A method and system for planning a flight path of an unmanned target drone

Abstract

The application discloses a kind of unmanned target aircraft flight path planning method and system, comprising: obtaining dynamic battlefield environment data, generates initial path node sequence by A* algorithm in combination with dynamic battlefield threat data, calculates communication quality weight sequence using path loss model and Monte Carlo method and optimizes path.Reevaluate deviation through Kalman filtering fusion navigation data, correct path by quadratic programming, generate smooth path using B-spline curve fitting, and evaluate path execution probability by Monte Carlo method, finally generate path instruction sequence containing speed and direction control signal.The application realizes intelligent path planning in complex battlefield environment, improves the safety, reliability and execution efficiency of path.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of path planning, and particularly relates to a flight path planning method and system for an unmanned target aircraft. BACKGROUND

[0002] The flight path planning of an unmanned target aircraft is a core technology in the field of modern military training and testing, and directly affects the authenticity of combat simulation and training effect. Its importance is reflected in improving the task execution capability of the target aircraft in a complex environment and the system resource utilization efficiency. At present, traditional path planning methods mostly rely on static models or single optimization objectives, and are difficult to adapt to dynamic battlefield environments and multi-task requirements. There are generally problems such as insufficient real-time performance, poor path adaptability, and low resource allocation efficiency. These limitations make it difficult for the target aircraft to respond flexibly when facing unexpected threats or communication interruptions, limiting the complexity and realism of the training scene.

[0003] At the same time, the dynamic battlefield environment also puts forward higher requirements for path planning, and the situation awareness capability becomes the primary technical bottleneck. The target aircraft needs to obtain environmental data in real time through multiple sensors and perform threat analysis, but the existing system is insufficient in data processing speed and accuracy in complex electromagnetic environments and terrain conditions, and it is difficult to generate a real-time optimal path. As a result, the demand for communication quality optimization arises, because path selection not only needs to consider threat avoidance, but also needs to ensure stable command transmission and control response, and the unevenness of signal strength distribution makes path planning need to dynamically balance communication quality and task objectives. Further, path deviation management becomes a key restricting factor, and real-time deviation monitoring and correction need to balance between tactical flexibility and safety, and too large deviation may lead to task failure or safety accidents. Finally, the smoothness and executability of path fitting directly affect the dynamic performance of the target aircraft, and multi-segment spline curves need to be dynamically adjusted between physical constraints and task requirements, and the fitting accuracy of the existing method is insufficient in high-mobility scenarios.

[0004] Therefore, how to integrate situation awareness, communication optimization, deviation management, and path fitting technology to build an unmanned target aircraft system that can realize real-time adaptive path planning in a complex dynamic environment has become the primary key problem to be studied. SUMMARY

[0005] To solve the above technical problems, the application provides a flight path planning method and system for an unmanned target aircraft. Among them, a flight path planning method for an unmanned target aircraft comprises:

[0006] Obtaining dynamic battlefield environment data, and generating a first path node sequence by assigning dynamic weights using an A* algorithm according to threat region coordinates in the dynamic battlefield environment data;

[0007] According to the first path node sequence, the signal strength of each node is calculated through a path loss model, the signal attenuation probability is evaluated by using a Monte Carlo method, and a communication quality weight sequence is generated;

[0008] If there is a node in the communication quality weight sequence that is lower than a preset threshold, a genetic algorithm is used to optimize the first path node sequence, and a second path node sequence is generated;

[0009] According to the second path node sequence, the Euclidean distance deviation is calculated by fusing inertial navigation and global positioning data through Kalman filtering, and a deviation data sequence is generated;

[0010] If there is a deviation value in the deviation data sequence that exceeds a preset threshold, a quadratic programming algorithm is used to locally correct the corresponding path segment, and a third path node sequence is generated;

[0011] According to the third path node sequence, a B-spline curve fitting algorithm is used to optimize the control point position in combination with motion constraints, and a smooth path curve parameter is generated;

[0012] According to the smooth path curve parameter, the path execution probability is evaluated in a dynamic battlefield simulation by using a Monte Carlo method, and if the execution probability is higher than a preset threshold, a final path instruction sequence containing speed and direction control signals is generated.

[0013] Preferably, the process of obtaining dynamic battlefield environment data includes:

[0014] Dynamic battlefield environment data is collected through radar, infrared, and optical sensors to obtain a multi-source sensor data set;

[0015] A multi-sensor fusion algorithm is used to preprocess the multi-source sensor data set to obtain preliminary fusion data;

[0016] For the preliminary fusion data, a Kalman filtering algorithm is applied for optimization processing to obtain smooth target position, threat region coordinates, and terrain height information;

[0017] If the distance between the target position and the threat region coordinates is less than a preset threshold, a high-risk target is determined through coordinate information analysis to obtain a high-risk target list;

[0018] According to the high-risk target list and the terrain height information, the relative height of the target relative to the terrain is calculated to obtain target terrain association data;

[0019] Through the target terrain association data, a clustering algorithm is used to spatially group the targets to obtain a target distribution pattern;

[0020] According to the target distribution pattern, real-time spatial data of the dynamic battlefield environment is generated to obtain battlefield situation information.

[0021] Preferably, the process of generating the first path node sequence by assigning dynamic weights using the A* algorithm comprises:

[0022] Based on the dynamic battlefield environment data, threat area coordinates and threat level information are extracted, and a threat area division result is determined;

[0023] Through a linear mapping method, the dynamic weights of the threat areas are calculated based on the threat level information to obtain a weight distribution table; wherein the dynamic weights are determined based on linear mapping of the threat level;

[0024] If the dynamic weight exceeds a preset threshold, an initial first path node sequence is generated based on the weight distribution table using the A* algorithm, and an initial path planning is determined;

[0025] According to the initial path planning, the trend of environmental data change in the path node sequence is obtained, and it is determined whether the path passes through a high-threat area, to obtain an optimization demand assessment;

[0026] If the optimization demand assessment shows that the path passes through a high-threat area, the path is recalculated by the Dijkstra algorithm, the node sequence is adjusted based on the dynamic weight, and an adjusted first path node sequence is generated;

[0027] The environmental data of the adjusted first path node sequence is obtained, the threat area coordinates and the dynamic weight are integrated using data fusion technology, and a final path planning is determined;

[0028] Through the final path planning, coordinate data of the path node sequence is generated, and an optimized first path node sequence is output.

[0029] Preferably, the process of generating the communication quality weight sequence comprises:

[0030] By analyzing the first path node sequence, the position distribution of each node and the signal propagation distance are obtained, and the initial signal strength is determined;

[0031] Using a path loss model, the signal strength of each node is calculated in combination with environmental interference factors, and a signal strength sequence is obtained;

[0032] If the signal strength is lower than a preset threshold, a signal attenuation probability distribution is generated by simulating the signal attenuation probability using the Monte Carlo method;

[0033] According to the attenuation probability distribution, the communication quality weight of each node is calculated, and a communication quality weight sequence is obtained;

[0034] For the weight sequence generation result, the communication quality is evaluated, and it is determined whether there is a low-quality node;

[0035] If there is a low-quality node, adjust the node position distribution, and recalculate the signal strength sequence;

[0036] Generate the final communication quality weight sequence by iteratively optimizing the weight sequence.

[0037] Preferably, the process of generating the second path node sequence comprises:

[0038] If there is a node below the preset threshold in the communication quality weight sequence, obtain the node identification below the preset threshold from the communication quality weight sequence to obtain a set of nodes to be optimized;

[0039] According to the set of nodes to be optimized, initialize the population of the first path node sequence using a genetic algorithm to generate an initial path node sequence set;

[0040] For the initial path node sequence set, calculate the weighted sum of the communication quality weight and the path length of each path to obtain a set of fitness function values;

[0041] If there is a path below the preset fitness threshold in the set of fitness function values, generate a new path node sequence set through the crossover and mutation operations of the genetic algorithm;

[0042] According to the new path node sequence set, recalculate the weighted sum of the communication quality weight and the path length to obtain an updated set of fitness function values;

[0043] By comparing the updated set of fitness function values, select the path node sequence with the highest fitness value to generate the second path node sequence;

[0044] According to the second path node sequence, adjust the node connection relationship of the communication network to determine the optimized communication path configuration.

