Reservoir bank slope inspection method and system based on unmanned aerial vehicle laser radar

Abstract

The invention discloses a reservoir bank slope inspection method and system based on an unmanned aerial vehicle laser radar, and relates to the technical field of unmanned aerial vehicle reservoir bank slope inspection, and the method comprises the following steps: 1, guaranteeing the normal working state of an unmanned aerial vehicle, starting the laser radar, connecting a ground GNSS reference station on the shore, calibrating the position of the unmanned aerial vehicle and a communication signal, an inspection area boundary and flight parameters are set; navigation data provided by a shoreside base station is utilized. According to the reservoir bank slope inspection method and system based on the unmanned aerial vehicle laser radar, the inspection efficiency is improved, and the endurance time is prolonged; the system is high in autonomy, reduces the labor cost, achieves the whole-course autonomous navigation, path optimization, obstacle avoidance and dynamic adjustment, does not need manual intervention, effectively reduces the workload and operation difficulty of inspection personnel, reduces the labor cost, improves the intelligent level of inspection tasks, and improves the autonomous adaptability of the unmanned aerial vehicle to complex terrains.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) reservoir bank slope inspection, and in particular to a reservoir bank slope inspection method and system based on unmanned aerial vehicle (UAV) laser radar. Background Art

[0002] After decades of rapid development, many large-scale dams, inter-basin water diversion projects, pumped storage power stations, etc. have been built in many major river basins. my country has become the country with the most water conservancy and hydropower projects in the world. Its construction scale has continuously broken world records, forming a number of representative large reservoirs and high dams. The number is extremely large, most of which were built in the early 21st century and have been stored and operated for more than 20 years. Especially in the western mountainous areas of my country, the large reservoirs formed on the rivers with obvious terrain conditions have steep slopes, large terrain drops, dense vegetation, extremely poor traffic conditions and changeable climate. After the reservoir is filled with water, the natural geological conditions of the reservoir area and the downstream slopes will change accordingly, which may cause a series of environmental geological problems such as collapse and landslides. In addition, during the operation of the reservoir, the reservoir slopes are long-term affected by the environment, water level rise and fall, scouring, soaking and rock mass strength weakening, and their morphology, geological structure, unfavorable structural surface combination, etc. make the stability worse, which is easy to cause a series of geological disasters such as collapse and landslides, which seriously threaten the safety of reservoir operation. For this reason, the safety of reservoir slopes has become a key focus of the water conservancy and hydropower industry. Reservoir landslides and bank collapses will become increasingly prominent. It is particularly important to identify the huge number of reservoir slope safety issues in advance, especially when mountain roads are blocked, to quickly and accurately discover the location and scale of disasters.

[0003] Traditional inspections of reservoir banks usually rely on manual observation or ground equipment inspections. However, with the rapid development and continuous improvement of UAV laser radars in recent years, they can quickly, non-contactly, obtain high-precision and high-density three-dimensional spatial information on the surface of alpine canyon areas, and have become an indispensable means of data acquisition. It has become popular all over the world because of its multiple advantages. It has inherent advantages in geological disaster investigations and safety inspections of large reservoir banks in alpine canyon areas, and has played a huge role, making great improvements in reservoir bank inspections.

[0004] However, there are still some urgent problems to be solved in the application of UAV lidar in reservoir slope inspection: due to the complex and changeable terrain of the reservoir slope, and the existence of steep slopes, cliffs, gullies and other terrains, the flight control and navigation of the UAV still have certain difficulties. In other words, in this case, higher requirements are placed on the flight control and navigation of the UAV. Not only do they need to have good maneuverability and stability, but they also need to have the ability to fly autonomously in complex terrain environments to avoid accidents such as collisions with obstacles and explosions. It is urgent to improve the autonomous adaptability of UAVs to complex terrain. Summary of the invention

[0005] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: a reservoir slope inspection method and system based on unmanned aerial vehicle laser radar, comprising the following steps:

[0006] Step 1: Ensure that the UAV is in normal working condition, start the LiDAR, connect to the ground GNSS base station on the shore, calibrate the UAV position and communication signal, and set the inspection area boundary and flight parameters; use the navigation data provided by the shore base station and the terrain data initially obtained by the LiDAR to generate the initial inspection path of the UAV; avoid dangerous areas during path design to ensure that the entire bank slope is covered;

[0007] Step 2: Start inspection according to the path planning, dynamically locate its own position and direction based on radio signals and terrain data, and continuously collect three-dimensional terrain data of the slope during flight to update the position and flight path in real time;

[0008] Step 3: During the inspection process, the drone identifies obstacles or dangerous terrain ahead based on terrain data obtained by the LiDAR and automatically adjusts the flight altitude and direction. If it encounters complex terrain, it replans the route to ensure the continuity and safety of the inspection.

