)+h(n)+r(n),

[0102] Where f(n) is the evaluation function of node n, g(n) is the actual cost from the starting node to n, h(n) is the estimated cost from n to the target node, and r(n) is an additional term that takes into account task priority and time window.

[0103] The target recognition module 4 and the target tracking module 5 work together to achieve real-time target recognition and tracking. The target recognition module 4 uses a pre-trained deep learning model, such as YOLOv5 or Faster R-CNN, to detect and recognize targets from real-time image data. The target tracking module 5 uses algorithms such as Kalman filters or particle filters to calculate the predicted trajectory of the target based on the target recognition results.

[0104] Algorithm optimization module 6 is another innovation of this invention. It optimizes the number of iterations and the mutation operator by analyzing historical data and using an adaptive mutation operator with dynamic weights. This adaptive optimization mechanism can significantly improve the efficiency and robustness of the algorithm. For example, the adaptive mutation operator can be expressed as:

[0105]

[0106] Where F is the adaptive mutation operator, α is the weight coefficient, Δ is the population variability, and σ is the weight coefficient. g For population stability, F0 is the initial mutation operator.

[0107] Model module 1 further includes an unmanned aerial vehicle (UAV) model unit 11, a mission model unit 12, a battlefield environment model unit 13, and a model decoupling unit 14. This detailed design enables the system to process various types of data more precisely, improving the accuracy and adaptability of the model.

[0108] The UAV model unit 11 stores and updates the UAV's constraints and capabilities data. For example, it can include parameters such as the UAV's maximum flight speed (e.g., 50 m / s), maximum flight altitude (e.g., 5000 m), and maximum payload (e.g., 10 kg). These parameters play a crucial role in mission planning, ensuring that the generated mission plan is within the UAV's capabilities.

[0109] Task model unit 12 stores and updates task location, region, timing, and priority data. For example, task location can be represented by latitude and longitude coordinates, task region can be described by polygonal regions, timing can include the start and end time windows of the task, and priority can be represented by integers from 1 to 10. This detailed task information enables the system to generate more accurate and efficient task planning schemes.

[0110] Battlefield environment model unit 13 stores and updates mission information, threat information, electromagnetic environment information, and terrain data within the battlefield area. For example, threat information may include the location and coverage of enemy air defense systems, electromagnetic environment information may include areas of communication jamming, and terrain data may include the distribution of natural obstacles such as mountains and rivers. Taking this environmental information into account makes the mission plans generated by the system safer and more reliable.

[0111] Model decoupling unit 14 is the core of model module 1. It is responsible for decoupling the received data from each unit to generate decoupled model data. This decoupling process enables the system to process various types of data more flexibly, improving the system's scalability and adaptability.

[0112] The optimization algorithm module 2 includes a multi-objective optimization unit 21, a decomposition optimization unit 22, a genetic algorithm unit 23, a non-dominated sorting unit 24, and an elite strategy unit 25. This modular design enables the system to efficiently handle complex multi-objective optimization problems.

[0113] The multi-objective optimization unit 21 constructs a multi-objective optimization problem based on the MOEA / D framework. In one embodiment of the present invention, the multi-objective optimization problem can be expressed as:

[0114] minF(x)=(f1(x),f2(x)),

[0115] Where f1(x) represents the track distance and f2(x) represents the mission execution time.

[0116] The decomposition and optimization unit 22 decomposes the multi-objective optimization problem into a trajectory optimization subproblem and a task sequence optimization subproblem. This decomposition strategy can significantly reduce the complexity of the problem and improve the solution efficiency. For example, the trajectory optimization subproblem can be expressed as:

[0117]

[0118] Where d(x) i ,x i+1 ) represents the distance between adjacent task points, and n is the number of task points.

[0119] Genetic algorithm unit 23 is responsible for solving the trajectory optimization subproblem and the task sequence optimization subproblem. In this invention, the chromosome encoding of the genetic algorithm can adopt real number encoding, the crossover operation can adopt simulated binary crossover (SBX), and the mutation operation can adopt polynomial mutation (PM). These advanced genetic operators can improve the search efficiency and solution quality of the algorithm.