[0045] Preferably, the process of generating the deviation data sequence comprises:

[0046] Through the second path node sequence, obtain inertial navigation data and global positioning data, and perform preliminary fusion using Kalman filtering to obtain initial fusion data;

[0047] For the initial fusion data, calculate the Euclidean distance deviation between each node to generate a preliminary deviation data sequence;

[0048] If there is an outlier in the preliminary deviation data sequence, remove the outlier through a preset threshold judgment to obtain a corrected deviation data sequence;

[0049] According to the corrected deviation data sequence, optimize data smoothness using a secondary Kalman filter to generate an optimized deviation data sequence;

[0050] extracting deviation statistical features from the optimized deviation data sequence, calculating deviation mean and variance, and obtaining deviation distribution characteristics;

[0051] fitting a deviation trend by using a least square method based on the deviation distribution characteristics, and constructing a deviation trend model;

[0052] predicting a deviation sequence of a subsequent path node according to the deviation trend model, and obtaining a deviation data sequence by prediction.

[0053] Preferably, the process of generating the third path node sequence comprises:

[0054] If there is a deviation value exceeding a preset threshold in the deviation data sequence, an abnormal deviation point is determined by threshold judgment to obtain an abnormal point set;

[0055] analyzing the corresponding relationship between the abnormal point set and the path segment, obtaining a path segment to be optimized, and determining a set of segments to be optimized;

[0056] performing local optimization on the set of segments to be optimized by using a quadratic programming algorithm to generate a temporary path node sequence;

[0057] If the deviation value of the temporary path node sequence still exceeds the preset threshold, the quadratic programming algorithm parameters are adjusted by iteration to obtain an optimized path node sequence;

[0058] According to the optimized path node sequence, the continuity feature of the node sequence is obtained, and the sequence smoothness is judged;

[0059] According to the smoothness judgment result, the optimized path node sequence is smoothed by using a spline interpolation algorithm to generate a third path node sequence;

[0060] If the deviation value of the third path node sequence meets the preset threshold, a final path node sequence is generated by sequence integration.

[0061] Preferably, the process of generating the smooth path curve parameters comprises:

[0062] Based on the third path node sequence, the node coordinates and connection order are extracted by preprocessing to determine an initial path point set;

[0063] The initial positions of control points are calculated according to the initial path point set by using a B-spline curve fitting algorithm to obtain preliminary curve parameters;

[0064] If the preliminary curve parameters meet the motion constraint, the control point positions are retained;

[0065] If not, the control points are adjusted according to the constraint conditions to generate an optimized control point set;

[0066] Re-calculate the B-spline curve by the optimized control point set to obtain the smooth path curve parameter;

[0067] According to the smooth path curve parameter, analyze the path smoothness, if the smoothness is lower than the preset threshold, iteratively adjust the control point position to determine the final control point set;

[0068] According to the final control point set, generate a smooth path curve, output the curve parameter and the path point sequence;

[0069] Extract the path geometric feature from the output curve parameter, generate the path description data, and determine the final smooth path.

[0070] Preferably, the process of generating the final path instruction sequence containing the speed and direction control signal by the Monte Carlo method in the dynamic battlefield simulation if the execution probability is higher than the preset threshold value comprises:

[0071] Obtain the probability distribution of the smooth path curve from the dynamic battlefield simulation by the Monte Carlo method to obtain the execution probability value;

[0072] Analyze the comparison result of the execution probability value and the preset threshold value by using the probability distribution model to determine the path feasibility;

[0073] If the path feasibility is determined to be executable, extract the speed control signal from the smooth path curve to generate a preliminary speed instruction set;

[0074] Extract the direction control signal from the smooth path curve to generate a preliminary direction instruction set;

[0075] Merge the preliminary speed instruction set and the preliminary direction instruction set by the instruction sequence optimization algorithm to obtain an optimized path instruction sequence;

[0076] For the optimized path instruction sequence, obtain the battlefield environment parameter, and judge the adaptability of the instruction sequence;

[0077] If the adaptability of the instruction sequence is passed, output the final path instruction sequence.

[0078] The application also provides an unmanned target aircraft flight path planning system, comprising:

[0079] A data acquisition module is configured to acquire dynamic battlefield environment data;

[0080] A path planning module is configured to generate a first path node sequence by assigning a dynamic weight using an A* algorithm according to threat region coordinates in the dynamic battlefield environment data;

[0081] A communication evaluation module is configured to calculate signal strength of each node by a path loss model according to the first path node sequence, evaluate signal attenuation probability by a Monte Carlo method, and generate a communication quality weight sequence;

[0082] A path optimization module is configured to optimize the first path node sequence by a genetic algorithm if there is a node lower than a preset threshold in the communication quality weight sequence, and generate a second path node sequence;

[0083] A deviation calculation module is configured to calculate Euclidean distance deviation by Kalman filtering fusion of inertial navigation and global positioning data according to the second path node sequence, and generate a deviation data sequence;

[0084] A local correction module is configured to perform local correction on a corresponding path segment by a quadratic programming algorithm if there is a deviation value exceeding a preset threshold in the deviation data sequence, and generate a third path node sequence;

[0085] A path smoothing module is configured to optimize control point positions by a B-spline curve fitting algorithm combined with motion constraints according to the third path node sequence, and generate a smooth path curve parameter;

[0086] A path execution module is configured to evaluate path execution probability in a dynamic battlefield simulation by a Monte Carlo method according to the smooth path curve parameter, and generate a final path instruction sequence containing speed and direction control signals if the execution probability is higher than a preset threshold.

[0087] Compared with the prior art, the present application has the following advantages and technical effects:

[0088] The application discloses an intelligent path planning method in a dynamic battlefield environment, battlefield data is obtained by multi-sensor fusion, an initial path is generated by an A* algorithm, genetic algorithm optimization is performed in combination with communication quality evaluation, deviation is corrected by Kalman filtering and quadratic programming, and finally a smooth path is generated by B-spline curve fitting. The method comprehensively considers factors such as threat avoidance, communication quality, navigation accuracy and the like, realizes intelligent path planning in a complex battlefield environment through multi-stage optimization, and improves the safety, reliability and execution efficiency of the path.

[0089] The application can provide effective decision support for autonomous navigation of an unmanned system in a dynamic confrontation environment, and has important military application value.

[0090] The application realizes intelligent and adaptive path planning in a dynamic battlefield environment, and provides strong support for autonomous movement of equipment such as unmanned systems. BRIEF DESCRIPTION OF DRAWINGS

[0091] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The embodiments of this application, and of the prior art, can be best understood from the accompanying drawings, wherein:

[0092] Figure 1 A flow chart of a method according to an embodiment of the application;

[0093] Figure 2 A schematic diagram of a system according to an embodiment of the application. DETAILED DESCRIPTION

[0094] It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0095] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described can be executed in an order different from that shown here.

[0096] Embodiment one

[0097] As shown in the drawings, Figure 1 A flight path planning method for unmanned target aircraft is provided in the embodiment, comprising:

[0098] Obtaining dynamic battlefield environment data, and using A* algorithm to assign dynamic weights to generate a first path node sequence according to threat region coordinates in the dynamic battlefield environment data;

[0099] According to the first path node sequence, calculating the signal strength of each node through a path loss model, and using a Monte Carlo method to evaluate the signal attenuation probability to generate a communication quality weight sequence;

[0100] If there is a node in the communication quality weight sequence that is lower than a preset threshold, then using a genetic algorithm to optimize the first path node sequence to generate a second path node sequence;

[0101] According to the second path node sequence, calculating the Euclidean distance deviation through Kalman filtering fusion of inertial navigation and global positioning data to generate a deviation data sequence;

[0102] If there is a deviation value in the deviation data sequence that exceeds a preset threshold, then using a quadratic programming algorithm to locally correct the corresponding path segment to generate a third path node sequence;

[0103] According to the third path node sequence, using a B-spline curve fitting algorithm to optimize the control point position in combination with motion constraints to generate a smooth path curve parameter;

[0104] According to the smooth path curve parameters, the path execution probability is evaluated in the dynamic battlefield simulation through the Monte Carlo method, and if the execution probability is higher than a preset threshold, a final path instruction sequence containing speed and direction control signals is generated.