[0009] Step 4: After the inspection is completed, the system conducts a comprehensive analysis of all collected data, extracts the structural characteristics of the slope, and identifies high-risk areas prone to landslides and collapses, and finally generates a complete inspection report; after the drone completes the inspection and returns, the system resets to prepare for the next inspection, and stores the analysis data for further research or maintenance reference.

[0010] A reservoir bank slope inspection system based on unmanned aerial vehicle laser radar comprises an unmanned aerial vehicle platform, a radio orientation module, a radio navigation module, a laser radar sensor, an obstacle avoidance module and a path planning module, wherein: the unmanned aerial vehicle platform is equipped with a radio orientation module and a radio navigation module, and performs position calibration by receiving radio signals sent by a radio reference station set up on the shore, so as to optimize the flight path in real time; the laser radar sensor is used to collect three-dimensional data of the reservoir bank slope to obtain terrain data; the obstacle avoidance module combines the radio navigation signal and the laser radar data to analyze the terrain around the unmanned aerial vehicle in real time to achieve accurate obstacle avoidance; the path planning module is used to generate a dynamic inspection route of the unmanned aerial vehicle through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module, so as to ensure the stable and autonomous flight of the unmanned aerial vehicle in complex terrain;

[0011] Among them, the radio direction-finding module is used to automatically identify the radio reference station set up on the shore, and perform triangulation calculation by measuring the radio signal strength, arrival time difference and frequency offset of different radio reference stations, calculate the position information of the UAV relative to the radio reference station, and obtain positioning data; the radio direction-finding module includes a signal enhancement unit and an environmental adaptation unit; the signal enhancement unit is used to analyze the influence of different working conditions on the accuracy of positioning data based on the UAV under different working conditions. For example, when far away from the reference station, the signal strength may weaken, resulting in a decrease in positioning accuracy; in complex terrain areas, the signal may be reflected, blocked, etc., affecting the signal The transmission and reception of signals affects the positioning results, and the orientation accuracy of the UAV is clarified and the orientation data is obtained by correcting and optimizing the positioning data, wherein the working conditions include being far away from the base station, signal interference, complex terrain and signal weak areas; the UAV is enabled to have stable positioning capabilities under various working conditions, providing a reliable basis for subsequent flight control and navigation; the environmental adaptation unit is used to identify changes in the signal environment based on the environmental adaptation algorithm, and adjust the communication frequency and transmission power of the radio orientation module to adapt to signal reflection and shielding in complex terrain, improve the effective ranging range of the radio orientation module, and enable the UAV to maintain a reliable navigation signal in obstacle areas such as cliffs and steep slopes.

[0012] Preferably, the process of adjusting the communication frequency and transmission power of the radio orientation module is as follows: the radio orientation module continuously collects communication-related data, the communication-related data includes the signal strength, signal frequency, and signal arrival time received from the shore radio reference station, and records the current position of the drone, the flight attitude, and some basic characteristics of the surrounding environment, such as the terrain type and whether it is close to obstacles. The collected communication-related data is pre-processed to remove noise interference and normalize the data to obtain communication processing data to improve the quality and availability of the data. The pre-processed communication processing data is analyzed using an environmental adaptation algorithm to identify changes in the signal environment. The environmental adaptation algorithm takes into account multiple factors, including fluctuations in signal strength, stability of signal frequency, and changes in signal arrival time. For example, if it is found that the signal strength drops sharply in a short period of time and the signal arrival time is significantly delayed, and the drone is in a complex terrain area, such as near a cliff or a steep slope, it may be judged that the signal environment has changed, which may be caused by terrain obstruction or signal reflection, and the analysis result of the signal environment change is obtained;