[0120] Non-dominated sorting unit 24 performs non-dominated sorting on the individuals in the population. In multi-objective optimization problems, non-dominated sorting is an important method for individual evaluation. This invention adopts a fast non-dominated sorting algorithm with a time complexity of O(MN^2), where M is the number of objectives and N is the population size.

[0121] Elite strategy unit 25 is responsible for retaining and updating the elite solution set. The elite strategy ensures that superior individuals in the population are not lost during evolution. In this invention, the size of the elite solution set can be set to 10% of the population size, which achieves a good balance between maintaining population diversity and improving the algorithm's convergence speed.

[0122] Through the coordinated operation of the above modules, the system of this invention can efficiently handle complex multi-target UAV mission planning problems and generate optimized mission execution schemes. This system not only considers the performance constraints and mission requirements of the UAV, but also fully takes into account the impact of the battlefield environment, demonstrating strong practicality and adaptability. Furthermore, the system's modular design and the application of advanced algorithms give it excellent scalability and optimization potential, laying the foundation for future functional expansion and performance improvements.

[0123] The system of the present invention further includes a path planning module 3, which comprises a local path planning network 31, a global path planning network 32, and a path optimization unit 33. This multi-level path planning structure can effectively handle complex mission environments and generate efficient and feasible flight paths.

[0124] The local path planning network 31 is responsible for planning local paths to the target point, calculating the flight path of the UAV to the target point. In one embodiment of the present invention, the local path planning network 31 can employ an improved Rapid Expanding Random Tree (RRT*) algorithm. The core idea of ​​this algorithm is to randomly sample in the state space and construct a tree by connecting the sampled points, ultimately finding a feasible path from the starting point to the target point. Preferably, the improved version of the RRT* algorithm can be expressed as:

[0125]

[0126] Where cost(x) new ) represents the cost of the new node, x near Let c(x) represent the set of nodes near the new node.near ,x new The cost is represented by ) to travel from the nearest neighbor node to the new node. This improvement can significantly enhance path quality and the algorithm's convergence speed.

[0127] A global path planning network 32 is used to establish flight paths between various target points for the UAV. In a preferred embodiment of the present invention, the global path planning network 32 employs an improved A* algorithm. Compared to the traditional A* algorithm, the improved version introduces dynamic weight factors, which can dynamically adjust the weights of the heuristic function according to the urgency of the task and the complexity of the environment. The improved A* algorithm can be expressed as:

[0128] f(n) = g(n) + w(n)·h(n),

[0129] Here, w(n) is a dynamic weighting factor, and its value range is usually [1,2]. When w(n) = 1, the algorithm degenerates into the traditional A' algorithm; when w(n) > 1, the algorithm will tend to explore nodes closer to the target, thereby speeding up the search.

[0130] The path optimization unit 33 is data-connected to the local path planning network 31 and the global path planning network 32, and is used to optimize the planning results of the global path planning network based on the output results of the local path planning network. Preferably, the path optimization unit 33 can use a B-spline curve smoothing algorithm to optimize the generated path, thereby improving the smoothness and flyability of the path. The B-spline curve can be represented as:

[0131]

[0132] Where, N i,p (t) is a p-th degree B-spline basis function, P i These are control points. By adjusting the positions of the control points, a smoother and more efficient flight path can be obtained.

[0133] The target recognition module 4 of this invention includes an image preprocessing unit 41, an attention model unit 42, a target classification unit 43, and a target localization unit 44. This modular design enables the system to efficiently and accurately identify and locate targets.

[0134] The image preprocessing unit 41 preprocesses the received real-time image data. In one embodiment of the present invention, the preprocessing process may include operations such as image scaling, histogram equalization, and data normalization. For example, the image may be scaled to 224x224 pixels to meet the input requirements of subsequent deep learning models. Histogram equalization can improve the contrast of the image, which helps in target recognition. Data normalization can normalize pixel values ​​to the range of [0, 1], which is beneficial for the training and inference of neural networks.

[0135] Attention model unit 42 is data-connected to image preprocessing unit 41 and is used to extract features from the preprocessed image. This invention preferably employs a self-attention mechanism to extract image features. The self-attention mechanism can be expressed as:

[0136]

[0137] Where Q, K, and V represent the query, key, and value matrices, respectively, and d k This represents the dimension of the key vector. This mechanism can capture long-range dependencies in an image, improving the effectiveness of feature extraction.

