UAV-based geological mapping method, system and storage medium

Abstract

The present application relates to the technical field of cooperative control of unmanned aerial vehicles (UAVs), and discloses a geological mapping method, system, and storage medium based on UAVs. The method includes: processing real-time wind field data and compensation control instruction sets to obtain a wind field compensation model, optimizing the wind field compensation model, and obtaining adaptive route data; analyzing and processing UAV battery status data, adaptive route data, and task completion data through an energy consumption prediction model to obtain task optimization parameters, dynamically allocating the task optimization parameters, and obtaining a task execution instruction set; processing the status information and task execution instruction sets of multiple UAVs to obtain cooperative control data, and performing real-time calculation and processing on the cooperative control data to obtain a multi-machine cooperative control instruction set. By establishing a distributed cooperative control mechanism, the present application realizes intelligent task allocation and anti-collision control during multi-machine collaborative operations, significantly improving the efficiency and reliability of geological mapping.

Description

Technical Field

[0001] The present application relates to the field of UAV collaborative control, and in particular to a UAV-based geological mapping method, system, and storage medium. Background Art

[0002] With the development of geological surveying and mapping technology, drones (UAVs) have been widely used in geological surveying and mapping due to their advantages such as high maneuverability and high surveying efficiency. Existing UAV geological surveying and mapping systems primarily operate in a stand-alone mode, collecting data via pre-set routes and then integrating UAV control and analysis with ground control points. This surveying and mapping method has achieved good results in areas with relatively simple terrain conditions, such as plains and hills. Furthermore, some systems also integrate GPS / IMU combined navigation and real-time image transmission, enhancing the controllability of the surveying process and the reliability of the data.

[0003] However, existing technologies still have some problems under complex terrain conditions: first, the single-machine operation mode is difficult to cope with large-scale and complex terrain surveying and mapping tasks, and the operation efficiency is low; second, the preset routes cannot be dynamically adjusted according to real-time wind field conditions, which can easily affect flight safety and data quality; third, there is a lack of accurate prediction and management of drone energy consumption, making it difficult to ensure long-term and continuous surveying and mapping operations; finally, there is a lack of effective task allocation and anti-collision mechanisms when multiple machines work together, which increases operational risks. Summary of the Invention

[0004] The present application provides a geological mapping method, system and storage medium based on drones, which are used to solve the technical problem that existing drone geological mapping systems lack intelligent control and multi-machine collaboration capabilities under complex terrain conditions.

[0005] In the first aspect, the present application provides a geological mapping method based on a drone, which includes: performing fuzzy controller processing on the terrain feature data and mapping parameter requirements of the survey area to obtain an initial control parameter group and a controller gain matrix; processing the digital elevation model of the survey area and the initial control parameter group through a partitioning algorithm to obtain partition control data, and performing closed-loop control processing on the partition control data and the controller gain matrix through a cascade PID algorithm to obtain a posture control instruction set; fusing the position data collected by multiple sensors and the posture control instruction set through an extended Kalman filter algorithm to obtain posture fusion data, and compensating the posture fusion data through a model predictive control algorithm to obtain a compensation control instruction set. Instruction set; the real-time wind field data collected by the airborne sensor and the compensation control instruction set are processed by a three-dimensional modeling algorithm to obtain a wind field compensation model, and the wind field compensation model is optimized by a dynamic programming algorithm to obtain adaptive route data; the UAV battery status data, the adaptive route data and the task completion data are analyzed and processed by an energy consumption prediction model to obtain task optimization parameters, and the task optimization parameters are dynamically allocated and processed by a task planning algorithm to obtain a task execution instruction set; the status information of multiple UAVs and the task execution instruction set are processed by a distributed collaborative algorithm to obtain collaborative control data, and the collaborative control data are calculated and processed in real time by a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

[0006] In a second aspect, the present application provides a geological mapping system based on a drone, the geological mapping system based on a drone comprising:

[0007] The processing module is used to perform fuzzy controller processing on the terrain feature data of the survey area and the surveying and mapping parameter requirements to obtain an initial control parameter group and a controller gain matrix;

[0008] A control module is used to process the digital elevation model of the survey area and the initial control parameter group through a partition algorithm to obtain partition control data, and to perform closed-loop control processing on the partition control data and the controller gain matrix through a cascade PID algorithm to obtain a posture control instruction set;

[0009] A fusion module is used to fuse the position data collected by multiple sensors and the attitude control instruction set through an extended Kalman filter algorithm to obtain attitude fusion data, and to compensate the attitude fusion data through a model predictive control algorithm to obtain a compensation control instruction set;

[0010] an optimization module for processing the real-time wind field data collected by the airborne sensor and the compensation control instruction set through a three-dimensional modeling algorithm to obtain a wind field compensation model, and optimizing the wind field compensation model through a dynamic programming algorithm to obtain adaptive route data;

[0011] An analysis module is used to analyze and process the UAV battery status data, the adaptive route data, and the mission completion data using an energy consumption prediction model to obtain mission optimization parameters, and dynamically allocate the mission optimization parameters using a mission planning algorithm to obtain a mission execution instruction set;

[0012] The computing module is used to process the status information of multiple drones and the task execution instruction set through a distributed collaborative algorithm to obtain collaborative control data, and to perform real-time computation and processing on the collaborative control data through a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

[0013] In a third aspect, the present application provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the above-mentioned drone-based geological mapping method.

[0014] The technical solution provided in this application uses a fuzzy controller to process the survey area's terrain feature data and surveying and mapping parameter requirements, enabling intelligent generation of initial control parameters under complex terrain conditions and improving the accuracy and adaptability of parameter settings. A partitioning algorithm and a cascaded PID algorithm are used to process the survey area's digital elevation model and initial control parameter set, enabling precise control of regions with different terrain features and enhancing the flight stability of the UAV in complex terrain. An extended Kalman filter algorithm is used to fuse position data collected by multiple sensors and an attitude control instruction set, combined with a model predictive control algorithm for compensation. This improves the accuracy of position and attitude estimation and enhances the robustness of the control system. A three-dimensional modeling algorithm and a dynamic programming algorithm are used to process real-time wind field data collected by onboard sensors and the compensation control instruction set, enabling precise compensation for wind field effects and real-time optimization of flight routes, thereby improving the reliability of surveying and mapping tasks under complex wind field conditions. An energy consumption prediction model and a mission planning algorithm are used to analyze and process the UAV's battery status data, adaptive route data, and mission completion data, optimizing energy efficiency and rationally planning mission execution. By processing the status information and task execution instruction sets of multiple drones through a distributed collaborative algorithm and a safe distance field algorithm, this method achieves intelligent task allocation and collision risk control during multi-drone collaboration, improving the efficiency and safety of multi-drone collaborative mapping. Overall, the proposed method significantly enhances the adaptability, reliability, and efficiency of UAV geological mapping in complex environments through multi-level closed-loop control and multi-algorithm collaboration. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0016] Figure 1 This is a schematic diagram of an embodiment of a geological mapping method based on a drone in an embodiment of the present application;

[0017] Figure 2 This is a schematic diagram of an embodiment of a geological mapping system based on a drone in an embodiment of the present application. DETAILED DESCRIPTION

[0018] Embodiments of the present application provide a method, system and storage medium for geological mapping based on drones. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiments of the present application, an embodiment of the geological mapping method based on a drone includes:

[0020] Step S101: Perform fuzzy controller processing on the survey area terrain feature data and surveying and mapping parameter requirements to obtain an initial control parameter group and a controller gain matrix;

[0021] Step S102: Processing the digital elevation model of the survey area and the initial control parameter group by a partition algorithm to obtain partition control data, and performing closed-loop control processing on the partition control data and the controller gain matrix by a cascade PID algorithm to obtain a posture control instruction set;

[0022] Step S103: fusing the position data collected by the multi-sensor and the attitude control instruction set through an extended Kalman filter algorithm to obtain attitude fusion data, and compensating the attitude fusion data through a model predictive control algorithm to obtain a compensation control instruction set;

[0023] Step S104: Processing the real-time wind field data and compensation control instruction set collected by the airborne sensor through a three-dimensional modeling algorithm to obtain a wind field compensation model, and optimizing the wind field compensation model through a dynamic programming algorithm to obtain adaptive route data;

[0024] Step S105: Analyze and process the drone battery status data, adaptive route data, and mission completion data using an energy consumption prediction model to obtain mission optimization parameters, and dynamically allocate the mission optimization parameters using a mission planning algorithm to obtain a mission execution instruction set.