[0013] According to the analysis results, the communication frequency is adjusted: a frequency band list and a preset threshold are preset, and the frequency bands are sorted from small to large according to the interference degree. When the interference is detected to exceed the preset threshold, the next frequency band in the list is switched to communicate in turn;

[0014] Adjustment of transmit power: According to the distance between the UAV and the radio base station and the signal strength, when the UAV is far away from the base station, the signal strength weakens. To ensure reliable communication, the transmit power is increased. The increase in transmit power is calculated based on the distance. The distance-based power adjustment formula P is used. d =P0+kd, where P d is the adjusted transmit power, P0 is the initial transmit power, k is the power adjustment coefficient, and d is the distance between the UAV and the base station.

[0015] Preferably, the radio navigation module combines the terrain data of the lidar sensor and adopts a multi-path evaluation algorithm to dynamically identify the signal paths sent by different radio reference stations, obtain navigation data, and plan the optimal flight path for the UAV, and automatically bypass obstacles or dangerous areas through the optimal flight path; once a path obstruction is detected, the flight altitude or direction is immediately adjusted to enable the UAV to avoid steep slopes, cliffs and gullies in real time.

[0016] Preferably, the process of the radio navigation module obtaining navigation data and planning the optimal flight path of the UAV is as follows:

[0017] The radio navigation module continuously receives signals sent from different radio reference stations on the shore, including the location information of the reference stations and the signal sending time. At the same time, the module records the signal strength, frequency and related characteristics of the signal propagation process. For example, each reference station will attach its own coordinate information when sending a signal. After receiving these signals, the radio navigation module first decodes the signals and extracts the location data of the reference station and the signal sending timestamp. At the same time, the radio navigation module obtains the terrain data of the lidar sensor. The terrain data includes the height, slope, location and shape of the terrain, and pre-processes the terrain data, including removing noise points, classifying the point cloud data, and distinguishing different types of objects such as the ground, vegetation, and buildings, so that these data can be used more effectively for path planning in the future;

[0018] A multipath evaluation algorithm is used to dynamically identify the signal paths sent by different radio reference stations. Based on the reflection, refraction and obstruction encountered by the signal during propagation, the propagation path of the signal from the reference station to the UAV is analyzed to obtain the identification result. For example, by comparing the actual received signal strength with the strength in the theoretical absence of interference, it is determined whether the signal has been reflected or obstructed. If the signal strength is significantly lower than expected and there are obstacles or complex terrain around, it may mean that the signal has been reflected multiple times or has been partially obstructed. Based on the identification result of the signal path, the triangulation principle is used, combined with the location information of multiple reference stations and the signal arrival time difference, to calculate the signal path. Calculate the position of the drone in three-dimensional space and obtain the navigation data of the drone, wherein the navigation data includes the position, speed and direction of the drone. For example, the coordinates of the three reference stations are known to be (x1, y1, z1), (x2, y2, z2), and (x3, y3, z3), and the time when the signal arrives at the drone from the three reference stations is t1, t2, and t3, respectively. By calculating the distance difference of signal propagation, according to the signal speed and time difference, the position coordinates (x, y, z) of the drone are calculated using the triangulation formula. At the same time, the speed and flight direction of the drone are calculated according to the continuous position data to comprehensively obtain the navigation data of the drone.

[0019] Optimal flight path planning: Based on the terrain data provided by the lidar and the calculated navigation data, the environment around the drone is analyzed to identify obstacles, including steep slopes, cliffs, and gullies. For example, the location of steep slopes and cliffs is determined by analyzing the height change rate in the terrain data; obstacles are marked as dangerous areas based on their shape and location information. Based on the complexity of the terrain, the distribution of dangerous areas, and the flight performance limitations of the drone, such as the minimum turning radius, maximum climbing angle, or maximum descent angle, the Dijkstra algorithm is used to search in three-dimensional space for feasible paths from the current position of the drone to the target area, such as the area to be inspected on the slope of a reservoir. During the search process, different paths are evaluated, and the path cost is calculated based on the path length, the risk level of passing through dangerous areas, and the energy consumption of the flight. For example, a path passing through a dangerous area will be assigned a higher cost, and a longer path will also have a corresponding increase in cost. By searching and comparing the costs of different paths, the path with the lowest cost is found as the optimal path; and during the flight of the drone, environmental changes are continuously monitored, such as terrain changes, the emergence of new obstacles, and the drone's own status, such as position, speed, and power. If it is found that the currently planned flight path is no longer feasible or is no longer the optimal path, the path planning is immediately re-performed until the best flight path is obtained.