[0138] The target classification unit 43 is data-connected to the attention model unit 42 for target classification based on extracted features. In a preferred embodiment of the invention, the target classification unit 43 employs a convolutional neural network (CNN) structure, such as ResNet-50 or EfficientNet-B0. These advanced CNN models can achieve high-accuracy target classification while maintaining low computational complexity. For example, the core idea of ​​ResNet-50 is to introduce residual connections, which can be represented as:

[0139] H(x) = F(x) + x,

[0140] Here, H(x) represents the desired underlying mapping, F(x) is the residual function, and x is the input. This structure can effectively alleviate the gradient vanishing problem in deep networks and improve model performance.

[0141] The target localization unit 44 is data-connected to the target classification unit 43 to determine the location of the classified target in the image. The system of this invention employs a region-based target detection algorithm, such as Faster R-CNN or YOLO. Taking Faster R-CNN as an example, its target localization process can be divided into two stages: the Region Proposal Network (RPN) generates candidate regions, and then these candidate regions are accurately classified and bounding box regressed. Bounding box regression can be expressed as:

[0142] t x =(xx) a ) / w a ,

[0143] t y =(yy) a ) / h a ,

[0144] t w =log(w / w) a ),

[0145] t h=log(h / h) a ),

[0146] Where (x, y, w, h) represent the predicted bounding box parameters, (x... a y a w a h a ) represents the anchor frame parameters.

[0147] The target tracking module 5 of this invention includes a trajectory prediction network 51, a trajectory smoothing unit 52, and a target guidance unit 53. This structural design enables the system to accurately predict target motion and provide effective guidance information for the UAV.

[0148] The trajectory prediction network 51 predicts the future trajectory of the target based on historical target location data. In one embodiment of the invention, the trajectory prediction network 51 employs a Long Short-Term Memory (LSTM) network structure. The core of LSTM is its memory unit, which can be represented as:

[0149] f t =σ(W f ·[h t-1 x t ]+b f ),

[0150] i t =σ(W i ·[h t-1 x t ]+b i ),

[0151]

[0152] o t =σ(W o ·[h t-1 x t ]+b o ),

[0153] h t =o t *tanh(C t ),

[0154] Among them, f t i t o t These represent the forget gate, input gate, and output gate, respectively. (C) t h represents the cell state. t This represents the hidden state. This structure can effectively capture the temporal characteristics of target motion and improve the accuracy of trajectory prediction.

[0155] The trajectory smoothing unit 52 is data-connected to the trajectory prediction network 51 and is used to smooth the predicted trajectory. Preferably, a Kalman filter is used for trajectory smoothing in this invention. The core equations of the Kalman filter include the prediction equation and the update equation:

[0156] Prediction equation:

[0157]

[0158] Update equation:

[0159]

[0160]

[0161] in, P represents state estimation. k K represents the covariance of the estimation error. k This represents the Kalman gain. This filtering process effectively reduces noise and fluctuations in trajectory prediction.

[0162] The target guidance unit 53 is data-connected to the trajectory smoothing unit 52 and is used to generate UAV guidance commands based on the smoothed predicted trajectory. In a preferred embodiment of the invention, the target guidance unit 53 uses Proportional Navigation (PN) to generate guidance commands. The acceleration command of the PN method can be expressed as:

[0163] a c =NV c λ,

[0164] Where N is the guidance coefficient (usually taken as 3-5), V c Let λ be the approach velocity and λ be the line-of-sight angular rate. This method enables effective interception of moving targets while being computationally simple.

[0165] The algorithm optimization module 6 of this invention includes an iterative optimization unit 61, a mutation operator optimization unit 62, and a parameter adaptive unit 63. This structural design enables the system to dynamically adjust algorithm parameters, improving optimization efficiency and adaptability.

[0166] The iterative optimization unit 61 determines the number of iterations based on the difference between the distance between the UAV and the target calculated each time and the current distance between the UAV and the target. In one embodiment of the invention, the number of iterations can be determined using an adaptive step size method, expressed as:

[0167]

[0168] Among them, T k d represents the number of iterations in the k-th iteration. kLet represent the target distance in the k-th iteration, and α be an adjustment coefficient (usually between 0.1 and 0.5). This method can dynamically adjust the number of iterations based on the convergence of the optimization process, thereby improving the algorithm's efficiency.