[0025] Step S106: The status information and task execution instruction sets of multiple UAVs are processed by a distributed collaborative algorithm to obtain collaborative control data, and the collaborative control data is processed in real time by a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

[0026] It is understandable that the execution subject of this application can be a geological surveying and mapping system based on a drone, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0027] Optionally, the terrain feature data and surveying parameter requirements for the survey area are first processed and then processed using a fuzzy controller to generate an initial control parameter set and a controller gain matrix. The terrain feature data includes elevation, slope, and land cover type data, while the surveying parameter requirements include parameters such as spatial resolution, overlap, and heading angle. The fuzzy controller calculates the gradient of the elevation data to obtain the terrain slope distribution. This is then combined with the resolution requirements in the surveying parameter requirements to establish a fuzzy rule base. For example, when the terrain slope is greater than 30 degrees and the resolution is required to be higher than 5 cm, the flight altitude must be set below 80 meters, and the speed must be controlled within 15 meters per second. These parameters constitute the initial control parameter set. Parameter sensitivity analysis determines the weight of each parameter's influence on the control effect, forming a controller gain matrix. A partitioning algorithm is then applied to the digital elevation model of the survey area and the initial control parameter set to generate partitioned control data. The partitioning algorithm first applies a mean filter to the elevation model to eliminate outliers. The survey area is then divided into flat, transitional, and complex areas based on the terrain characteristics. Different control strategies are employed for different zones, achieving closed-loop control through a cascaded PID algorithm. This algorithm comprises an inner attitude loop and an outer position loop, responsible for rapid response and stable tracking, respectively. The attitude control instruction set includes control variables for pitch, roll, and yaw angles, ensuring flight stability in complex terrain.

[0028] After acquiring multi-sensor position data and attitude control instruction sets, an extended Kalman filter (EKF) algorithm is used for data fusion. This algorithm estimates the state of a nonlinear system through two phases: state prediction and measurement update. The state vector contains position, velocity, and attitude information, while the observation vector contains GPS position data and IMU attitude data. The fused position and attitude data is compensated using a model predictive control algorithm. The predictive control time domain typically takes 2-3 seconds, during which the drone's state is predicted and optimized, generating a compensation control instruction set. Three-dimensional modeling is performed on the wind field data collected by onboard sensors and the compensation control instruction set to establish a wind field compensation model. The wind field data includes information such as wind speed, wind direction, and turbulence intensity. An interpolation algorithm is used to construct a continuous three-dimensional wind field distribution. A dynamic programming algorithm is used to optimize the wind field compensation model and generate adaptive route data. The dynamic programming algorithm calculates a route cost matrix, comprehensively considering flight distance, energy consumption, and wind field effects, to select the optimal route.

[0029] Energy consumption prediction and analysis are performed on drone battery status data, adaptive route data, and mission completion data. Battery status data includes voltage, current, and temperature information, and the remaining available time is estimated using an energy consumption prediction model. A task planning algorithm dynamically allocates tasks based on task optimization parameters and generates task execution instruction sets. This allocation process considers factors such as task priority, energy balance, and time constraints. Finally, distributed collaborative processing is performed on the status information and task execution instruction sets of multiple drones. The distributed collaborative algorithm establishes a communication topology to enable information exchange between multiple drones. The safe distance field algorithm calculates the relative distance and speed between drones, constructs a collision risk assessment model, and generates a multi-drone collaborative control instruction set.

[0030] For example, the survey area is 2 square kilometers, with a terrain elevation difference of 200 meters, requiring a spatial resolution better than 5 centimeters. First, the terrain slope distribution is calculated based on elevation data, with a maximum slope of 45 degrees. The fuzzy controller generates initial control parameters based on this data: flight altitude of 100 meters and speed of 20 meters per second in the flat area; flight altitude of 80 meters and speed of 15 meters per second in the transition area; and flight altitude of 60 meters and speed of 10 meters per second in the complex area. Zoning control data indicates that the survey area consists of 40% flat areas, 35% transition areas, and 25% complex areas. The extended Kalman filter algorithm, integrating GPS and IMU data, achieves position accuracy better than 10 centimeters and attitude accuracy better than 0.5 degrees. Wind field data indicates that wind speeds vary between 5 and 8 meters per second in the survey area. The adaptive route generated by the dynamic programming algorithm saves 15% of flight distance compared to conventional routes. The mission is assigned to three drones, each with a remaining battery level between 75% and 85%. Energy consumption forecasts indicate sufficient flight time to complete their respective mission areas. Using a distributed collaborative algorithm, drones maintained a safe distance of 50-100 meters between each other, successfully avoiding the risk of flight trajectory deviation due to wind field fluctuations. The entire mapping mission was completed within 2 hours, meeting the required spatial resolution and data quality.

[0031] In this embodiment, a fuzzy controller processes the survey area's terrain feature data and surveying and mapping parameter requirements, enabling intelligent generation of initial control parameters under complex terrain conditions and improving the accuracy and adaptability of parameter settings. A partitioning algorithm and a cascaded PID algorithm are used to process the survey area's digital elevation model and initial control parameter set, enabling precise control of regions with different terrain features and enhancing the flight stability of the UAV in complex terrain. An extended Kalman filter algorithm is used to fuse position data collected by multiple sensors and an attitude control instruction set, combined with a model predictive control algorithm for compensation. This improves the accuracy of position and attitude estimation and enhances the robustness of the control system. A three-dimensional modeling algorithm and a dynamic programming algorithm are used to process real-time wind field data collected by onboard sensors and the compensation control instruction set, enabling precise compensation for wind field effects and real-time optimization of flight routes, thereby improving the reliability of surveying and mapping tasks under complex wind field conditions. An energy consumption prediction model and a task planning algorithm are used to analyze and process the UAV's battery status data, adaptive route data, and mission completion data, optimizing energy efficiency and rationally planning mission execution. By processing the status information and task execution instruction sets of multiple drones through a distributed collaborative algorithm and a safe distance field algorithm, this method achieves intelligent task allocation and collision risk control during multi-drone collaboration, improving the efficiency and safety of multi-drone collaborative mapping. Overall, the proposed method significantly enhances the adaptability, reliability, and efficiency of UAV geological mapping in complex environments through multi-level closed-loop control and multi-algorithm collaboration.

[0032] In an optional embodiment, the process of executing step S101 may specifically include the following steps:

[0033] (1) Perform gradient calculation on the elevation data in the survey area terrain feature data to obtain terrain slope data, and then perform interval division on the terrain slope data to obtain terrain complexity classification data;

[0034] (2) Quantify the spatial resolution data in the surveying and mapping parameter requirements to obtain the reference flight altitude data, and correlate the reference flight altitude data with the terrain complexity classification data to obtain the altitude adjustment coefficient;

[0035] (3) Analyze the overlap data in the surveying and mapping parameter requirements to obtain speed constraints, and then define the flight speed range based on the speed constraints to obtain speed range data;

[0036] (4) Performing fuzzy rule generation processing on the speed interval data and the height adjustment coefficient to obtain a control rule base, and constructing a membership function on the control rule base to obtain a fuzzy inference set;

[0037] (5) Defuzzifying the fuzzy inference set to obtain the original control parameters, and performing equilibrium constraint processing on the original control parameters to obtain the initial control parameter group;

[0038] (6) Perform sensitivity analysis on the parameters in the initial control parameter group to obtain the parameter influence weights, and generate the controller gain matrix based on the parameter influence weights.