[0020] Preferably, the path planning module is used to generate a dynamic inspection route for the UAV through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module; during the generation of the dynamic inspection route, the path planning module uses a prediction algorithm to predict the complexity of the terrain to be encountered based on the current position, flight speed and terrain structure of the UAV, and then adjusts the flight parameters of the UAV in real time to obtain a dynamic inspection route, thereby ensuring the stability and efficiency of the UAV inspection route and preventing inspection interruptions due to terrain changes.

[0021] Preferably, the process of generating a dynamic inspection route of a drone is as follows:

[0022] The path planning module obtains orientation data from the radio orientation module. This data clarifies the position and orientation information of the UAV relative to the radio reference station, so that the path planning module can accurately grasp the position of the UAV in space. At the same time, it receives navigation data from the radio navigation module, which contains the precise position, speed, flight direction and surrounding environment information of the UAV. These data provide comprehensive basic information for subsequent path planning. The current position of the UAV is determined by using the acquired orientation and navigation data, and the terrain structure features around the position are extracted in combination with the terrain data provided by the lidar sensor. The terrain structure features include the degree of undulation of the terrain, such as slope and height difference, the type of terrain, such as mountains, plains, water edges, and whether there are obstacles, such as trees, buildings and cliffs;

[0023] Using a prediction algorithm, based on the current position, flight speed and direction of the drone, as well as the analyzed current terrain structure characteristics, the terrain complexity of the area that the drone will fly through in the future is predicted, and a prediction result is obtained;

[0024] According to the predicted results of terrain complexity, the flight parameters of the UAV are adjusted in real time. If a steep slope or complex terrain is predicted ahead, the flight speed of the UAV is reduced to ensure flight safety and stability. If it is necessary to cross a narrow valley or bypass a large obstacle, the flight altitude and direction of the UAV are adjusted to enable it to pass smoothly. At the same time, based on the endurance of the UAV, the battery power is reasonably allocated according to the terrain and mission requirements. The energy consumption is planned in advance in areas that require climbing or acceleration to ensure that the UAV has enough power to complete the task during the entire inspection process and avoid interruption of the task due to insufficient power. Based on the adjusted flight parameters, combined with radio orientation and navigation data and terrain information, the dynamic inspection route of the UAV is planned. The sampling-based rapid exploration random tree algorithm searches for feasible paths from the current position of the UAV to the target inspection area in three-dimensional space, randomly generates a series of path points, and screens and optimizes them according to constraints to construct a dynamic inspection route that meets the requirements. During the planning period, the route is continuously adjusted based on new data obtained in real time, such as terrain changes and obstacle movements. Adjustment and optimization ensure that the route always adapts to actual environmental changes. Constraints include terrain, obstacles and flight parameters. When the UAV flies along the dynamic inspection route, the path planning module continuously monitors the flight status of the UAV and changes in the surrounding environment. The flight status includes position, speed and attitude, and changes in the surrounding environment include changes in terrain and the appearance of new obstacles. By maintaining real-time data interaction with the radio navigation module and the lidar sensor, the latest information is obtained in a timely manner. Once changes in the surrounding environment or changes in the flight status of the UAV are detected, resulting in the current inspection route no longer being optimal or no longer feasible, the path planning module immediately replans the path. For example, if sudden severe weather such as strong winds and heavy rains are encountered during the flight, affecting the flight safety of the UAV, or new obstacles such as suddenly collapsed trees are found, blocking the original route, the path planning module will quickly recalculate the flight parameters according to the new situation, generate a new dynamic inspection route, and transmit the adjusted route information to the flight control system of the UAV, so that the UAV can change the flight trajectory in time and continue to perform inspection tasks safely and efficiently.