[0105] Further, the process of acquiring dynamic battlefield environment data includes:

[0106] The dynamic battlefield environment data is collected through radar, infrared and optical sensors to obtain a multi-source sensor data set; a multi-sensor fusion algorithm is used to preprocess the multi-source sensor data set to obtain preliminary fusion data.

[0107] The Kalman filtering algorithm is applied to the preliminary fusion data for optimization processing to obtain smooth target position, threat region coordinates and terrain height information.

[0108] If the distance between the target position and the threat region coordinates is less than a preset threshold, a high-risk target is determined through coordinate information analysis to obtain a high-risk target list.

[0109] According to the high-risk target list and the terrain height information, the relative height of the target relative to the terrain is calculated to obtain target terrain correlation data.

[0110] Through the target terrain correlation data, a clustering algorithm is used to group the targets in space to obtain a target distribution pattern; according to the target distribution pattern, real-time spatial data of the dynamic battlefield environment is generated to obtain battlefield situation information.

[0111] Specifically, in a dynamic battlefield environment, multi-source sensor data fusion is the core technology to achieve situation awareness. Radar, infrared and optical sensors capture the distance, speed, thermal characteristics and image information of the target respectively to form a multi-source data set. For example, radar can provide speed information of the target within 1000 meters, infrared sensor detects the target thermal signal strength, and optical sensor generates a two-dimensional image of the target. In the preprocessing stage, a multi-sensor fusion algorithm such as weighted average or Bayesian inference is used to integrate these data into a set of preliminary fusion data.

[0112] Specifically, the target coordinates measured by the radar are (500, 300, 50), the infrared confirms the thermal signal, and the unified target position data is generated after fusion. This fusion can improve data consistency and reduce the influence of single sensor noise. For the preliminary fusion data, the Kalman filtering algorithm is used for optimization processing to generate smooth target position, threat region coordinates and terrain height information.

[0113] For example, the initial position of a target is (510, 305, 55), which is corrected to (508, 302, 53) through Kalman filtering combined with historical trajectory and sensor noise characteristics, while the threat area coordinates (600, 400, 60) and the terrain height of 50 meters are generated.

[0114] Preferably, this method can effectively filter out random errors and improve target tracking accuracy to provide reliable data for subsequent analysis.

[0115] In one possible implementation, the distance between the target position and the threat area coordinates is calculated, and if it is less than a preset threshold, such as 200 meters, it is determined as a high-risk target.

[0116] For example, the distance between the target position (508, 302, 53) and the threat area (600, 400, 60) is about 141 meters, which is less than the threshold, and is included in the high-risk target list.

[0117] It can be understood that this determination can quickly screen potential threats to assist command decision-making. According to the high-risk target list and the terrain height information, the relative height of the target relative to the terrain is calculated to form target terrain correlation data.

[0118] In one embodiment, the target height is 53 meters, the terrain height is 50 meters, and the relative height is 3 meters, indicating that the target may be a low-altitude aircraft. This correlation data can reveal the activity characteristics of the target and improve the understanding of the battlefield situation. Through target terrain correlation data, a clustering algorithm such as K-means is used to spatially group targets to generate target distribution patterns.

[0119] Specifically, 10 high-risk targets are divided into 3 groups according to their positions and relative heights, which are located in the northwest, southeast, and central regions of the battlefield, respectively.

[0120] It should be noted that this grouping can reveal the enemy's deployment rules and facilitate the prediction of their intentions.

[0121] In one embodiment, according to the target distribution pattern, real-time spatial data of the dynamic battlefield environment is generated to form battlefield situation information. For example, the target in the northwest region is dense and has a low relative height, which may be a ground troop concentration; the target in the southeast region is sparse and has a high height, which may be a reconnaissance drone. This situation information can provide commanders with an intuitive spatial situation map to support tactical adjustments.

[0122] Preferably, the above technical process significantly improves the accuracy and real-time performance of battlefield situation awareness through multi-source data fusion, optimization processing, and pattern analysis to support rapid decision-making. Each technical link supports each other to form an efficient link from data acquisition to situation generation, ensuring the comprehensiveness and reliability of battlefield information.

[0123] Further, the process of assigning dynamic weights by using the A* algorithm to generate the first path node sequence comprises:

[0124] Based on the dynamic battlefield environment data, threat region coordinates and threat level information are extracted, and a threat region division result is determined.

[0125] Through a linear mapping method, the dynamic weights of the threat regions are calculated according to the threat level information, and a weight distribution table is obtained. The dynamic weights are determined based on linear mapping of the threat level. If the dynamic weights exceed a preset threshold, the A* algorithm is used to generate an initial first path node sequence based on the weight distribution table, and an initial path planning is determined.

[0126] According to the initial path planning, the trend of the environmental data in the path node sequence is obtained, and it is determined whether the path passes through a high-threat region, and an optimization demand assessment is obtained.

[0127] If the optimization demand assessment shows that the path passes through a high-threat region, the Dijkstra algorithm is used to recalculate the path, the node sequence is adjusted based on the dynamic weights, and an adjusted first path node sequence is generated.

[0128] The environmental data of the adjusted first path node sequence is obtained, the threat region coordinates and the dynamic weights are integrated by using data fusion technology, and a final path planning is determined.

[0129] Through the final path planning, coordinate data of the path node sequence is generated, and an optimized first path node sequence is output.

[0130] Specifically, in a dynamic battlefield environment, obtaining threat region coordinates and threat level information is the core of building battlefield situation awareness.

[0131] For example, assuming that through multi-sensor data fusion, it is determined that the center coordinates of a certain region are (2000, 3000), and the threat level is high. Based on intelligence analysis and sensor detection, the threat region is divided into a circular region with a center at the coordinates and a radius of 800 meters. The threat level information is usually derived from the enemy target density detected by radar, the heat signal strength captured by infrared sensors, and the equipment type identified by optical sensors.

[0132] Specifically, a high threat level may correspond to the intensive deployment of enemy heavy equipment, and the heat signal shows continuous activity. Through these data, the command system can clearly divide the threat region and provide a basis for subsequent path planning.

[0133] Preferably, when calculating the dynamic weights based on the threat level, a linear mapping method is used.

[0134] For example, the threat level is divided into three levels: low, medium, and high, which are mapped to weight values of 0.3, 0.6, and 0.9, respectively. Assuming that the threat level of the (2000, 3000) region is high, with a weight of 0.9, and the threat level of another region (2500, 3500) is medium, with a weight of 0.6. The weight distribution table records the coordinates of each region and its weight value, forming a dynamic weight distribution. This method is simple and efficient, and can intuitively reflect the differences in threat levels, providing a quantitative basis for path planning.

[0135] In one possible implementation, if the dynamic weight of a certain region exceeds the threshold value 0.8, the A* algorithm is triggered to generate an initial path.

[0136] For example, planning a path from the starting point (1000, 1000) to the end point (3000, 3000), the A* algorithm prioritizes nodes with lower weights based on the weight distribution table, generating a node sequence such as (1000, 1000), (1500, 1500), (2000, 2000), (3000, 3000).

[0137] It should be noted that the A* algorithm quickly generates a feasible path by heuristic search, considering both path length and threat weight.

[0138] Specifically, when analyzing the trend of environmental data changes in the initial path, the threat level and terrain information of each point in the node sequence are extracted. For example, the node (2000, 2000) is close to a high-threat region, with a threat weight of 0.85, which exceeds the safety threshold, indicating that the path may pass through a high-threat region, triggering the need for optimization. Optimization evaluation is based on data fusion, integrating threat level, terrain height, and enemy activity intelligence to confirm that the path needs to be adjusted.