[0169] The mutation operator optimization unit 62 is used to calculate the differences between individuals in the population and those in the previous generation, as well as the stability of the individuals themselves, to generate an adaptive mutation operator. Preferably, the adaptive mutation operator can be expressed as:

[0170]

[0171] Among them, F m For the adaptive mutation operator, F base The base variation rate is β (typically 0.01-0.05), the weighting coefficient is β (typically 0.1-0.5), Δ is the population variability, and σ is the population stability. This adaptive mechanism can achieve a balance between maintaining population diversity and improving convergence speed.

[0172] The parameter adaptation unit 63 is data-connected to the iterative optimization unit 61 and the mutation operator optimization unit 62, and is used to dynamically adjust the parameters in the MOEA / D framework. For example, the weight vector in MOEA / D can be adaptively adjusted according to the current Pareto front distribution.

[0173]

[0174] Among them, w i This represents the i-th weight vector. Let γ represent the ideal weight distribution, and γ be the learning rate (usually between 0.01 and 0.1). This adaptive adjustment allows the algorithm to better adapt to the characteristics of the problem and improve the quality of the solution.

[0175] Through the coordinated operation of the above modules, the system of this invention can achieve efficient and accurate multi-target UAV mission planning. The system not only considers path planning optimization but also integrates advanced target recognition and tracking technologies, and achieves adaptive optimization through an algorithm optimization module. This comprehensive design enables the system to perform excellently in complex and dynamic mission environments, providing strong support for the intelligent application of UAVs.

[0176] The genetic algorithm unit 23 of the present invention further includes an encoding unit 231, a crossover unit 232, a mutation unit 233, a fitness calculation unit 234, and a selection unit 235. This sophisticated modular design enables the genetic algorithm to handle complex optimization problems more efficiently.

[0177] Encoding unit 231 is responsible for encoding the UAV track and mission sequence into a chromosome. In a preferred embodiment of the invention, a real-number encoding method is used. For example, a track containing n mission points can be represented by a 2n-dimensional real-number vector:

[0178] C = (x1, y1, x2, y2, ..., x n ,y n ),

[0179] Among them, (x i ,y i () represents the coordinates of the i-th task point. This encoding method is intuitive and simple, facilitating subsequent genetic operations.

[0180] Crossover unit 232 is used to perform an SBX (Simulated Binary Crossover) operation on the selected parent chromosome. The SBX crossover operation can be represented as:

[0181] C 1,k =0.5[(1+β k )p 1，k +(1-β k )p 2,k ],

[0182] C 2,k =0.5[(1-β k )p 1，k +(1+β k )p 2,k ],

[0183] Among them, c 1,k and c 2,k p is the kth gene on the offspring chromosome. 1,k and p 2,k The κ-th gene on the paternal chromosome, β k The crossover operation is a random number whose distribution is controlled by a parameter η. Typically, η takes a value between 20 and 100. This crossover operation can well preserve the characteristics of the parent generation while producing sufficient diversity.

[0184] Mutation unit 233 is used to perform PM (Polynomial Mutation) mutation operations on the offspring chromosomes after crossover. The PM mutation operation can be represented as:

[0185] c k =p k +δ k (u k -l k ),

[0186] Among them, c kp represents the mutated gene value. k The gene value before mutation, u k and l k These represent the upper and lower bounds of the gene, δ k For a small perturbation, its distribution is controlled by a parameter η_m. Typically, η_m takes a value between 20 and 100. This mutation operation can maintain local search capability while also having a certain probability of producing larger changes.

[0187] The fitness calculation unit 234 is used to calculate the multi-objective fitness value of a chromosome. In this invention, the fitness function includes two objectives: track length and task completion time. This can be expressed as:

[0188]

[0189] Where d(x) i x i+1 ) represents the distance between adjacent task points, t i Let v represent the dwell time at the i-th task point, and v represent the average flight speed of the UAV.