[0025] Preferably, the process of planning energy consumption in advance in areas where climbing or acceleration is required is as follows:

[0026] The path planning module obtains terrain data, including the slope, height difference, and distance of the area that needs to be climbed or accelerated, through data interaction with the laser radar sensor and the radio navigation module, and clarifies the task requirements of the inspection task for these areas, such as whether a detailed inspection is required at a specific height. For example, through laser radar scanning, the slope of the hillside ahead is 30 degrees, the climbing height needs to reach 100 meters, and the horizontal distance is 500 meters. According to the task requirements, high-definition images of the slope surface need to be collected during the climbing process. According to the technical specifications of the drone, the energy consumption characteristic parameters during climbing and acceleration are obtained to obtain the performance parameters of the drone, where the energy consumption characteristic parameters include the power consumption curve of the drone at different speeds and climbing rates. For example, the power consumption is 500W when the level flight speed is 10m / s, and the power consumption increases to 800W when the climbing rate is 2m / s; at the same time, the capacity and discharge efficiency of the drone battery are considered, such as the battery capacity is 10000mAh and the discharge efficiency is 80%;

[0027] Based on terrain data and drone performance parameters, calculate the energy consumption in areas where climbing or acceleration is required:

[0028] For the climbing process, using formula E climb = mgh / P climb Calculate the required energy, where m is the mass of the drone, g is the acceleration due to gravity, h is the climbing height, and P climb is the climbing power. For example, if the mass of the drone is 5kg, the acceleration due to gravity is 9.8m / s 2 , the climbing height is 100 meters, the climbing power is 800W, then the energy required for climbing is E climb =5×9.8×100 / 800=6.125Wh;

[0029] For the acceleration process, according to the kinetic formula Calculate the energy consumption, where m is the mass of the drone, v is the speed after acceleration, and P accel is the acceleration power. Assuming that the UAV accelerates from an initial speed of 5 m / s to 10 m / s, the acceleration power is 600 W. The energy required for acceleration can be calculated as

[0030]

[0031] After calculating the basic climbing and acceleration energy consumption, combined with the impact of the flight path and mission requirements on energy consumption, if additional operations are required in the climbing or acceleration area, including hovering photography and turning, the corresponding energy consumption calculation is increased. For example, during the climbing process, it is necessary to hover for 10 seconds for image acquisition. Assuming the hovering power is 400W, the hovering energy consumption is E hover=400×10 / 3600=1.11Wh, where the time is converted into hours for calculation. At the same time, the energy is allocated based on the priority and time requirements of the inspection task and the total energy consumption does not exceed the available energy of the drone battery to obtain the total energy consumption. If there is a higher task requirement in a certain area, the energy consumption needs to be increased to ensure the quality of task completion;

[0032] According to the calculated total energy consumption, optimize the flight parameters of the drone in the climbing or acceleration area. If the energy consumption exceeds expectations, reduce the climbing speed or acceleration rate to reduce power consumption. For example, reduce the climbing speed from 2m / s to 1.5m / s, recalculate the climbing power and energy consumption, and find a balance point that can meet the mission requirements and reasonably control energy consumption. During the entire inspection process, continuously monitor the battery power of the drone. Before approaching the area where climbing or acceleration is required, re-evaluate the feasibility of the flight plan based on the remaining power and the calculated energy consumption. If the remaining power is not enough to support the energy consumption demand of the area, adjust the inspection route, bypass the area or choose an alternative path with lower energy consumption, or return to the charging station for charging before continuing the mission. For example, when the remaining battery power is 4000mAh, equivalent to 4Wh, and the calculated total energy consumption in the climbing and acceleration area is 6Wh, including the climbing, acceleration and hovering energy consumption calculated above, it is necessary to re-plan the path or take energy-saving measures, such as lowering the flight altitude, reducing unnecessary operations, etc., to ensure that the drone can safely complete the inspection mission and complete the formulation of the energy management strategy.