[0139] In one embodiment, the Dijkstra algorithm is used to replan the path, adjusting the node sequence based on the dynamic weight. For example, the Dijkstra algorithm aims to minimize the cumulative weight, avoiding high-threat regions, and generates a second path node sequence such as (1000, 1000), (1200, 1300), (1800, 1800), (3000, 3000). This method ensures higher path safety through global search. The final path planning integrates threat region coordinates and dynamic weights, outputting optimized node sequence coordinate data, providing reliable support for command decision-making.

[0140] It can be understood that the advantage of the above method is that it dynamically adapts to changes in the battlefield environment through multi-level data processing and algorithm optimization. Threat region division and weight calculation provide accurate situational awareness, and the combination of A* and Dijkstra algorithms takes into account efficiency and safety, generating a path node sequence that effectively avoids high-threat regions and improves the success rate of task execution.

[0141] Further, the process of generating the communication quality weight sequence comprises:

[0142] By analyzing the first path node sequence, the position distribution of each node and the signal propagation distance are obtained, and the initial signal strength is determined;

[0143] Using the path loss model, combined with the environmental interference factor, the signal strength of each node is calculated to obtain the signal strength sequence;

[0144] If the signal strength is lower than the preset threshold, the signal attenuation probability distribution is generated by simulating the signal attenuation probability through the Monte Carlo method;

[0145] According to the attenuation probability distribution, the communication quality weight of each node is calculated to obtain the communication quality weight sequence;

[0146] For the weight sequence generation result, the communication quality is evaluated to determine whether there is a low-quality node;

[0147] If there is a low-quality node, the node position distribution is adjusted, and the signal strength sequence is recalculated;

[0148] The final communication quality weight sequence is generated by iteratively optimizing the weight sequence.

[0149] Specifically, in a dynamic battlefield environment, path planning and signal optimization of a communication system are crucial.

[0150] Illustratively, the analysis of the path node sequence can obtain the position distribution and signal propagation distance by analyzing the spatial coordinates of each node. Assuming that in a certain battlefield scenario, the node sequence is A(10, 10), B(30, 30), C(50, 50), D(70, 70), and E(90, 90), the Euclidean distance between adjacent nodes is calculated to obtain the propagation distance of each segment, such as 28.28 units for the AB segment, 28.28 units for the BC segment, etc. These distances provide a basis for subsequent signal strength calculation.

[0151] Specifically, the determination of the initial signal strength is based on the relationship between the transmission power and the distance.

[0152] In one embodiment, assuming that the transmission power is 100 mW, combined with the free space path loss model, the signal strength of node A can be set as a reference value, and the remaining nodes are attenuated according to the distance.

[0153] For example, the signal strength of node B is slightly attenuated due to the propagation distance of 28.28 units, estimated at 90 mW.

[0154] It should be noted that environmental interference factors such as terrain and electromagnetic noise need to be considered.

[0155] Preferably, an interference coefficient can be introduced, such as a plain terrain interference coefficient of 0.8, and the actual signal strength of the Node B is adjusted to 90 mW x 0.8 = 72 mW. By a similar method, the signal strength sequence is calculated: A (100 mW), B (72 mW), C (65 mW), D (60 mW), E (55 mW). For example, if the preset signal strength threshold is 60 mW, then the 55 mW of Node E is lower than the threshold, and further analysis of its communication quality is required.

[0156] In one possible implementation, the Monte Carlo method is used to simulate the signal attenuation probability. By multiple random sampling, the attenuation of Node E under different interference conditions is simulated to generate an attenuation probability distribution. Assuming that in 1000 simulations, 500 times show that the attenuation exceeds 50%, then the attenuation probability of Node E is 50%. According to the distribution, the communication quality weight is calculated.

[0157] For example, the weight of Node E can be set to 1-0.5 = 0.5, and the weights of the remaining nodes are generated based on similar logic to obtain the weight sequence: A (0.9), B (0.8), C (0.7), D (0.6), E (0.5).

[0158] It can be understood that the weight sequence is used to evaluate the communication quality. The low weight of Node E indicates that it is a low-quality node and needs to be optimized.

[0159] In one embodiment, the position of Node E is adjusted to (85, 85), the distance to D is shortened, and the signal strength is recalculated to be 62 mW, meeting the threshold requirement. After iterative optimization, the weight sequence is updated to A (0.9), B (0.8), C (0.7), D (0.6), E (0.7). This method ensures the stability and reliability of the communication link through dynamic adjustment.

[0160] Preferably, the above scheme can be extended to multi-node cooperative optimization. For example, a backup node F (80, 80) is added to share the communication pressure of E and further improve the link quality. This logic from core optimization to extended cooperation ensures the integrity and diversity of the scheme. The above method improves the robustness of battlefield communication through refined signal management.

[0161] Further, the process of generating the second path node sequence comprises:

[0162] If there is a node in the communication quality weight sequence that is lower than the preset threshold, the node identifier that is lower than the preset threshold is obtained from the communication quality weight sequence to obtain a set of nodes to be optimized;

[0163] According to the set of nodes to be optimized, a genetic algorithm is used to initialize the population of the first path node sequence to generate an initial path node sequence set;

[0164] For the initial path node sequence set, the weighted sum of the communication quality weight and the path length of each path is calculated to obtain a fitness function value set;

[0165] If there is a path with a fitness function value lower than a preset fitness threshold in the fitness function value set, a new path node sequence set is generated through the crossover and mutation operations of the genetic algorithm;

[0166] According to the new path node sequence set, the weighted sum of the communication quality weight and the path length is recalculated to obtain an updated fitness function value set;

[0167] By comparing the updated fitness function value set, the path node sequence with the highest fitness value is selected to generate a second path node sequence;

[0168] According to the second path node sequence, the node connection relationship of the communication network is adjusted to determine the optimized communication path configuration.

[0169] By way of example, the communication quality weight sequence reflects the communication reliability of each node in the network, and the preset threshold is used to filter out nodes with poor communication quality. Assuming that the communication quality weight threshold is 0.6, and the weights of nodes A, B, C, and D in the sequence are 1.0, 0.8, 0.5, and 0.3 respectively, then nodes C and D are below the threshold and constitute the set of nodes to be optimized. The key to this step is to accurately identify the nodes that need to be optimized to ensure that the subsequent optimization is highly targeted.

[0170] In one possible implementation, the genetic algorithm is used to initialize the population of path node sequences. For the nodes to be optimized C and D, an initial set containing multiple paths is generated, such as path 1 A-B-C-D, path 2 A-C-D, and path 3 A-B-D. When the population is initialized, multiple node connection methods are randomly generated to maintain path diversity. The advantage of the genetic algorithm is that it gradually filters out better path configurations by simulating biological evolution.

[0171] Specifically, the fitness function value of the initial path node sequence set needs to consider both the communication quality weight and the path length. The communication quality weight reflects signal stability, while the path length affects transmission delay. Assuming that the total length of path A-B-C-D is 450 meters and the weight sum is 2.6, and the length of path A-C-D is 300 meters and the weight sum is 1.8. The weighted sum can be calculated by linear combination of weight and length, with higher weight proportion to prioritize communication quality. Paths with low fitness values, such as A-B-C-D, are marked as to be optimized due to low weight.

[0172] Preferably, the new path set is generated by crossover and mutation operations of the genetic algorithm. The crossover operation exchanges part of the nodes of two paths, for example, A-B-C-D and A-C-D are exchanged to generate A-B-D and A-C-B-D. The mutation operation randomly adjusts the order of the nodes, for example, A-B-D is mutated to A-D-B. The new set retains the advantages of the original path while introducing new possibilities.

[0173] It should be noted that the crossover and mutation need to be controlled in proportion to avoid excessive randomization leading to slow convergence.

[0174] In an embodiment, the fitness values of the new path set are recalculated.