[0039] Optionally, the elevation data in the terrain feature data of the survey area is processed by gradient calculation. The elevation data is obtained through a digital elevation model, which contains the altitude value of each grid point in the survey area. The elevation data is subjected to gradient calculation, and the terrain slope data is calculated by dividing the elevation difference of adjacent grid points by the horizontal distance. According to the statistical distribution characteristics of the terrain slope data, the slope is divided into multiple intervals: 0-15 degrees is a gentle area, 15-30 degrees is a transition area, and above 30 degrees is a steep area, thereby obtaining terrain complexity classification data. For the spatial resolution data in the surveying and mapping parameter requirements, quantification processing is required to obtain the benchmark flight altitude data. Spatial resolution represents the actual distance corresponding to one pixel on the ground, which is directly related to the flight altitude and camera parameters. The calculation of the benchmark flight altitude data takes into account parameters such as camera focal length and sensor size, and is calculated using the following height adjustment coefficient formula:

[0040]

[0041] Among them, H adj represents the height adjustment factor, GSD represents the target ground resolution (meters / pixel), f represents the camera focal length (mm), p s represents the sensor pixel size (micrometers), α represents the terrain slope (radians), K t Represents the terrain complexity correction factor. For the overlap data required by surveying and mapping parameters, speed constraints must be obtained through analytical processing. Overlap data includes heading overlap and lateral overlap. Typically, heading overlap is required to be at least 80%, and lateral overlap is required to be at least 60%. Based on the overlap requirement and the camera capture interval, the maximum flight speed limit can be determined. Speed ​​range data is derived by comprehensively considering flight altitude, overlap requirements, and camera performance parameters.

[0042] The speed range data and altitude adjustment coefficients are processed using fuzzy rule generation to establish a control rule base. The fuzzy rules divide the speed and altitude parameters into several linguistic variables, such as "low," "medium," and "high," establishing a mapping relationship between input and output. A membership function is constructed for the control rule base, and the membership value is calculated using the following formula:

[0043]

[0044] Here, μ(x) represents the membership function value, x represents the input variable, c represents the membership function center value, σ represents the standard deviation parameter, β represents the shape adjustment factor, and γ represents the weight coefficient. This method generates a fuzzy inference set. The fuzzy inference set is defuzzified, converting the fuzzy inference results into clear control variables to obtain the original control parameters. Defuzzification uses the center of gravity method to apply equilibrium constraints to the original control parameters, ensuring that the parameter values ​​are within a reasonable range. This results in an initial control parameter set. A sensitivity analysis is performed on the parameters in the initial control parameter set. By varying individual parameters and observing changes in the system response, the parameter influence weights are calculated, ultimately generating a controller gain matrix.

[0045] For example, the survey area is 3 square kilometers, and the elevation data resolution is 5 meters. Gradient calculations are performed on the elevation data to determine the terrain slope distribution: 45% of the terrain is flat, with slopes between 0 and 15 degrees; 35% of the terrain is transitional, with slopes between 15 and 30 degrees; and 20% of the terrain is steep, with slopes exceeding 30 degrees. Setting a target ground resolution of 3 cm / pixel and using a camera with a focal length of 35 mm and a pixel size of 3.76 microns, the baseline flight altitude for the flat area is calculated to be 120 meters. Using the altitude adjustment coefficient formula, the flight altitudes for the transition and steep areas are adjusted to 90 and 60 meters, respectively. The heading overlap is set to 85%, the lateral overlap is 65%, and the minimum camera interval is 0.5 seconds. Based on these parameters, speed constraints are calculated: a maximum speed of 18 meters / second in the flat area, 14 meters / second in the transition area, and 10 meters / second in the steep area. A fuzzy rule base is established, containing 27 rules (three levels each for speed and altitude, and three levels for terrain complexity). Through membership function calculation, a complete fuzzy reasoning set is obtained.

[0046] After defuzzification, specific control parameters for the flat, transition, and steep regions were obtained. Sensitivity analysis showed that flight altitude had a weight of 0.45 on surveying accuracy, flight speed had a weight of 0.35, and overlap had a weight of 0.20. This information was used to generate a controller gain matrix for subsequent flight control. The entire process automated the generation of flight control parameters from terrain data, ensuring reliable execution of surveying tasks.

[0047] In an optional embodiment, the process of executing step S102 may specifically include the following steps:

[0048] (1) Perform mean filtering on the elevation data in the digital elevation model of the survey area to obtain smooth elevation data, and perform slope calculation on the smooth elevation data to obtain terrain slope distribution data;

[0049] (2) Perform cluster analysis on the terrain slope distribution data to obtain terrain feature partitions, and perform boundary extraction on the terrain feature partitions to obtain partition boundary data;

[0050] (3) Perform parameter matching processing on the partition boundary data and the initial control parameter group to obtain a partition parameter set, and perform interpolation processing on the partition parameter set to obtain a continuous parameter field;

[0051] (4) performing boundary constraint processing on the continuous parameter field to obtain constraint parameter data, and generating partition control data based on the constraint parameter data;

[0052] (5) performing error calculation processing on the partition control data to obtain position error data, and performing differential processing on the position error data to obtain speed error data;

[0053] (6) Integrating the speed error data to obtain integrated error data, and performing proportional operation on the integrated error data according to the controller gain matrix to obtain the basic control quantity;

[0054] (7) The basic control quantity is limited to obtain a bounded control quantity, and the bounded control quantity is subjected to attitude solution processing to obtain an attitude control instruction set.

[0055] Optionally, the processing of the digital elevation model data of the survey area first requires mean filtering to eliminate outliers and noise. The mean filter uses a sliding window method to average the data within the surrounding 3×3 or 5×5 grid range for each elevation point to obtain smoothed elevation data. The smoothed elevation data is used for slope calculation. By calculating the ratio of the elevation difference between adjacent grid points to the horizontal distance, the terrain slope distribution data is obtained. The terrain slope distribution data is analyzed using the K-means clustering algorithm to cluster areas with similar slope characteristics to form terrain feature partitions. The K-means clustering process includes four steps: initial cluster center selection, distance calculation, category division, and center update, and ultimately obtains terrain partitions of different categories. Subsequently, the boundaries of the terrain feature partitions are extracted, and the Sobel operator is used to detect the region boundaries to obtain partition boundary data.

[0056] The partition boundary data is matched against the initial control parameter set, and the control parameters of different partitions are linked to the boundary data to generate a partition parameter set. This partition parameter set contains parameters such as flight altitude, speed, and heading angle for each partition. To eliminate sudden changes in parameters between partitions, bilinear interpolation is performed on the partition parameter set to generate a continuous parameter field, ensuring smooth parameter transitions at partition boundaries.

[0057] Boundary constraints are applied to the continuous parameter field, primarily considering the UAV's performance limitations and flight safety requirements. These constraints include maximum flight altitude, minimum flight altitude, and maximum speed, generating constraint parameter data. Based on these constraint parameter data, partition control data is generated, including the specific control strategy for each partition. This partition control data is compared with the actual flight position to calculate position error data. The position error represents the deviation between the UAV's current position and the planned route position. The position error data is differentiated to generate velocity error data, reflecting the rate of change of the position error. The velocity error data is then integrated to generate integral error data, which is used to eliminate steady-state errors. The integral error data is then proportionally calculated using the controller gain matrix to generate the basic control variable.