[0033] Preferably, the obstacle avoidance module includes a collision prediction unit and an obstacle avoidance execution unit; the collision prediction unit is used to analyze the terrain around the UAV in real time according to the terrain data of the radio navigation module and the laser radar sensor, build a collision prediction model, identify potential obstacles or sharply changing areas in the terrain, and generate a collision prediction result; the obstacle avoidance execution unit adjusts the flight attitude and flight speed of the UAV according to the collision prediction result to achieve safe terrain tracking and automatic avoidance;

[0034] The process of building a collision prediction model is as follows:

[0035] Obtain the navigation data of the drone from the radio navigation module, and obtain terrain data from the lidar sensor. The terrain data includes detailed information of the surrounding terrain and obstacles, including the height, slope, location, shape and size of the terrain. Preprocess the collected data to remove noise points and outliers; improve the accuracy and reliability of the data. For example, for lidar point cloud data, a filtering algorithm can be used to remove isolated points caused by measurement errors or environmental interference. At the same time, the data is coordinate transformed and normalized to unify them to the same coordinate system and be in a suitable numerical range for subsequent calculation and analysis. Based on the preprocessed data, extract the features related to collision prediction, including the distance, relative speed, relative direction between the drone and the surrounding obstacles, and the geometric shape features of the obstacles, such as length, width, height, area, volume, and slope change rate of the terrain. Through the correlation analysis in the feature selection algorithm, determine the most valuable feature subset for collision prediction, reduce the data dimension, and improve the calculation efficiency and accuracy of the model;

[0036] Select a decision tree to build a collision prediction model, divide the extracted feature subset into a training set and a test set, use the training set to train the model, and based on the node splitting rule of the decision tree, ensure that the model learns the relationship between features and the possibility of collision. During the training process, use the validation set to verify the model to prevent overfitting and ensure that the model has good generalization ability. When the UAV flies in different environments and working conditions, collect new sample data, add it to the training set, and retrain the model so that the model can adapt to more diverse scenarios and conditions. Regularly evaluate the model's accuracy, recall rate, and F1 value indicators to obtain evaluation results, and improve the model based on the evaluation results to obtain a collision prediction model. For example, if it is found that the model has a low prediction accuracy under certain specific terrains or obstacle types, analyze the reasons and adjust the model's feature extraction method or algorithm parameters in a targeted manner to improve the model's performance in these situations.

[0037] Preferably, the process of generating collision prediction results is:

[0038] During the flight of the UAV, the collision prediction unit preprocesses and extracts features from the navigation data and lidar data obtained in real time to obtain the feature vector at the current moment, where the navigation data includes the position, speed, and direction of the UAV; the feature vector at the current moment is input into the constructed collision prediction model, which calculates the probability of the UAV colliding with the surrounding terrain or obstacles in the future based on the input feature vector. The probability value is between 0 and 1, where 0 means that a collision is almost impossible and 1 means that a collision is very likely to occur. According to the set collision threshold: 0.5, it can be adjusted according to actual needs and model performance to determine whether the UAV has a collision risk. If the calculated value is If the collision probability obtained is greater than or equal to the collision threshold, it is considered that there is a collision risk; otherwise, the current flight state is considered to be relatively safe. For example, the model calculates that the probability of the drone colliding with the obstacle in front within the next 3 seconds is 0.6. If the threshold is set to 0.5, the drone is determined to have a collision risk. If it is determined that there is a collision risk, the collision prediction unit further generates a collision prediction result, which includes the expected time and location of the collision, the location is relative to the drone or geographic coordinates, the collision object, such as the specific obstacle type or terrain features, and the estimated severity of the collision, such as a minor collision, a moderate collision, or a severe collision; that is, it can be evaluated based on factors such as speed, relative position, and obstacle characteristics.

[0039] The present invention provides a reservoir slope inspection method and system based on unmanned aerial vehicle laser radar, which has the following beneficial effects:

[0040] 1. The reservoir slope inspection method and system based on UAV laser radar ensures safety through real-time dynamic obstacle avoidance. Based on the collision prediction model and the terrain data of the laser radar, the UAV can automatically detect and avoid obstacles during inspection, ensuring safe flight in steep or narrow environments, reducing the risk of damage to the UAV and data loss; highly adaptable communication and stability, the system can adjust the communication frequency and power according to the signal interference situation, and adapt to the complex signal environment, thereby realizing stable communication of the UAV at long distances, signal interference or occlusion, and ensuring the integrity and real-time nature of the inspection data.