[0175] For example, the weight sum of A-B-D is 2.3 and the length is 350 meters, and its fitness value is higher than that of A-C-B-D, which is 1.5 and 400 meters. By comparison, A-B-D with the highest fitness value is selected as the second path node sequence. This step ensures that the path takes into account both communication quality and transmission efficiency.

[0176] It can be understood that adjusting the network connection relationship according to the second path node sequence requires reconfiguring the communication link between nodes. For example, node B and D are directly connected, bypassing C, reducing signal attenuation. The optimized path configuration improves the overall reliability of the network, especially in high-load scenarios, effectively reducing the data packet loss rate.

[0177] For example, the optimized path A-B-D has improved signal transmission stability in actual deployment, and the communication quality weight of node D has increased from 0.3 to 0.7. This improvement is due to the global search capability of the genetic algorithm for paths and the comprehensive consideration of communication quality and path length. Through multiple iterations, the network configuration gradually tends to be optimal, significantly improving communication efficiency.

[0178] Further, the process of generating the bias data sequence includes:

[0179] Through the second path node sequence, inertial navigation data and global positioning data are obtained, and Kalman filtering is used for preliminary fusion to obtain initial fusion data; for the initial fusion data, the Euclidean distance bias between each node is calculated to generate a preliminary bias data sequence;

[0180] If there are outliers in the preliminary bias data sequence, the outliers are removed by a pre-set threshold to obtain a corrected bias data sequence; according to the corrected bias data sequence, a quadratic Kalman filter is used to optimize data smoothness to generate an optimized bias data sequence;

[0181] The deviation statistical features are extracted from the optimized deviation data sequence, the deviation mean and variance are calculated, and the deviation distribution characteristics are obtained; the least square method is used to fit the deviation trend through the deviation distribution characteristics, and a deviation trend model is constructed; and the deviation sequence of the subsequent path nodes is predicted according to the deviation trend model, and the deviation data sequence is obtained by prediction.

[0182] In one possible implementation, the process of acquiring inertial navigation data and global positioning data generally involves sensor data collection. An inertial navigation system records the motion trajectory of a device through an accelerometer and a gyroscope, providing high-frequency but easily drifting position data.

[0183] For example, a certain unmanned aerial vehicle cruises along a path A-F-C-G-E, and the inertial navigation system collects 100 acceleration and angular velocity data per second. The global positioning system provides low-frequency but high-precision absolute position through satellite signals.

[0184] Exemplarily, the unmanned aerial vehicle acquires global positioning coordinates every 5 seconds, and the precision is about 2 meters. After alignment through timestamps, the two are input into a Kalman filter for preliminary fusion. The Kalman filter balances the error characteristics of the two types of data by using state prediction and observation update, to generate initial fusion data.

[0185] Preferably, the fused data can reflect the position of the unmanned aerial vehicle at a relatively high frequency while maintaining good precision.

[0186] Specifically, when calculating the Euclidean distance deviation based on the initial fusion data, the geometric distance between the fusion positions of adjacent nodes and the actual path needs to be compared.

[0187] For example, the fusion position of the unmanned aerial vehicle from node A to F shows a distance of 500 meters, while the actual path is 510 meters, and the deviation is 10 meters. The deviations of A-F, F-C, C-G, and G-E are calculated in turn to form a preliminary deviation data sequence, such as [10, 8, 15, 12]. If there is an abnormal value in the sequence, for example, the deviation between C-G is 15 meters, which is much higher than the average value, a threshold value of the average value plus 2 times the standard deviation can be set, and the modified deviation data sequence [10, 8, 12] is obtained after removing the abnormal value. This removal method can effectively reduce the influence of sensor noise or environmental interference.

[0188] In one embodiment, a secondary Kalman filter is used to optimize the smoothness of the modified deviation data sequence. Compared with the initial fusion, the secondary filter focuses on the dynamic characteristics of the deviation data, and reduces random fluctuations.

[0189] For example, the filtered deviation sequence may become [9.8, 8.2, 11.9], and the data is smoother, which is suitable for subsequent analysis.

[0190] It should be noted that the smoothness improvement helps to accurately extract the deviation statistical features. The deviation mean is about 9.97 meters, and the variance is about 2.25, reflecting the concentration and dispersion of the deviation. These features provide a quantitative basis for the deviation distribution characteristics.

[0191] For example, the deviation trend fitting adopts the least squares method, and a linear or polynomial model is constructed based on the optimized deviation data sequence. Assuming that the fitting result shows that the deviation gradually increases to 12 meters from A to E. This trend may be related to the cumulative error of the unmanned aerial vehicle navigation system. The deviation trend model provides a basis for predicting the deviation of subsequent nodes.

[0192] Preferably, the predicted deviation data sequence can be generated by extrapolation of the model.

[0193] For example, the predicted deviation of the subsequent node E to the new node H is about 12.5 meters. This prediction result can be used to adjust the navigation parameters in advance and reduce the positioning error.

[0194] It can be understood that the above method gradually optimizes the reliability of navigation data through multi-level data processing. The implementation of each technical subject supports each other, forming a complete chain from data acquisition to deviation prediction.

[0195] For example, the initial fusion ensures data accuracy, the secondary filtering improves smoothness, and the trend fitting and prediction provide a basis for dynamic adjustment. This progressive logic can effectively improve the robustness of path planning, especially in complex environments.

[0196] Further, the process of generating the third path node sequence includes:

[0197] If there is a deviation value in the deviation data sequence that exceeds the preset threshold, an abnormal deviation point is determined through threshold judgment to obtain an abnormal point set;

[0198] The corresponding relationship between the abnormal point set and the path segment is analyzed to obtain a to-be-optimized path segment, and a to-be-optimized segment set is determined;

[0199] The to-be-optimized segment set is locally optimized using a quadratic programming algorithm to generate a temporary path node sequence;

[0200] If the deviation value of the temporary path node sequence and the original path segment still exceeds the preset threshold, the quadratic programming algorithm parameters are adjusted through iteration to obtain an optimized path node sequence;

[0201] According to the optimized path node sequence, the continuity feature of the node sequence is obtained, and the smoothness is judged;

[0202] According to the smoothness judgment result, the optimized path node sequence is smoothed using a spline interpolation algorithm to generate a third path node sequence;

[0203] If the deviation value of the third path node sequence meets the preset threshold value, the final path node sequence is generated through sequence integration.

[0204] For example, the processing of the deviation data sequence needs to identify abnormal deviation points first. Assuming that the deviation data sequence of a certain path node is [0.2, 0.3, 0.9, 0.4, 0.7]m, and the preset threshold value is 0.5m. Through threshold judgment, the deviation values 0.9m and 0.7m exceed the threshold value and are marked as abnormal points, forming an abnormal point set. When analyzing the corresponding relationship between the abnormal points and the path segments, the abnormal points can be associated through timestamps or node numbers to determine the path segment where the abnormal points are located.

[0205] For example, the deviation 0.9m corresponds to the 3rd node, which is located in the path segment S1; the deviation 0.7m corresponds to the 5th node, which is located in the path segment S2, thereby obtaining the set of segments to be optimized {S1, S2}.

[0206] In one possible implementation, a quadratic programming algorithm is used to locally optimize S1 and S2. Quadratic programming adjusts the node coordinates by setting an objective function to minimize the path deviation while maintaining the continuity of the path. Assuming that S1 contains node coordinates [100, 50] and [102, 52], the temporary path node sequence is generated after optimization and is adjusted to [101, 51], etc. If the temporary path deviation still exceeds the threshold value, the weight parameters of the quadratic programming can be iteratively adjusted, for example, the smoothness weight is increased, to generate an optimized path node sequence, such as [100.8, 50.9], etc. This process ensures that the path is closer to the reference trajectory.

[0207] Specifically, the continuity feature of the optimized path node sequence is judged by calculating the distance and angle change between nodes.