[0058] The basic control quantity is limited to ensure that the control instruction is within the executable range of the UAV and a bounded control quantity is obtained. The calculation formula of the bounded control quantity is:

[0059]

[0060] Among them, U bound represents the bounded control quantity, K1, K2, K3 represent the weight coefficients of position error, speed error and integral error respectively, E p represents the normalized position error, E v represents the normalized velocity error, E i represents the normalized integral error, U max Indicates the maximum control amount.

[0061] Finally, the bounded control quantity is processed for attitude solution, and the control quantity is converted into a specific attitude angle instruction. The calculation formula for attitude solution is:

[0062]

[0063] in, θ, ψ represent the roll angle, pitch angle and yaw angle respectively, W1, W2, W3, W4 and W5 are attitude weight coefficients, ω is the angular velocity, t is the time parameter, U x 、U y 、U z is a three-axis bounded control quantity.

[0064] For example, in a geological mapping mission covering a 4-square-kilometer area, the digital elevation model resolution was 2 meters. A 5×5 mean filter was used to eliminate 98% of abnormal elevation values. Slope was calculated from the smoothed elevation data, yielding distribution data with a slope range of 0-65 degrees. K-means clustering (K=3) was used to categorize the terrain into gentle zones (0-20 degrees), transition zones (20-40 degrees), and steep zones (above 40 degrees). Boundary extraction yielded zone boundary data totaling approximately 25 kilometers. The initial control parameter set included a flight altitude of 120 meters and a speed of 18 meters per second in the gentle zone, a flight altitude of 90 meters and a speed of 15 meters per second in the transition zone, and a flight altitude of 70 meters and a speed of 12 meters per second in the steep zone. A continuous parameter field was generated using bilinear interpolation, with the parameter change rate at zone boundaries controlled to no more than 2% per meter. Considering the performance constraints of the drone, the maximum flight altitude was set to 150 meters, the minimum flight altitude to 50 meters, and the maximum speed to 20 meters per second.

[0065] During actual flight, position error is controlled within ±3 meters, and velocity error is within ±2 meters per second. After weighting the integral error, the resulting basic control variable is between -1 and 1. Limiting ensures that roll angles are within ±30 degrees, pitch angles within ±25 degrees, and yaw angles within ±180 degrees. The resulting attitude control command set is updated at a frequency of 50 Hz, ensuring stable flight in complex terrain.

[0066] In an optional embodiment, the process of executing step S103 may specifically include the following steps:

[0067] (1) Performing timestamp alignment processing on the position data collected by multiple sensors to obtain synchronized position data, and constructing state vectors on the synchronized position data and attitude control instruction sets to obtain initial state estimation;

[0068] (2) Performing state prediction processing on the initial state estimate to obtain a priori state quantity, and performing covariance transfer processing on the priori state quantity to obtain a priori covariance;

[0069] (3) Observe and predict the prior state quantity to obtain the predicted observation value, and calculate the residual between the predicted observation value and the actual observation value to obtain the observation residual;

[0070] (4) Perform gain calculation on the observation residual and prior covariance to obtain the Kalman gain, and perform state update processing on the Kalman gain to obtain the pose fusion data;

[0071] (5) Perform error prediction processing on the pose fusion data to obtain a prediction error sequence, and construct a cost function on the prediction error sequence to obtain a control objective function;

[0072] (6) performing constraint condition setting processing on the control objective function to obtain a constraint condition set, and performing rolling time domain partition processing on the constraint condition set to obtain a time domain control interval;

[0073] (7) Perform sequence prediction processing on the state within the time domain control interval to obtain the predicted control quantity, and perform feedback correction processing on the predicted control quantity to obtain the compensation control instruction set.

[0074] Optionally, processing multi-sensor position data first requires timestamp alignment. Multiple sensors, such as GPS, IMUs, and lidars, each have different data sampling frequencies: 10 Hz for GPS, 200 Hz for IMUs, and 20 Hz for lidars. Timestamp alignment uses an interpolation algorithm to align data of varying frequencies to a common time base, generating synchronized position data. A state vector is constructed from the synchronized position data and attitude control instruction set. The state vector contains 12 components, including position coordinates, velocity, attitude angle, and angular velocity, forming the initial state estimate.

[0075] The initial state estimate is processed for state prediction, and the UAV's dynamic model is used to predict the state at the next moment. State prediction takes into account the UAV's motion characteristics and external forces, including gravity, thrust, and aerodynamics. The predicted prior state represents the system's best estimate before the measurement update. A covariance transfer process is performed on the prior state to calculate the uncertainty of the state prediction and obtain the prior covariance matrix. The covariance matrix reflects the correlation between the state quantities and the reliability of the prediction. The prior state is substituted into the observation equation and the observation prediction process is performed to obtain the predicted observation value. The observation equation describes the relationship between the state quantity and the sensor measurement value. The predicted observation value is compared with the actual observation value to calculate the observation residual. The observation residual reflects the deviation between the predicted value and the actual measurement value and is an important indicator for evaluating prediction accuracy.

[0076] The Kalman gain is calculated based on the observation residual and prior covariance. The Kalman gain determines the weight assigned to the predicted and observed values; a larger value indicates a greater tendency to trust the observed value. The Kalman gain is then updated to perform a weighted fusion of the predicted and observed values ​​to produce pose fusion data. This pose fusion data combines the advantages of the prediction model and observation data, providing a more accurate state estimate. Error prediction is performed on the pose fusion data to predict the state error at future moments based on the system model, resulting in a predicted error sequence. This predicted error sequence is used to evaluate control effectiveness and optimize control strategies. By setting error weights and control costs, a cost function is constructed for the predicted error sequence to produce the control objective function. This control objective function comprehensively considers multiple aspects, including tracking accuracy and control energy consumption.

[0077] Constraints are set for the control objective function, including state constraints and control constraints. State constraints ensure that the drone's motion remains within a safe range, while control constraints ensure that the control output does not exceed the actuator's capabilities. The constraint set is partitioned over a rolling time domain to produce a time domain control interval. This time domain control interval is typically 2-3 seconds, during which prediction and optimization are performed. Sequence prediction of the states within the time domain control interval is performed, and dynamic programming is used to search for the optimal control sequence, resulting in a predicted control variable. The predicted control variable undergoes feedback correction to compensate for model errors and external disturbances, ultimately generating a set of compensated control instructions.

[0078] For example, in a complex terrain mapping mission, a drone was equipped with three sensors: GPS, IMU, and LiDAR. Raw data sampling showed: GPS position accuracy of ±2 meters, updated at a 10Hz frequency; IMU attitude accuracy of ±0.1 degrees, updated at a 200Hz frequency; and LiDAR ranging accuracy of ±0.05 meters, updated at a 20Hz frequency. Timestamp alignment was used to standardize all data to a 50Hz frequency. The initial values ​​of the constructed state vector were: position (100, 200, 50) meters, velocity (5, 0, 0) meters / second, attitude angle (2, 3, 30) degrees, and angular velocity (0, 0, 2) degrees / second. During the state prediction phase, using a prediction step of 0.02 seconds, the calculated prior states were: position (100.1, 200, 50) meters, velocity (4.8, 0.2, 0) meters / second, and attitude angle (2, 3.1, 30.04) degrees.

[0079] Comparing the actual observed values ​​with the predicted values ​​revealed observation residuals of 0.3 meters for position, 0.2 meters per second for velocity, and 0.2 degrees for attitude angle. The calculated Kalman gain ensures that 75% of the position estimate is based on observations and 25% on predictions; while the attitude estimate is based on 60% of observations and 40% on predictions. After fusion, position accuracy improves to ±0.5 meters, and attitude accuracy improves to ±0.05 degrees. Based on the fused data, the error trend over the next two seconds is predicted, and the control objective function is set with a weight of 0.5 for position error, 0.3 for velocity error, and 0.2 for attitude error. Constraints include a maximum velocity of 20 meters per second, a maximum acceleration of 4 meters per square second, and a maximum attitude angle change rate of 60 degrees per second. A 0.1-second control cycle is used for time domain partitioning, generating 20 control points. The resulting compensation control instruction set maintains a trajectory tracking error within 0.8 meters and attitude stability better than 0.5 degrees.