[0190] Selection unit 235 is used to select superior individuals for the next generation based on fitness values. This invention employs an elite selection strategy based on non-dominated ranking. First, the population is ranked non-dominated, and then individuals are selected from highest to lowest non-dominated level until the population size reaches a preset value. When individuals at a certain non-dominated level cannot all be selected, crowding ranking is used to select individuals at that level. This selection strategy ensures both the preservation of elite individuals and the maintenance of population diversity.

[0191] The system of this invention also includes a constraint processing module 7, which is data-connected to the optimization algorithm module 2. The introduction of the constraint processing module 7 enables the system to better handle various constraints in practical applications, thereby improving the feasibility of the optimization results.

[0192] The constraint processing module 7 first receives candidate solutions generated by the optimization algorithm module 2. In one embodiment of the present invention, a candidate solution can be represented as a structure containing a task sequence and corresponding tracks:

[0193] S={(T1,P1),(T2,P2),...,(T n ,P n )},

[0194] Among them, T i Let P represent the i-th task. i This indicates the flight path used to complete the mission.

[0195] Then, constraint processing module 7 evaluates the feasibility of candidate solutions based on preset constraints. The constraints considered in this invention mainly include:

[0196] 1. Limitations on drone battery life:

[0197]

[0198] Among them, T max This represents the maximum flight time of the drone.

[0199] 2. Task time window constraints:

[0200]

[0201] Where $t_{start,i}$ and $t_{end,i}$ are the start and end time windows of the i-th task, respectively, and $t_{arrive,i}$ is the time when the drone arrives at the i-th task point.

[0202] 3. Track safety constraints:

[0203]

[0204] Where d(P,O) represents the distance from waypoint P to obstacle °, d safe This is the safe distance threshold.

[0205] The constraint processing module 7 feeds back the evaluation results to the optimization algorithm module 2 to guide the next round of optimization. Preferably, a penalty function method is used to process the constraints, converting the degree of constraint violation into a penalty term added to the objective function:

[0206]

[0207] Where F(x) is the original objective function, g i (x) is the i-th constraint function, r i The corresponding penalty coefficient can be used to gradually increase the influence of constraints during the optimization process, eventually obtaining a solution that satisfies all constraints.

[0208] This invention also provides a multi-objective UAV mission planning method based on a heuristic optimization algorithm. The method includes multiple steps, each carefully designed to achieve efficient and reliable mission planning.

[0209] Step S1 obtains the UAV's constraint and capability model, as well as the location, region, timing, and priority model for each task. This step provides the necessary foundational data for subsequent optimization processes. For example, the UAV's constraint and capability model may include parameters such as maximum flight speed, maximum flight altitude, and maximum payload, while the task model contains information such as task coordinates, execution time windows, and task priorities.

[0210] In step S2, this invention establishes a multi-objective optimization function with the shortest UAV flight path distance and the shortest UAV mission execution time as optimization objectives and the UAV's flight constraints as constraints. This multi-objective design can better balance mission efficiency and resource utilization. The optimization function can be expressed as:

[0211] minF(x)=(f1(x),f2(x)),

[0212] stg i (x)≤0,i=1,2,...,m,

[0213] Where f1(x) represents the track distance, f2(x) represents the mission execution time, and g i (x) represents various constraints.

[0214] Step S3 uses the MOEA / D framework to construct a task optimization model, decomposing the multi-objective optimization function into two sub-problems: trajectory optimization and task sequence optimization. The core idea of ​​MOEA / D is to transform the multi-objective problem into a series of single-objective sub-problems, each sub-problem corresponding to a weight vector. For example, the Chebyshev decomposition method can be used.

[0215]

[0216] Where w is the weight vector, z * For the ideal point.

[0217] In step S4, this invention uses a genetic algorithm to solve the subproblems of trajectory optimization and task sequence optimization. This process includes initializing the population, calculating the multi-objective fitness values ​​of individuals, performing non-dominated sorting, applying an elite selection strategy, SBX crossover, and PM mutation. Specifically, the non-dominated sorting uses a fast non-dominated sorting algorithm with a time complexity of O(MN^2), where M is the number of objectives and N is the population size.