[0208] For example, the distance between nodes is stable within 2m, and the angle change is less than 15 degrees, indicating that the sequence continuity is good. If the continuity is insufficient, it can be smoothed by a spline interpolation algorithm. Spline interpolation fits a cubic polynomial to generate a smooth third path node sequence.

[0209] For example, after interpolation, the original node sequence [100, 50], [102, 52] adds an intermediate node [101, 51], and the path is smoother.

[0210] Preferably, the third path node sequence needs to be verified again whether the deviation meets the threshold value. Assuming that the deviation after verification is [0.3, 0.4, 0.2], which is lower than 0.5m, the final path node sequence is generated through sequence integration. When integrating, adjacent nodes can be merged to reduce redundancy, for example, [100.8, 50.9] and [101, 51] close to each other are merged into [100.9, 51]. The final path node sequence is more efficient and suitable for use by navigation equipment.

[0211] It should be noted that the above method forms a complete optimization process through abnormal point identification, path segment optimization and smoothing processing. Each step supports each other to ensure the accuracy and continuity of the path.

[0212] For example, abnormal point identification provides accurate targets for subsequent optimization, and quadratic programming and spline interpolation improve path quality from different aspects. This multi-step collaborative optimization method can effectively improve the practicality of the navigation path.

[0213] Further, the process of generating smooth path curve parameters includes:

[0214] Based on the third path node sequence, the node coordinates and connection order are extracted through preprocessing to determine the initial path point set;

[0215] The B-spline curve fitting algorithm is used to calculate the initial position of the control point based on the initial path point set to obtain the preliminary curve parameters;

[0216] If the preliminary curve parameters meet the motion constraints, the control point position is retained;

[0217] If not, adjust the control point according to the constraint condition to generate an optimized control point set;

[0218] By optimizing the control point set, the B-spline curve is recalculated to obtain the smooth path curve parameters;

[0219] According to the smooth path curve parameters, analyze the path smoothness, if the smoothness is lower than the preset threshold, iteratively adjust the control point position to determine the final control point set;

[0220] According to the final control point set, generate a smooth path curve, output the curve parameters and path point sequence;

[0221] Extract the path geometric features from the output curve parameters to generate path description data and determine the final smooth path.

[0222] Specifically, in the field of path planning, generating a smooth path is the key to ensuring efficiency and accuracy. To obtain the third path node sequence, first, the node coordinates and connection order need to be extracted through preprocessing.

[0223] For example, the initial path point set may contain a series of node coordinates, such as (1, 2), (2, 3), (4, 4). These nodes are connected by timestamps or path order to form a preliminary path. During preprocessing, the system analyzes the distance and direction change between nodes to ensure the continuity of the point set.

[0224] Preferably, if the distance between two nodes is too large, such as more than 2 units, the system will insert an intermediate node to improve the resolution of the path. When using the B-spline curve fitting algorithm, the initial positions of the control points are calculated based on the initial set of path nodes.

[0225] It can be understood that the advantage of the B-spline curve lies in its local adjustment capability, and the movement of a single control point only affects the local area of the curve.

[0226] For example, assuming that the initial set of path nodes contains 5 nodes, the system will generate a set of control points, such as (0.8, 1.8), (2.2, 3.1), (4.1, 4.2), to preliminarily form a curve. If the parameters of the preliminary curve meet the motion constraints, such as the radius of curvature is not less than 0.5, the positions of the control points are retained. Otherwise, the positions of the control points need to be adjusted.

[0227] For example, if the curvature of a certain segment is too small, the system will offset the control points outward through an optimization algorithm, such as adjusting (2.2, 3.1) to (2.4, 3.3), to generate an optimized set of control points.

[0228] In one possible implementation, the B-spline curve is recalculated by optimizing the set of control points to obtain the parameters of the smooth path curve. The system will analyze the smoothness of the path.

[0229] Specifically, the smoothness can be evaluated by the rate of change of curvature. If the rate of change exceeds a preset threshold, such as 0.3, the system will iteratively adjust the control points.

[0230] For example, a control point can be fine-tuned from (4.1, 4.2) to (4.0, 4.3) until the smoothness meets the standard. The final set of control points generates a smooth path curve, and the output curve parameters and path node sequence are, for example, (1, 2), (1.5, 2.5), (2, 3), (4, 4). For the extraction of path geometric features, the system will analyze the length, curvature, and direction change of the curve.

[0231] For example, the total length of the path can be 10 units, and the maximum curvature appears near (2, 3). When generating path description data, the system will record these features to form structured data, which facilitates subsequent path optimization or navigation applications.

[0232] It should be noted that the completeness of the path description data directly affects the planning efficiency.

[0233] For example, a navigation system can adjust the speed in advance based on the curvature data to ensure smooth operation.

[0234] In one embodiment, the generation process of the smooth path embodies high flexibility and adaptability.

[0235] Exemplarily, if the initial path point set needs to be temporarily adjusted due to environmental changes, such as the addition of obstacles causing path point deviation, the B-spline algorithm can quickly recalculate the control points to generate a new path. The advantage of this method is that it does not need to reconstruct the entire path, but only needs to make local adjustments to meet the constraint conditions.

[0236] Preferably, the system can also iteratively optimize according to real-time feedback to ensure that the path always meets dynamic requirements.

[0237] Specifically, the above process forms a complete link from the initial point set to the final smooth path through multi-level analysis and adjustment. The implementation of each technical subject focuses on path smoothness and constraint satisfaction, with logical rigor and mutual support.

[0238] For example, control point adjustment and smoothness analysis complement each other to ensure path quality. The final output of path description data provides a reliable foundation for subsequent applications, demonstrating the practicality and efficiency of the technology.

[0239] Further, the process of generating a final path instruction sequence including speed and direction control signals if the execution probability is higher than the preset threshold value in the dynamic battlefield simulation through the Monte Carlo method to evaluate the path execution probability includes:

[0240] Obtain the execution probability value by the Monte Carlo method to obtain the probability distribution of the smooth path curve from the dynamic battlefield simulation;

[0241] Determine the path feasibility by analyzing the comparison result of the execution probability value and the preset threshold value using the probability distribution model;

[0242] If the path feasibility is determined to be executable, extract the speed control signal from the smooth path curve to generate a preliminary speed instruction set;

[0243] Extract the direction control signal from the smooth path curve to generate a preliminary direction instruction set;

[0244] Merge the preliminary speed instruction set and the preliminary direction instruction set to obtain an optimized path instruction sequence through an instruction sequence optimization algorithm;

[0245] For the optimized path instruction sequence, obtain battlefield environment parameters to determine the adaptability of the instruction sequence;

[0246] If the instruction sequence adaptability passes the judgment, output the final path instruction sequence.

[0247] More specifically, when obtaining the execution probability value by the Monte Carlo method to obtain the probability distribution of the smooth path curve from the dynamic battlefield simulation, the dynamic battlefield simulation is constructed based on battlefield environment parameters, the Monte Carlo method performs random sampling through simulation iteration times to generate a probability distribution model, and the execution probability value of the smooth path curve is calculated;

[0248] When determining the path feasibility by comparing the execution probability value with the preset threshold value using the probability distribution model, the probability distribution model generates a statistical distribution according to the execution probability value, and if the execution probability value is higher than the preset threshold value, it is determined that the path is feasible, and a feasibility result is output;

[0249] If the path feasibility is determined to be executable, when extracting the speed control signal from the smoothed path curve to generate a preliminary speed instruction set, the speed-related parameters are analyzed from the smoothed path curve, and the preliminary speed instruction set containing the time stamp and the speed value is generated through the mapping function;

[0250] When extracting the direction control signal from the smoothed path curve to generate a preliminary direction instruction set, the direction-related parameters are analyzed from the smoothed path curve, and the preliminary direction instruction set containing the time stamp and the direction angle is generated through vector transformation;

[0251] The preliminary speed instruction set and the preliminary direction instruction set are merged through an instruction sequence optimization algorithm to obtain an optimized path instruction sequence; the instruction sequence optimization algorithm uses a dynamic programming algorithm to fuse the speed instruction set and the direction instruction set to generate a time-synchronized optimized path instruction sequence;

[0252] For the optimized path instruction sequence, battlefield environment parameters are obtained to determine the adaptability of the instruction sequence, including:

[0253] The battlefield environment parameters are extracted from the dynamic battlefield simulation, and if the optimized path instruction sequence matches the environment parameters, it is determined that the instruction sequence adaptability is passed;

[0254] If the instruction sequence adaptability is passed, the final path instruction sequence is output;

[0255] The optimized path instruction sequence is formatted into a standard instruction format to output the final path instruction sequence, including the speed and direction control signals.