[0041] 2. The reservoir bank slope inspection method and system based on UAV lidar extends the energy management of battery life. Through real-time monitoring and optimization of energy consumption, the UAV can reasonably allocate electricity, especially in climbing or acceleration areas, to optimize battery utilization; the energy management strategy effectively improves the inspection efficiency and extends the battery life; it has strong autonomy and reduces labor costs. The system realizes full-process autonomous navigation, path optimization, obstacle avoidance and dynamic adjustment without human intervention, effectively reducing the workload and operation difficulty of inspection personnel, reducing labor costs and improving the intelligence level of inspection tasks, and improving the autonomous adaptability of UAVs to complex terrain; through comprehensive terrain data collection and continuous path optimization, the system can provide accurate terrain change monitoring and environmental assessment data for reservoir slopes, and provide reliable data support for reservoir management decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of a reservoir slope inspection method based on UAV laser radar of the present invention;

[0043] Figure 2 The present invention is a flowchart of a reservoir slope inspection system based on UAV laser radar. DETAILED DESCRIPTION

[0044] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0045] like Figure 1 and Figure 2 As shown, the present invention provides a technical solution: a reservoir bank slope inspection method and system based on unmanned aerial vehicle laser radar, including an unmanned aerial vehicle platform, a radio orientation module, a radio navigation module, a laser radar sensor, an obstacle avoidance module and a path planning module, wherein: the unmanned aerial vehicle platform is equipped with a radio orientation module and a radio navigation module, and performs position calibration by receiving radio signals sent by a radio reference station set up on the shore to optimize the flight path in real time; the laser radar sensor is used to collect three-dimensional data of the reservoir bank slope to obtain terrain data; the obstacle avoidance module combines the radio navigation signal and the laser radar data to analyze the terrain around the unmanned aerial vehicle in real time to achieve accurate obstacle avoidance; the path planning module is used to generate a dynamic inspection route of the unmanned aerial vehicle through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module, so as to ensure the stable and autonomous flight of the unmanned aerial vehicle in complex terrain;

[0046] Among them, the radio direction-finding module is used to automatically identify the radio reference station set up on the shore, and perform triangulation calculations by measuring the radio signal strength, arrival time difference and frequency offset of different radio reference stations, calculate the position information of the UAV relative to the radio reference station, and obtain positioning data; the radio direction-finding module includes a signal enhancement unit and an environmental adaptation unit; the signal enhancement unit is used to analyze the influence of different working conditions on the accuracy of positioning data based on the UAV under different working conditions. For example, when far away from the reference station, the signal strength may weaken, resulting in a decrease in positioning accuracy; in complex terrain areas, the signal may be reflected, blocked, etc., affecting the transmission and reception of the signal, and then affecting the positioning result, and by correcting and optimizing the positioning data , clarify the orientation accuracy of the UAV and obtain orientation data, among which the working conditions include being far away from the base station, signal interference, complex terrain and weak signal areas; enable the UAV to have stable positioning capabilities under various working conditions, and provide a reliable basis for subsequent flight control and navigation; the environmental adaptation unit is used to identify changes in the signal environment based on the environmental adaptation algorithm, and adjust the communication frequency and transmission power of the radio orientation module to adapt to the signal reflection and obstruction of complex terrain, improve the effective ranging range of the radio orientation module, and enable the UAV to maintain a reliable navigation signal in obstacle areas such as cliffs and steep slopes; by combining multiple modules, multi-functional collaboration is achieved, which can realize real-time inspection in complex terrain, effectively improve the accuracy and reliability of inspection, and reduce the need for manual intervention.

[0047] The process of adjusting the communication frequency and transmission power of the radio directional module is as follows:

[0048] The radio orientation module continuously collects communication-related data, including the signal strength, signal frequency, and signal arrival time received from the shore radio reference station, and records the current position of the drone, flight attitude, and some basic characteristics of the surrounding environment, such as terrain type and whether it is close to obstacles;

[0049] Preprocess the collected communication-related data to remove noise interference and normalize the data to obtain communication processing data to improve data quality and availability. Use the environment adaptation algorithm to analyze the preprocessed communication processing data to identify changes in the signal environment. The environment adaptation algorithm takes into account multiple factors, including the fluctuation of signal strength, the stability of signal frequency, and the change law of signal arrival time. For example, if it is found that the signal strength drops sharply in a short period of time and the signal arrival time is significantly delayed, and the drone is in a complex terrain area, such as near a cliff or a steep slope, it may be judged that the signal environment has changed, which may be caused by terrain obstruction or signal reflection, and the analysis result of the signal environment change is obtained;