[0218] Step S5 determines whether the termination condition is met. Common termination conditions include reaching the maximum number of iterations, reaching the preset computation time, and no significant improvement in the optimal solution over multiple generations. If the termination condition is not met, return to step S4 to continue iteration; otherwise, output the optimal solution in the non-dominated set as the planning result for the UAV trajectory and mission sequence.

[0219] In step S6, based on the planning results, the present invention generates detailed UAV flight paths using local path planning networks and global path planning networks. Local path planning employs an improved RRT* algorithm, while global path planning uses an improved A* algorithm. This multi-level path planning strategy can effectively handle complex mission environments.

[0220] In step S7, during task execution, the target recognition module detects and identifies targets in real time, and the target tracking module predicts the target trajectory and guides the drone. The target recognition module uses a deep learning model, such as YOLOv5 or Faster R-CNN, while the target tracking module uses a Kalman filter or particle filter.

[0221] In step S8, the present invention uses an algorithm optimization module to dynamically adjust the number of iterations and the mutation operator to improve the algorithm's efficiency and adaptability. For example, the number of iterations can be dynamically adjusted based on the convergence of the optimization process:

[0222]

[0223] Among them, T k f(x) represents the iteration number of the k-th iteration. k ) represents the objective function value of the k-th iteration, and α is the adjustment coefficient.

[0224] Finally, step S9 repeats steps S2 to S8 until all tasks are completed or a preset termination condition is met. This iterative optimization strategy enables the system to adapt to dynamically changing task environments and continuously improve the planning scheme.

[0225] Through the above steps, the method of this invention can efficiently handle complex multi-target UAV mission planning problems and generate optimized mission execution schemes. This method comprehensively considers UAV performance, mission characteristics, and environmental constraints, exhibiting strong practicality and adaptability. Furthermore, the numerous innovative technologies incorporated into the method, such as the MOEA / D framework, improved path planning algorithms, deep learning-based target recognition and tracking, and adaptive optimization strategies, give it significant advantages in both efficiency and effectiveness.

[0226] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-objective unmanned aerial vehicle mission planning method based on heuristic optimization algorithm, characterized in that, The method comprises the following steps: S1, obtaining the constraint and capability model of the unmanned aerial vehicle and the position, area, time sequence and priority model of each task; S2, establishing a multi-objective optimization function with the shortest flight path distance of the unmanned aerial vehicle and the shortest execution time of the unmanned aerial vehicle as the optimization objectives and the flight constraint of the unmanned aerial vehicle as the constraint condition; S3, constructing a task optimization model by using the MOEA / D framework and decomposing the multi-objective optimization function into two sub-problems of flight path optimization and task sequence optimization; S4, solving the flight path optimization and task sequence optimization sub-problems by using a genetic algorithm, wherein: initializing a population and calculating the multi-objective fitness value of each individual; performing non-dominated sorting to obtain a non-dominated set, an external set and remaining individuals; generating new individuals by using an elite selection strategy, SBX crossover and PM mutation on the remaining individuals; updating the non-dominated set and the external set; S5, judging whether a termination condition is met, and if not, returning to step S4, otherwise outputting the optimal solution in the non-dominated set as the planning result of the flight path and the task sequence of the unmanned aerial vehicle; S6, generating a detailed flight path of the unmanned aerial vehicle by using a local path planning network and a global path planning network based on the planning result; S7, detecting and identifying a target in real time by using a target identification module and predicting a target trajectory and guiding the unmanned aerial vehicle by using a target tracking module during task execution; S8, dynamically adjusting the iteration calculation times and the mutation operator by using an algorithm optimization module to improve the algorithm efficiency and adaptability; S9, repeating steps S2 to S8 until all tasks are completed or a preset termination condition is met.

2. A heuristic-driven UAV multi-objective mission planning and path planning system implementing the method of claim 1, characterized in that, The method comprises the following steps: A model module is configured to: decouple a task model into an unmanned aerial vehicle model, a task model and a battlefield environment model; transmit the decoupled model data to an optimization algorithm module; The optimization algorithm module is connected with the model module and configured to: receive the decoupled model data transmitted by the model module; determine a planning target and generate an optimization result based on the decoupled model data; feed back the optimization result to the model module; A path planning module is connected with the optimization algorithm module and configured to: receive the optimization result generated by the optimization algorithm module; generate a flight path of the unmanned aerial vehicle based on the optimization result; A target identification module is connected with the path planning module and configured to: receive real-time image data; detect and identify a target from the real-time image data based on a pre-trained target identification model; A target tracking module is connected with the target identification module and configured to: receive target information identified by the target identification module; calculate a target predicted trajectory based on a trajectory prediction model; An algorithm optimization module is connected with the optimization algorithm module and configured to: optimize the iteration calculation times and the mutation operator of the algorithm based on a historical data analysis result and a dynamic weight adaptive mutation operator.