[0256] Exemplarily, in the dynamic battlefield simulation, the Monte Carlo method generates a probability distribution of the smoothed path curve by random sampling. The core of the Monte Carlo method is to simulate the uncertainty of the battlefield environment, such as terrain changes or enemy movements, through a large number of random iterations.

[0257] In an embodiment, it is assumed that the battlefield simulation is set to 10000 iterations, and the path point coordinates are randomly sampled each time to generate a probability distribution model of the path curve. The path execution probability value, for example, 0.85, indicates that the path successfully passes through the obstacle in 85% of the simulations. If the preset threshold value is 0.8, the path is determined to be feasible. The advantage of this method is that the reliability of the path in a complex environment can be quantified.

[0258] Specifically, when the probability distribution model analyzes the execution probability value, the path feasibility can be determined based on the statistical distribution.

[0259] In a possible implementation, the probability distribution model adopts a normal distribution with a mean of 0.85 and a standard deviation of 0.05. If the execution probability value is higher than the threshold value 0.8, the path feasibility passes.

[0260] For example, the probability value of a certain path is 0.9, which is significantly higher than the threshold value, indicating that it has high stability in the dynamic environment. The advantage of this analysis method is that the robustness of the path is intuitively reflected through statistical means.

[0261] In an embodiment, when extracting the speed control signal from the smoothed path curve, the geometric characteristics of the path curve can be analyzed.

[0262] For example, the curvature of a certain segment of the path curve is small, which is mapped to a speed value of 10 meters per second, generating an instruction containing a timestamp of 1 second and a speed of 10 meters per second. The preliminary speed instruction set is thus formed, covering the entire path. The advantage of this method is to ensure that the speed instruction is consistent with the path geometric characteristics, improving the execution accuracy.

[0263] Preferably, the extraction of the direction control signal depends on the tangent vector of the path curve.

[0264] For example, the tangent vector direction angle of a certain point of the path is 30 degrees, and a direction instruction is generated through vector transformation, containing a timestamp of 1 second and a direction angle of 30 degrees. The preliminary direction instruction set is thus generated, ensuring the continuity of the path direction. The advantage of this method is that the direction instruction is highly matched with the path geometry, reducing the deviation.

[0265] For example, the instruction sequence optimization algorithm adopts dynamic programming to integrate the speed and direction instruction sets.

[0266] In a possible implementation, the dynamic programming algorithm aims to synchronize time, combines the speed instruction 10 meters per second and the direction instruction 30 degrees, and generates an optimized path instruction sequence. The advantage of this method is the time consistency of the instruction sequence, avoiding execution conflicts.

[0267] It can be understood that when judging the adaptability of the instruction sequence, the battlefield environment parameters need to be obtained, such as wind speed 5 meters per second or terrain slope 10 degrees. If the speed and direction instructions of the optimized path instruction sequence adapt to these parameters, such as the speed not exceeding the terrain limit, then the adaptability is determined to pass. The advantage of this judgment method is to ensure that the instruction sequence is compatible with the actual environment.

[0268] It should be noted that the output of the final path instruction sequence needs to be formatted into a standard instruction format.

[0269] For example, a command sequence might include a timestamp of 1 second, a speed of 10 meters per second, and a direction angle of 30 degrees, formatted as a JSON structure. The advantage of this formatting method is that it facilitates system parsing and execution, improving command transmission efficiency.

[0270] Example 2

[0271] like Figure 2 As shown, based on the same inventive concept, this embodiment also provides an unmanned target drone flight path planning system, including:

[0272] The data acquisition module is used to acquire dynamic battlefield environment data;

[0273] The path planning module is used to generate the first path node sequence by assigning dynamic weights based on the threat area coordinates in the dynamic battlefield environment data using the A* algorithm.

[0274] The communication evaluation module is used to calculate the signal strength of each node based on the first path node sequence using the path loss model, evaluate the signal attenuation probability using the Monte Carlo method, and generate a communication quality weight sequence.

[0275] The path optimization module is used to optimize the first path node sequence and generate the second path node sequence by using a genetic algorithm if there are nodes in the communication quality weight sequence that are below a preset threshold.

[0276] The deviation calculation module is used to calculate the Euclidean distance deviation based on the second path node sequence by fusing inertial navigation and global positioning data through Kalman filtering, and generate a deviation data sequence.

[0277] The local correction module is used to perform local correction on the corresponding path segment by using a quadratic programming algorithm if there is a deviation value in the deviation data sequence that exceeds a preset threshold, and to generate a third path node sequence.

[0278] The path smoothing module is used to generate smooth path curve parameters based on the third path node sequence, using a B-spline curve fitting algorithm combined with motion constraints to optimize the control point positions.

[0279] The path execution module is used to evaluate the path execution probability in dynamic battlefield simulation using the Monte Carlo method based on the smooth path curve parameters. If the execution probability is higher than a preset threshold, it generates a final path instruction sequence containing speed and direction control signals.

[0280] The unmanned target drone flight path planning system provided in this embodiment has all the advantages of the unmanned target drone flight path planning method provided in Embodiment 1.

[0281] The above merely provides the preferred embodiments of the present application, and the protection scope of the present application is not limited thereto, and any changes or substitutions within the technical scope disclosed by the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for planning a flight path of a target drone, characterized in that, Comprise: Obtain dynamic battlefield environment data, and according to threat area coordinates in the dynamic battlefield environment data, generate a first path node sequence by using A* algorithm to distribute dynamic weights; According to the first path node sequence, calculate the signal strength of each node by the path loss model, evaluate the signal attenuation probability by the Monte Carlo method, and generate a communication quality weight sequence; If there is a node in the communication quality weight sequence that is lower than a preset threshold, then optimize the first path node sequence by using genetic algorithm to generate a second path node sequence; According to the second path node sequence, calculate the Euclidean distance deviation by Kalman filtering fusion of inertial navigation and global positioning data to generate a deviation data sequence; If there is a deviation value in the deviation data sequence that exceeds a preset threshold, then use a quadratic programming algorithm to locally correct the corresponding path segment to generate a third path node sequence; According to the third path node sequence, use B-spline curve fitting algorithm to optimize the control point position combined with motion constraints to generate a smooth path curve parameter; According to the smooth path curve parameter, evaluate the path execution probability in the dynamic battlefield simulation by the Monte Carlo method, and if the execution probability is higher than a preset threshold, then generate a final path instruction sequence containing speed and direction control signals.

2. The method of claim 1, wherein the process of obtaining dynamic battlefield environment data comprises: Collecting dynamic battlefield environment data through radar, infrared and optical sensors to obtain a multi-source sensor data set; Using a multi-sensor fusion algorithm to preprocess the multi-source sensor data set to obtain preliminary fusion data; Applying Kalman filtering algorithm to optimize the preliminary fusion data to obtain smooth target position, threat area coordinates and terrain height information; If the distance between the target position and the threat area coordinates is less than a preset threshold, then determine the high-risk target through coordinate information analysis to obtain a high-risk target list; According to the high-risk target list and the terrain height information, calculate the relative height of the target relative to the terrain to obtain target terrain association data; Using clustering algorithm to group the targets in space through the target terrain association data to obtain target distribution mode; According to the target distribution mode, generate real-time spatial data of the dynamic battlefield environment to obtain battlefield situation information.