[0080] In an optional embodiment, the process of executing step S104 may specifically include the following steps:

[0081] (1) The wind speed data collected by the airborne sensor is decomposed into components to obtain three-dimensional wind speed components, and the three-dimensional wind speed components are interpolated to obtain continuous wind field data;

[0082] (2) Perform density analysis on the continuous wind field data to obtain the wind field density distribution, and perform gradient calculation on the wind field density distribution to obtain the wind field change characteristics;

[0083] (3) Correlation processing is performed on the wind field change characteristics and the compensation control instruction set to obtain the wind field influence coefficient, and threshold division processing is performed on the wind field influence coefficient to obtain the wind field compensation model;

[0084] (4) Quantify the change amplitude in the wind farm compensation model to obtain a route cost matrix, and perform path search on the route cost matrix to obtain an initial route set;

[0085] (5) Calculate the energy consumption of the initial route set to obtain energy consumption distribution data, and sort the energy consumption distribution data to obtain a route sequence;

[0086] (6) The route sequence is traversed to obtain an alternative route group, and the alternative route group is smoothed to obtain adaptive route data.

[0087] Optionally, the wind speed data collected by the airborne sensor is decomposed into components. The airborne sensor includes an anemometer and a barometer, which measure the wind speed magnitude and direction data. The wind speed data is decomposed into an east component, a north component and a vertical component to form a three-dimensional wind speed component. The three-dimensional wind speed components are processed by cubic spline interpolation to expand the discrete measurement points into continuous wind field data with an interpolation interval of 10 meters to ensure the continuity and smoothness of the wind field data. Density analysis is performed on the continuous wind field data to calculate the wind speed change within the unit space to obtain the wind field density distribution. The wind field density distribution reflects the concentration degree and change trend of the wind field. The wind field density distribution data is gradient calculated to obtain the wind field change characteristics. The calculation formula for the wind field change characteristics is:

[0088]

[0089] Among them, G wind represents the wind field variation characteristics, λ i Represents the weight coefficient of each direction, V i Represents the wind speed components in three directions, ρ air Indicates the air density.

[0090] The wind field variation characteristics are correlated with the compensation control instruction set to calculate the wind field impact coefficient. The wind field impact coefficient indicates the degree of influence of the wind field on flight control. The wind field impact coefficient is then divided into thresholds and multiple risk levels are set to form a wind field compensation model. The wind field compensation model is used to guide route planning and flight control. The magnitude of the change in the wind field compensation model is quantitatively analyzed to establish a route cost matrix. The route cost matrix considers factors such as wind field intensity, change gradient, and flight direction. An A* algorithm path search is performed on the route cost matrix. The search process takes into account the starting point, end point, and obstacle avoidance requirements to obtain an initial set of routes that meet the surveying and mapping requirements.

[0091] Calculate the energy consumption of each route in the initial route set to obtain energy consumption distribution data. The formula for energy consumption calculation is:

[0092]

[0093] Among them, E total represents the total energy consumption, m represents the mass of the drone, g represents the acceleration of gravity, h j Indicates the altitude change of the flight segment, v j represents the segment speed, ξ j Indicates the drag coefficient, F drag represents resistance, d j represents the flight distance, η j represents the energy conversion efficiency, and n represents the number of flight segments.

[0094] The energy consumption distribution data is sorted and arranged from low to high energy consumption to create a route sequence. This route sequence is then traversed to select route combinations that meet the surveying and mapping requirements and have low energy consumption to form a candidate route group. This candidate route group is then smoothed using cubic spline processing to eliminate sudden changes at route inflection points, ultimately generating adaptive route data.

[0095] For example, the survey area covers 5 square kilometers with an altitude difference of 300 meters. Airborne anemometers collect wind speed data at a frequency of 5 Hz, with raw data showing wind speeds ranging from 3 to 12 meters per second. The wind speed data is decomposed into three directional components: an average easterly wind speed of 6 meters per second, an average northerly wind speed of 4 meters per second, and an average vertical wind speed of 1 meter per second. Using cubic spline interpolation, continuous wind field data is generated for a 100-meter x 100-meter grid. Density analysis reveals three major wind field concentrations within the survey area, located at ridges and valleys, with wind speed gradients reaching a maximum of 0.2 meters per second per meter. Calculations of wind field variation characteristics indicate that wind field variations are most pronounced at ridges, with an influence coefficient of 0.8, requiring special consideration. The wind field compensation model divides the survey area into four risk levels, with high-risk areas accounting for 15% of the total area. The route cost matrix constructed based on the wind field compensation model is 500 x 500, and the A* algorithm searches for 12 feasible initial routes. Energy consumption calculations showed that these routes consumed between 800 and 1200 watt-hours. After sorting the routes, the top five were selected as alternatives. After smoothing, the optimal route had a total length of 15 kilometers, an estimated flight time of 45 minutes, and an energy consumption of 950 watt-hours, meeting the mapping accuracy and coverage requirements.

[0096] In an optional embodiment, the process of executing step S105 may specifically include the following steps:

[0097] (1) performing data normalization processing on the voltage, current and temperature data in the battery status data to obtain standardized battery parameters, and performing feature extraction processing on the standardized battery parameters to obtain battery status features;

[0098] (2) Correlation processing is performed on the battery status characteristics and the adaptive route data to obtain route energy consumption data, and statistical analysis and processing are performed on the route energy consumption data to obtain a unit energy consumption index;

[0099] (3) Perform interval segmentation processing on the task completion data to obtain segmented task data, and perform weight calculation processing on the segmented task data to obtain the task weight coefficient;

[0100] (4) Comprehensively calculate and process the unit energy consumption index and the task weight coefficient to obtain weighted energy consumption data, and perform threshold analysis on the weighted energy consumption data to obtain task optimization parameters;

[0101] (5) performing sequential arrangement processing on the task optimization parameters to obtain a task sequence table, and performing time allocation processing on the task sequence table to obtain a time sequence task table;

[0102] (6) Perform task division processing on the sequential task table to obtain task subsets, and perform priority allocation processing on the task subsets to obtain a task execution instruction set.

[0103] Optionally, battery status data includes three parameters: voltage, current, and temperature. The voltage range is 20-25V, the current range is 0-40A, and the temperature range is 10-50°C. Data normalization uses the maximum-minimum method to normalize all parameters to a range of 0-1. After normalization, the battery parameters undergo feature extraction, including calculating statistical features such as mean, variance, and rate of change, to obtain battery status characteristics. Feature extraction highlights key indicators of the battery's operating status, such as remaining capacity, discharge rate, and temperature stability. Correlation analysis is performed between the battery status characteristics and the adaptive route data to calculate energy consumption for each route segment, generating route energy consumption data. Route energy consumption data reflects energy consumption during different flight phases, including climb, cruise, and turns. Statistical analysis of the route energy consumption data calculates energy consumption per unit distance and per unit time to generate a unit energy consumption index. This unit energy consumption index reflects energy utilization efficiency and provides an important basis for mission planning.

[0104] Task completion data is segmented and divided into different completion stages based on indicators such as survey area coverage and data collection quality. The segmented task data includes the work content, completion status, and remaining workload for each stage. Weighted task data is weighted, taking into account factors such as task importance, time urgency, and resource consumption, to generate task weight coefficients. These task weight coefficients are used to guide resource allocation and task prioritization. Unit energy consumption indicators and task weight coefficients are combined and weighted summed to generate weighted energy consumption data. Weighted energy consumption data comprehensively considers energy consumption and task value, providing a basis for decision-making in task optimization. Threshold analysis is performed on the weighted energy consumption data, setting multiple energy consumption levels to generate task optimization parameters. These task optimization parameters guide the rational allocation of tasks and the scheduling of their execution order. The task optimization parameters are then sorted in descending order of priority to generate a task sequence table. The task sequence table includes the execution order, estimated duration, and resource requirements for each task. Time allocation is performed on the task sequence table, determining the execution time window for each task based on the drone's endurance and task urgency, to generate a time-series task table. The time-series task table provides a detailed task execution plan.