3. The system of claim 2, wherein, The model module comprises: an unmanned aerial vehicle model unit configured to store and update constraint and capability data of the unmanned aerial vehicle; a task model unit configured to store and update position, area, time sequence and priority data of the task; a battlefield environment model unit configured to store and update task information, threat information, electromagnetic environment information and terrain data in a battlefield area. The model decoupling unit is connected with the unmanned aerial vehicle model unit, the task model unit and the battlefield environment model unit, and is configured to decouple the received unit data to generate decoupled model data.

4. The system of claim 2, wherein, The optimization algorithm module comprises: a multi-objective optimization unit configured to construct a multi-objective optimization problem based on an MOEA / D framework; a decomposition optimization unit connected with the multi-objective optimization unit and configured to decompose the multi-objective optimization problem into a flight path optimization sub-problem and a task sequence optimization sub-problem; a genetic algorithm unit connected with the decomposition optimization unit and configured to solve the flight path optimization sub-problem and the task sequence optimization sub-problem; a non-dominated sorting unit connected with the genetic algorithm unit and configured to perform non-dominated sorting on population individuals; an elitist strategy unit connected with the non-dominated sorting unit and configured to retain and update an elitist solution set.

5. The system of claim 2, wherein, The path planning module comprises: a local path planning network configured to perform local path planning on target points to calculate flight paths of the unmanned aerial vehicle to the target points; a global path planning network configured to establish flight paths of the unmanned aerial vehicle between the target points; a path optimization unit connected with the local path planning network and the global path planning network and configured to optimize planning results of the global path planning network based on output results of the local path planning network.

6. The system of claim 2, wherein, The target identification module comprises: an image preprocessing unit configured to preprocess received real-time image data; an attention model unit connected with the image preprocessing unit and configured to extract features from the preprocessed image; a target classification unit connected with the attention model unit and configured to perform target classification based on the extracted features; a target positioning unit connected with the target classification unit and configured to determine positions of the classified targets in the image.

7. The system of claim 2, wherein, The target tracking module comprises: a trajectory prediction network configured to predict future trajectories of the target based on historical target position data; a trajectory smoothing unit connected with the trajectory prediction network and configured to perform smoothing processing on the predicted trajectory; a target guiding unit connected with the trajectory smoothing unit and configured to generate unmanned aerial vehicle guiding instructions based on the smoothed predicted trajectory.

8. The system of claim 2, wherein, The algorithm optimization module comprises: an iteration optimization unit configured to determine an iteration calculation number based on a difference between a distance of the unmanned aerial vehicle from the target in each calculation and a current distance of the unmanned aerial vehicle from the target; a mutation operator optimization unit configured to calculate a difference between population individuals and last-generation population individuals and stability of the population individuals to generate an adaptive mutation operator; a parameter self-adaption unit connected with the iteration optimization unit and the mutation operator optimization unit and configured to dynamically adjust parameters in the MOEA / D framework.

9. The system of claim 4, wherein, The genetic algorithm unit comprises: an encoding unit configured to encode the unmanned aerial vehicle flight path and the task sequence into a chromosome; a crossover unit configured to perform SBX crossover operation on selected parent chromosomes; a mutation unit configured to perform PM mutation operation on the crossed child chromosomes; a fitness calculation unit configured to calculate multi-objective fitness values of the chromosomes; a selection unit configured to select excellent individuals into the next generation based on the fitness values.

10. The system of claim 2, wherein, Also included is a constraint processing module, in data connection with the optimization algorithm module, for: receiving a candidate solution generated by the optimization algorithm module; evaluating the feasibility of the candidate solution based on pre-set constraints; feeding back the evaluation result to the optimization algorithm module for guiding the next round of optimization.

Images