3. The method of claim 1, wherein the process of generating a first path node sequence by using A* algorithm to distribute dynamic weights comprises: Extracting threat area coordinates and threat level information based on the dynamic battlefield environment data to determine threat area division results; Calculating the dynamic weights of the threat areas according to the threat level information by linear mapping method to obtain a weight distribution table; wherein the dynamic weights are determined based on linear mapping of threat level; If the dynamic weight exceeds a preset threshold, then generate an initial first path node sequence based on the weight distribution table by using A* algorithm to determine an initial path planning; According to the initial path planning, obtain the trend of environmental data change in the path node sequence, judge whether the path crosses a high-threat area, and obtain optimization demand evaluation. ​ ​ If the optimization demand evaluation shows that the path passes through a high-threat area, the path is recalculated by the Dijkstra algorithm, the node sequence is adjusted based on the dynamic weight, and an adjusted first path node sequence is generated; Obtain the environmental data of the adjusted first path node sequence, integrate the threat area coordinates and the dynamic weight using data fusion technology, and determine the final path planning; Through the final path planning, the coordinate data of the path node sequence is generated, and the optimized first path node sequence is output.

4. The method of claim 1, wherein the process of generating a communication quality weight sequence comprises: By analyzing the first path node sequence, the position distribution of each node and the signal propagation distance are obtained, and the initial signal strength is determined; Using a path loss model, combined with environmental interference factors, the signal strength of each node is calculated to obtain a signal strength sequence; If the signal strength is lower than the preset threshold, the signal attenuation probability is simulated by the Monte Carlo method to generate an attenuation probability distribution; According to the attenuation probability distribution, the communication quality weight of each node is calculated to obtain the communication quality weight sequence; For the weight sequence generation result, evaluate the communication quality and determine whether there is a low-quality node; If there is a low-quality node, adjust the node position distribution and recalculate the signal strength sequence; Iteratively optimize the weight sequence to generate the final communication quality weight sequence.

5. The method of claim 1, wherein the process of generating a second path node sequence comprises: If there is a node in the communication quality weight sequence that is lower than the preset threshold, the node identifier of the node lower than the preset threshold is obtained from the communication quality weight sequence to obtain a set of nodes to be optimized; According to the set of nodes to be optimized, a population of the first path node sequence is initialized using a genetic algorithm to generate an initial path node sequence set; For the initial path node sequence set, the weighted sum of the communication quality weight and the path length of each path is calculated to obtain a set of fitness function values; If there is a path in the set of fitness function values that is lower than the preset fitness threshold, a new path node sequence set is generated by the crossover and mutation operations of the genetic algorithm; According to the new path node sequence set, the weighted sum of the communication quality weight and the path length is recalculated to obtain an updated set of fitness function values; By comparing the updated set of fitness function values, the path node sequence with the highest fitness value is selected to generate the second path node sequence; According to the second path node sequence, the node connection relationship of the communication network is adjusted to determine the optimized communication path configuration.

6. The method of claim 1, wherein the process of generating a bias data sequence comprises: Through the second path node sequence, inertial navigation data and global positioning data are obtained, and Kalman filtering is used for preliminary fusion to obtain initial fusion data; For the initial fusion data, the Euclidean distance bias between each node is calculated to generate a preliminary bias data sequence; If there is an outlier in the preliminary bias data sequence, the outlier is removed by a preset threshold judgment to obtain a corrected bias data sequence; ​ ​ ​ According to the modified deviation data sequence, a quadratic Kalman filter is used to optimize data smoothness to generate an optimized deviation data sequence; From the optimized deviation data sequence, deviation statistical features are extracted, and deviation mean and variance are calculated to obtain deviation distribution characteristics; Through the deviation distribution characteristics, a least squares method is used to fit the deviation trend to construct a deviation trend model; According to the deviation trend model, the deviation sequence of the subsequent path node is predicted, and the deviation data sequence is obtained.

7. The method of claim 1, wherein the process of generating the third path node sequence comprises: If there is a deviation value in the deviation data sequence that exceeds a preset threshold, an abnormal deviation point is determined through threshold judgment to obtain an abnormal point set; The corresponding relationship between the abnormal point set and the path segment is analyzed to obtain the path segment to be optimized, and a set of segments to be optimized is determined; A quadratic programming algorithm is used to locally optimize the set of segments to be optimized to generate a temporary path node sequence; If the deviation value of the temporary path node sequence still exceeds the preset threshold, the parameters of the quadratic programming algorithm are adjusted through iteration to obtain an optimized path node sequence; According to the optimized path node sequence, the continuity of the node sequence is obtained, and the smoothness of the sequence is determined; According to the smoothness determination result, a spline interpolation algorithm is used to smooth the optimized path node sequence to generate the third path node sequence; If the deviation value of the third path node sequence meets the preset threshold, the final path node sequence is generated through sequence integration.

8. The method of claim 1, wherein the process of generating the smooth path curve parameter comprises: Based on the third path node sequence, the node coordinates and connection order are extracted through preprocessing to determine an initial path point set; A B-spline curve fitting algorithm is used to calculate the initial position of the control point based on the initial path point set to obtain preliminary curve parameters; If the preliminary curve parameters meet the motion constraints, the control point position is retained; If not, the control point is adjusted according to the constraint conditions to generate an optimized control point set; Through the optimized control point set, the B-spline curve is recalculated to obtain the smooth path curve parameter; According to the smooth path curve parameter, the path smoothness is analyzed, and if the smoothness is lower than the preset threshold, the control point position is iteratively adjusted to determine the final control point set; According to the final control point set, a smooth path curve is generated, and the curve parameter and path point sequence are outputted; From the output curve parameter, the path geometric features are extracted to generate path description data, and the final smooth path is determined.

9. The method of claim 1, wherein the process of generating the final path instruction sequence containing speed and direction control signals if the execution probability is higher than the preset threshold in the dynamic battlefield simulation through the Monte Carlo method comprises: The probability distribution of the smooth path curve is obtained from the dynamic battlefield simulation through the Monte Carlo method to obtain the execution probability value; The execution probability value is compared with the preset threshold using a probability distribution model to determine the path feasibility; ​ ​ ​ If the path feasibility is determined as executable, extract a speed control signal from the smooth path curve to generate a preliminary speed instruction set; Extract a direction control signal from the smooth path curve to generate a preliminary direction instruction set; Merge the preliminary speed instruction set and the preliminary direction instruction set through an instruction sequence optimization algorithm to obtain an optimized path instruction sequence; Obtain battlefield environment parameters for the optimized path instruction sequence and determine the adaptability of the instruction sequence; If the adaptability of the instruction sequence is determined as passed, output a final path instruction sequence.

10. An unmanned target aircraft flight path planning system, characterized by, The method comprises the following steps: A data acquisition module is configured to obtain dynamic battlefield environment data; A path planning module is configured to generate a first path node sequence by assigning dynamic weights using an A* algorithm based on threat region coordinates in the dynamic battlefield environment data; A communication evaluation module is configured to calculate the signal strength of each node by a path loss model based on the first path node sequence, and to generate a communication quality weight sequence by evaluating the signal attenuation probability using a Monte Carlo method; A path optimization module is configured to optimize the first path node sequence using a genetic algorithm to generate a second path node sequence if there is a node in the communication quality weight sequence that is lower than a preset threshold; A deviation calculation module is configured to calculate the Euclidean distance deviation by fusing inertial navigation and global positioning data through Kalman filtering based on the second path node sequence to generate a deviation data sequence; A local correction module is configured to perform local correction on the corresponding path segment using a quadratic programming algorithm to generate a third path node sequence if there is a deviation value in the deviation data sequence that exceeds a preset threshold; A path smoothing module is configured to optimize the control point position by combining motion constraints using a B-spline curve fitting algorithm based on the third path node sequence to generate a smooth path curve parameter; A path execution module is configured to evaluate the path execution probability in a dynamic battlefield simulation by a Monte Carlo method based on the smooth path curve parameter, and to generate a final path instruction sequence containing speed and direction control signals if the execution probability is higher than a preset threshold.

Images