[0105] The timed task list is partitioned, breaking down large tasks into independently executable subtasks to create task subsets. Task subsets facilitate multi-machine collaborative execution and dynamic task adjustment. Priorities are assigned to the task subsets, taking into account task dependencies and resource constraints, to generate a task execution instruction set. This task execution instruction set provides clear operational guidance for the drone. For example, consider a drone powered by a lithium battery performing a mission. The initial battery status displays: voltage 24.5V, current 5A, and temperature 25°C. After data normalization, the voltage standard values ​​are 0.9, current standard values ​​0.125, and temperature standard values ​​0.375. Feature extraction indicates that the battery capacity remains at 85%, and the temperature change rate is less than 0.5°C / minute, indicating good operating condition.

[0106] The adaptive route is 20 kilometers long and divided into 25 segments. Analysis of route energy consumption data shows an average power consumption of 380W, a peak power consumption of 550W during the climb phase, and a power consumption of 320W during the cruise phase. The calculated unit energy consumption index is 70 watt-hours per kilometer and 1150 watt-hours per hour. Mission completion data indicates that 40% of the survey area has been covered. The segmentation results in three mission phases: rolling terrain (weight 0.4), flat terrain (weight 0.3), and transitional terrain (weight 0.3). Weighted energy consumption calculations show that the unit area energy consumption is 1.2 watt-hours per square meter in the rolling terrain phase, 0.8 watt-hours per square meter in the flat terrain phase, and 1.0 watt-hours per square meter in the transitional terrain phase. Threshold analysis categorizes missions into three energy consumption levels: high, medium, and low. The mission sequence is arranged in the order of "high energy consumption - low energy consumption - medium energy consumption," with time allocation prioritizing the execution of high-energy-consuming missions during the cooler morning hours.

[0107] The final mission execution instruction set included 15 subtasks, with priority assigned based on remaining battery power, weather conditions, and mission urgency. High-priority tasks were scheduled during periods of sufficient battery capacity (over 70% remaining), with a 20% battery margin reserved to ensure safe mission completion. The entire surveying and mapping mission was estimated to take three hours and consume 3,500 watt-hours of energy, meeting the survey area coverage and data quality requirements.

[0108] In an optional embodiment, the process of executing step S106 may specifically include the following steps:

[0109] (1) Perform data synchronization processing on the position and velocity data of multiple UAVs to obtain synchronization status data, and then associate the synchronization status data with the task execution instruction set to obtain the initial coordination parameters;

[0110] (2) Performing communication topology construction processing on the initial coordination parameters to obtain communication link data, and performing bandwidth allocation processing on the communication link data to obtain a resource allocation table;

[0111] (3) performing time slot division processing on the resource allocation table to obtain a communication timing table, and performing conflict detection processing on the communication timing table to obtain a set of available time slots;

[0112] (4) Performing information interaction processing on the data in the available time slot set to obtain interaction status data, and performing consistency verification processing on the interaction status data to obtain collaborative control data;

[0113] (5) Perform distance calculation on the position information in the collaborative control data to obtain a relative distance matrix, and then perform safety domain construction on the relative distance matrix to obtain collision risk data;

[0114] (6) Classify the collision risk data to obtain a risk level table, and perform trajectory planning on the risk level table to obtain a set of alternative tracks;

[0115] (7) Perform interference assessment on the alternative trajectory set to obtain the multi-machine collaborative control instruction set.

[0116] Optionally, implementing multi-drone coordinated control first requires data synchronization. Each drone's position and velocity data is collected at a different frequency: position data is acquired via GPS at a frequency of 10 Hz, and velocity data is measured via a pitot tube at a frequency of 20 Hz. Data synchronization utilizes a time window method, using a 50-millisecond synchronization window. Data within this window is linearly interpolated to generate synchronized state data based on a unified time base. This synchronized state data is then correlated with the mission execution instruction set to extract mission information, position information, and velocity information for each drone, forming the initial coordination parameters. A communication topology is constructed based on these initial coordination parameters, and a minimum spanning tree algorithm is used to establish communication connections between the drones. This communication topology considers inter-drone distance and communication quality to optimize the layout of communication links. Bandwidth allocation is performed on communication link data. Based on data transmission requirements and link quality, appropriate bandwidth resources are allocated to each communication link, generating a resource allocation table. Dynamic bandwidth allocation is implemented to ensure the timely transmission of critical data.

[0117] The resource allocation table is divided into time slots, and communications are organized using TDMA (Time Division Multiple Access). The time slot length is set to 10 milliseconds, and specific data transmission tasks are scheduled within each time slot to generate a communication schedule. The communication schedule is checked for conflicts to see if the same time slot is occupied by multiple communication tasks. Conflict detection uses a sliding window method with a window size of 100 milliseconds. Detected conflicts are adjusted to obtain a set of available time slots. Information exchange is performed on the data in the set of available time slots, including the exchange of information such as position, speed, and task status. A distributed consistency protocol is used for information exchange to ensure the synchronization and consistency of information between multiple machines. The interaction status data is verified for consistency, checking the timeliness and reliability of the data, eliminating abnormal data, and obtaining collaborative control data. The collaborative control data contains verified global state information, providing a foundation for subsequent multi-machine collaboration.

[0118] Distance calculations are performed on the location information in the collaborative control data, using Euclidean distance to calculate the relative distances between UAVs, forming a relative distance matrix. This relative distance matrix reflects the spatial distribution of the UAV swarm. A safety domain is constructed based on the relative distance matrix, with a minimum safety distance threshold set. The collision risk between UAVs is analyzed to generate collision risk data. The safety domain dynamically adjusts the safety distance requirements by taking into account the UAV's motion characteristics and environmental factors. The collision risk data is classified into three levels: high, medium, and low, based on relative distance, relative speed, and environmental conditions. A risk level table is generated. Trajectory planning is performed based on the risk level table, using the artificial potential field method to generate obstacle avoidance trajectories and form a set of candidate trajectories. The artificial potential field method treats other UAVs as sources of repulsion and the target point as a source of attraction, generating safe trajectories through force field superposition.

[0119] Interference assessment is performed on the candidate trajectory sets, analyzing the mutual impact of the trajectories and the effectiveness of mission execution. After comprehensive evaluation, a multi-aircraft collaborative control instruction set is generated. The multi-aircraft collaborative control instruction set contains information such as trajectory planning, speed control, and task allocation for each UAV.

[0120] For example, three drones are dispatched simultaneously to perform a mission. Their initial positions are (0, 0, 100) meters, (200, 0, 100) meters, and (100, 200, 100) meters, respectively, with a flight speed of 15 meters per second. During data synchronization, data within a 50-millisecond time window is linearly interpolated to a 20-Hz frequency. The communication topology is constructed using a minimum spanning tree algorithm, establishing three primary communication links with a total bandwidth of 10 Mbps, allocated across the three links in a 4:3:3 ratio. Time slot allocation divides the 100-millisecond communication cycle into 10 time slots, each 10 milliseconds. Conflict detection identifies two time slot conflicts, which are resolved by adjusting the timing, resulting in eight available time slots. During information exchange, position data occupies four time slots, velocity data occupies two time slots, and mission status data occupies two time slots.

[0121] Relative distance calculations showed that the minimum distance between the three drones was 200 meters and the maximum distance was 282 meters. The safety distance threshold was set at 100 meters, and collision risk analysis showed that all relative distances were within a safe range. During trajectory planning, taking into account that one drone encountered a crosswind of 5 meters per second, its trajectory was adjusted, deviating 30 meters from the original route to ensure flight safety. The final generated multi-machine collaborative control instruction set assigned different mapping areas to the three drones. The working areas were spaced 150 meters apart, the flight altitude difference was set to 20 meters, and an interlaced "S"-shaped mapping route was adopted to avoid mutual interference. The entire mapping task was completed within 45 minutes, with a data transmission success rate of 99.5% and a position control accuracy of better than ±2 meters, meeting the accuracy requirements of geological mapping.

[0122] The above describes the geological mapping method based on drone in the embodiment of the present application. The following describes the geological mapping system based on drone in the embodiment of the present application. Figure 2 In the embodiments of the present application, an embodiment of a geological mapping system based on a drone includes:

[0123] The processing module is used to perform fuzzy controller processing on the terrain feature data of the survey area and the surveying and mapping parameter requirements to obtain an initial control parameter group and a controller gain matrix;

[0124] A control module is used to process the digital elevation model of the survey area and the initial control parameter group through a partition algorithm to obtain partition control data, and to perform closed-loop control processing on the partition control data and the controller gain matrix through a cascade PID algorithm to obtain a posture control instruction set;

[0125] A fusion module is used to fuse the position data collected by multiple sensors and the attitude control instruction set through an extended Kalman filter algorithm to obtain attitude fusion data, and to compensate the attitude fusion data through a model predictive control algorithm to obtain a compensation control instruction set;

[0126] an optimization module for processing the real-time wind field data collected by the airborne sensor and the compensation control instruction set through a three-dimensional modeling algorithm to obtain a wind field compensation model, and optimizing the wind field compensation model through a dynamic programming algorithm to obtain adaptive route data;

[0127] An analysis module is used to analyze and process the UAV battery status data, the adaptive route data, and the mission completion data using an energy consumption prediction model to obtain mission optimization parameters, and dynamically allocate the mission optimization parameters using a mission planning algorithm to obtain a mission execution instruction set;

[0128] The computing module is used to process the status information of multiple drones and the task execution instruction set through a distributed collaborative algorithm to obtain collaborative control data, and to perform real-time computation and processing on the collaborative control data through a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

[0129] Through the collaborative efforts of these components, a fuzzy controller processes the survey area's terrain feature data and mapping parameter requirements, enabling intelligent generation of initial control parameters for complex terrain conditions and improving the accuracy and adaptability of parameter settings. A partitioning algorithm and a cascaded PID algorithm are used to process the survey area's digital elevation model and initial control parameter set, enabling precise control of regions with different terrain features and enhancing the UAV's flight stability in complex terrain. An extended Kalman filter algorithm is used to fuse position data collected by multiple sensors and the attitude control instruction set, combined with a model predictive control algorithm for compensation. This improves the accuracy of position and attitude estimation and enhances the robustness of the control system. Real-time wind field data collected by onboard sensors and the compensation control instruction set are processed using a three-dimensional modeling algorithm and a dynamic programming algorithm, enabling precise compensation for wind field effects and real-time flight path optimization, thereby enhancing the reliability of surveying and mapping missions in complex wind field conditions. An energy consumption prediction model and a mission planning algorithm are used to analyze and process the UAV's battery status data, adaptive route data, and mission completion data, optimizing energy efficiency and rationally planning mission execution. By processing the status information and task execution instruction sets of multiple drones through a distributed collaborative algorithm and a safe distance field algorithm, this method achieves intelligent task allocation and collision risk control during multi-drone collaboration, improving the efficiency and safety of multi-drone collaborative mapping. Overall, the proposed method significantly enhances the adaptability, reliability, and efficiency of UAV geological mapping in complex environments through multi-level closed-loop control and multi-algorithm collaboration.

[0130] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the UAV-based geological mapping method.

[0131] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0132] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A geological mapping method based on drone, characterized in that: The UAV-based geological mapping method includes: The terrain feature data of the survey area and the surveying and mapping parameter requirements are processed by a fuzzy controller to obtain an initial control parameter group and a controller gain matrix, including: performing gradient calculation processing on the elevation data in the terrain feature data of the survey area to obtain terrain slope data, and performing interval division processing on the terrain slope data to obtain terrain complexity classification data; quantizing the spatial resolution data in the surveying and mapping parameter requirements to obtain reference flight altitude data, and correlating the reference flight altitude data with the terrain complexity classification data to obtain an altitude adjustment coefficient; performing analytical processing on the overlap data in the surveying and mapping parameter requirements to obtain a speed constraint condition. and performing a range-defining process on the flight speed according to the speed constraint condition to obtain speed interval data; performing a fuzzy rule generation process on the speed interval data and the altitude adjustment coefficient to obtain a control rule base, and performing a membership function construction process on the control rule base to obtain a fuzzy inference set; performing a defuzzification process on the fuzzy inference set to obtain original control parameters, and performing a balance constraint process on the original control parameters to obtain an initial control parameter group; performing a sensitivity analysis process on the parameters in the initial control parameter group to obtain parameter influence weights, and generating a controller gain matrix according to the parameter influence weights; The digital elevation model of the survey area and the initial control parameter group are processed by a partition algorithm to obtain partition control data, and the partition control data and the controller gain matrix are closed-loop controlled by a cascade PID algorithm to obtain a posture control instruction set; The position data collected by the multi-sensor and the attitude control instruction set are fused by an extended Kalman filter algorithm to obtain attitude fusion data, and the attitude fusion data are compensated by a model predictive control algorithm to obtain a compensation control instruction set; The real-time wind field data collected by the airborne sensor and the compensation control instruction set are processed by a three-dimensional modeling algorithm to obtain a wind field compensation model, and the wind field compensation model is optimized by a dynamic programming algorithm to obtain adaptive route data; The UAV battery status data, the adaptive route data, and the mission completion data are analyzed and processed using an energy consumption prediction model to obtain mission optimization parameters, and the mission optimization parameters are dynamically allocated and processed using a mission planning algorithm to obtain a mission execution instruction set; The status information of multiple UAVs and the task execution instruction set are processed by a distributed collaborative algorithm to obtain collaborative control data, and the collaborative control data is calculated and processed in real time by a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

2. The geological surveying method based on UAV according to claim 1, characterized in that: The digital elevation model of the measurement area and the initial control parameter group are processed by a partition algorithm to obtain partition control data, and the partition control data and the controller gain matrix are closed-loop controlled by a cascade PID algorithm to obtain a posture control instruction set, including: Performing mean filtering on the elevation data in the digital elevation model of the survey area to obtain smoothed elevation data, and performing slope calculation on the smoothed elevation data to obtain terrain slope distribution data; Performing cluster analysis on the terrain slope distribution data to obtain terrain feature partitions, and performing boundary extraction on the terrain feature partitions to obtain partition boundary data; Performing parameter matching processing on the partition boundary data and the initial control parameter group to obtain a partition parameter set, and performing interpolation processing on the partition parameter set to obtain a continuous parameter field; performing boundary constraint processing on the continuous parameter field to obtain constraint parameter data, and generating partition control data according to the constraint parameter data; performing error calculation processing on the partition control data to obtain position error data, and performing differentiation processing on the position error data to obtain speed error data; Integrating the speed error data to obtain integrated error data, and performing proportional operation on the integrated error data according to the controller gain matrix to obtain a basic control amount; The basic control amount is subjected to a limiting process to obtain a bounded control amount, and the bounded control amount is subjected to an attitude solution process to obtain an attitude control instruction set.

3. The geological surveying method based on UAV according to claim 1, characterized in that: The position data collected by the multi-sensor and the attitude control instruction set are fused by an extended Kalman filter algorithm to obtain attitude fusion data, and the attitude fusion data are compensated by a model predictive control algorithm to obtain a compensation control instruction set, including: Performing timestamp alignment processing on position data collected by multiple sensors to obtain synchronized position data, and performing state vector construction processing on the synchronized position data and the attitude control instruction set to obtain an initial state estimate; Performing state prediction processing on the initial state estimate to obtain a priori state quantity, and performing covariance transfer processing on the priori state quantity to obtain a priori covariance; Performing observation prediction processing on the priori state quantity to obtain a predicted observation value, and performing residual calculation processing on the predicted observation value and the actual observation value to obtain an observation residual; Performing gain calculation processing on the observation residual and the prior covariance to obtain a Kalman gain, and performing state update processing on the Kalman gain to obtain pose fusion data; Performing error prediction processing on the posture fusion data to obtain a prediction error sequence, and performing cost function construction processing on the prediction error sequence to obtain a control objective function; Performing constraint condition setting processing on the control objective function to obtain a constraint condition set, and performing rolling time domain division processing on the constraint condition set to obtain a time domain control interval; A sequence prediction process is performed on the state within the time domain control interval to obtain a predicted control amount, and a feedback correction process is performed on the predicted control amount to obtain a compensation control instruction set.

4. The method for geological surveying and mapping based on an unmanned aerial vehicle according to claim 1, characterized in that: The real-time wind field data collected by the airborne sensor and the compensation control instruction set are processed by a three-dimensional modeling algorithm to obtain a wind field compensation model, and the wind field compensation model is optimized by a dynamic programming algorithm to obtain adaptive route data, including: Performing component decomposition processing on wind speed data collected by the airborne sensor to obtain three-dimensional wind speed components, and performing interpolation processing on the three-dimensional wind speed components to obtain continuous wind field data; Performing density analysis on the continuous wind field data to obtain wind field density distribution, and performing gradient calculation on the wind field density distribution to obtain wind field change characteristics; performing correlation processing on the wind field change characteristics and the compensation control instruction set to obtain a wind field influence coefficient, and performing threshold division processing on the wind field influence coefficient to obtain a wind field compensation model; quantifying the change amplitude in the wind farm compensation model to obtain a route cost matrix, and performing path search processing on the route cost matrix to obtain an initial route set; performing energy consumption calculation processing on the initial route set to obtain energy consumption distribution data, and sorting the energy consumption distribution data to obtain a route sequence; The route sequence is traversed to obtain an alternative route group, and the alternative route group is smoothed to obtain adaptive route data.

5. The geological mapping method based on drone according to claim 1, characterized in that: The UAV battery status data, the adaptive route data, and the mission completion data are analyzed and processed using an energy consumption prediction model to obtain mission optimization parameters, and the mission optimization parameters are dynamically allocated and processed using a mission planning algorithm to obtain a mission execution instruction set, including: performing data normalization processing on the voltage, current, and temperature data in the battery status data to obtain standardized battery parameters, and performing feature extraction processing on the standardized battery parameters to obtain battery status features; Correlating the battery status characteristics with the adaptive route data to obtain route energy consumption data, and performing statistical analysis on the route energy consumption data to obtain a unit energy consumption index; Performing interval segmentation processing on the task completion data to obtain segmented task data, and performing weight calculation processing on the segmented task data to obtain a task weight coefficient; Performing comprehensive calculation processing on the unit energy consumption index and the task weight coefficient to obtain weighted energy consumption data, and performing threshold analysis processing on the weighted energy consumption data to obtain task optimization parameters; Performing sequential arrangement processing on the task optimization parameters to obtain a task sequence table, and performing time allocation processing on the task sequence table to obtain a time sequence task table; The sequential task table is processed by task division to obtain task subsets, and the task subsets are processed by priority assignment to obtain a task execution instruction set.

6. The geological surveying method based on UAV according to claim 1, characterized in that: The state information of the multiple UAVs and the task execution instruction set are processed by a distributed collaborative algorithm to obtain collaborative control data, and the collaborative control data are processed in real time by a safe distance field algorithm to obtain a multi-machine collaborative control instruction set, including: Performing data synchronization processing on the position and speed data of the multiple UAVs to obtain synchronization state data, and performing association processing on the synchronization state data and the task execution instruction set to obtain initial coordination parameters; Performing communication topology construction processing on the initial coordination parameters to obtain communication link data, and performing bandwidth allocation processing on the communication link data to obtain a resource allocation table; Performing time slot division processing on the resource allocation table to obtain a communication timing table, and performing conflict detection processing on the communication timing table to obtain an available time slot set; Performing information interaction processing on the data in the available time slot set to obtain interaction state data, and performing consistency verification processing on the interaction state data to obtain collaborative control data; performing distance calculation processing on the position information in the collaborative control data to obtain a relative distance matrix, and performing safety domain construction processing on the relative distance matrix to obtain collision risk data; performing a classification process on the collision risk data to obtain a risk level table, and performing a track planning process on the risk level table to obtain an alternative track set; Interference assessment processing is performed on the alternative track set to obtain a multi-machine cooperative control instruction set.

7. A geological mapping system based on a drone, used to implement the geological mapping method based on a drone according to any one of claims 1 to 6, characterized in that: The UAV-based geological mapping system includes: The processing module is used to perform fuzzy controller processing on the terrain feature data of the survey area and the surveying and mapping parameter requirements to obtain the initial control parameter group and the controller gain matrix, including: performing gradient calculation processing on the elevation data in the terrain feature data of the survey area to obtain terrain slope data, and performing interval division processing on the terrain slope data to obtain terrain complexity classification data; performing quantization processing on the spatial resolution data in the surveying and mapping parameter requirements to obtain reference flight altitude data, and performing correlation processing on the reference flight altitude data and the terrain complexity classification data to obtain altitude adjustment coefficient; performing analytical processing on the overlap data in the surveying and mapping parameter requirements to obtain speed speed constraint conditions, and based on the speed constraint conditions, the flight speed is range-defined to obtain speed interval data; the speed interval data and the altitude adjustment coefficient are subjected to fuzzy rule generation processing to obtain a control rule base, and the control rule base is subjected to membership function construction processing to obtain a fuzzy inference set; the fuzzy inference set is defuzzified to obtain original control parameters, and the original control parameters are subjected to equilibrium constraint processing to obtain an initial control parameter group; the parameters in the initial control parameter group are subjected to sensitivity analysis processing to obtain parameter influence weights, and a controller gain matrix is ​​generated based on the parameter influence weights; A control module is used to process the digital elevation model of the survey area and the initial control parameter group through a partition algorithm to obtain partition control data, and to perform closed-loop control processing on the partition control data and the controller gain matrix through a cascade PID algorithm to obtain a posture control instruction set; A fusion module is used to fuse the position data collected by multiple sensors and the attitude control instruction set through an extended Kalman filter algorithm to obtain attitude fusion data, and to compensate the attitude fusion data through a model predictive control algorithm to obtain a compensation control instruction set; an optimization module for processing the real-time wind field data collected by the airborne sensor and the compensation control instruction set through a three-dimensional modeling algorithm to obtain a wind field compensation model, and optimizing the wind field compensation model through a dynamic programming algorithm to obtain adaptive route data; An analysis module is used to analyze and process the UAV battery status data, the adaptive route data, and the mission completion data using an energy consumption prediction model to obtain mission optimization parameters, and dynamically allocate the mission optimization parameters using a mission planning algorithm to obtain a mission execution instruction set; The computing module is used to process the status information of multiple drones and the task execution instruction set through a distributed collaborative algorithm to obtain collaborative control data, and to perform real-time computation and processing on the collaborative control data through a safe distance field algorithm to obtain a multi-machine collaborative control instruction set.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the UAV-based geological mapping method according to any one of claims 1 to 6 is implemented.

Images