[0050] According to the analysis results, determine the adjustment of the communication frequency: pre-set the frequency band list and preset threshold, sort them according to the interference degree from small to large, and when the interference is detected to exceed the preset threshold, switch to the next frequency band in the list for communication; for example, the initial communication frequency is 2.4GHz band, if the stronger interference signal A is detected in this band, and after analysis, it is determined that the interference in the 5.8GHz band is smaller, and the preset threshold is B, when A≥B, the communication frequency is adjusted to 5.8GHz. During the adjustment process, the equipment compatibility between the drone and the radio base station must also be considered to ensure that both parties can support the new communication frequency;

[0051] Adjustment of transmit power: According to the distance between the UAV and the radio base station and the signal strength, when the UAV is far away from the base station, the signal strength weakens. To ensure reliable communication, the transmit power is increased. The increase in transmit power is calculated based on the distance. The distance-based power adjustment formula P is used. d =P0+kd, where P d is the adjusted transmission power, P0 is the initial transmission power, k is the power adjustment coefficient, and d is the distance between the UAV and the base station; it should be further explained that, in the specific implementation process, the signal enhancement unit can maintain the directional accuracy in an environment with weak signals or interference, and the environmental adaptation unit improves the communication quality by adjusting the frequency and power, which ensures the stable operation of the UAV in complex terrain and signal environments; the communication frequency and transmission power are adjusted through the environmental adaptation algorithm: this technology realizes automatic signal interference management, improves communication reliability, avoids frequent signal loss, and ensures the continuity and stability of data transmission.

[0052] The radio navigation module combines the terrain data of the lidar sensor with a multi-path evaluation algorithm to dynamically identify the signal paths sent by different radio reference stations, obtain navigation data, and plan the best flight path for the UAV. It automatically bypasses obstacles or dangerous areas through the best flight path. Once an obstruction is detected on the path, the flight altitude or direction is immediately adjusted to enable the UAV to avoid steep slopes, cliffs and gullies in real time.

[0053] The process of the radio navigation module obtaining navigation data and planning the best flight path for the UAV is as follows: the radio navigation module continuously receives signals sent from different radio reference stations on the shore, including the location information of the reference stations and the signal sending time. At the same time, the module records the signal strength, frequency and related characteristics during the signal propagation process. For example, when each reference station sends a signal, it will attach its own coordinate information. After receiving these signals, the radio navigation module first decodes the signal and extracts the location data of the reference station and the signal sending timestamp. At the same time, the radio navigation module obtains the terrain data of the lidar sensor. The terrain data includes the height, slope, location and shape of the obstacles, and pre-processes the terrain data, including removing noise points, classifying the point cloud data, and distinguishing different types of objects such as the ground, vegetation, and buildings, so that these data can be used more effectively for path planning in the future;

[0054] A multipath evaluation algorithm is used to dynamically identify the signal paths sent by different radio reference stations. Based on the reflection, refraction and obstruction encountered by the signal during propagation, the propagation path of the signal from the reference station to the UAV is analyzed to obtain the identification result. For example, by comparing the actual received signal strength with the strength in the theoretical absence of interference, it is determined whether the signal has been reflected or obstructed. If the signal strength is significantly lower than expected and there are obstacles or complex terrain around, it may mean that the signal has been reflected multiple times or has been partially obstructed. Based on the identification result of the signal path, the triangulation principle is used to combine the location information of multiple reference stations and the signal arrival time difference to calculate the UAV's position in three dimensions. The position in space is used to calculate the navigation data of the drone. The navigation data includes the position, speed and direction of the drone. It should be further explained that in the specific implementation process, the coordinates of the three reference stations are known to be (x1, y1, z1), (x2, y2, z2), and (x3, y3, z3). The time when the signal arrives at the drone from these three reference stations is t1, t2, and t3 respectively. The position coordinates (x, y, z) of the drone are calculated by calculating the distance difference of signal propagation, according to the signal speed and time difference, and the triangulation formula is used to calculate the position coordinates (x, y, z) of the drone. At the same time, the speed and flight direction of the drone are calculated according to the continuous position data to comprehensively obtain the navigation data of the drone.

[0055] Optimal flight path planning: Based on the terrain data provided by the lidar and the calculated navigation data, the environment around the drone is analyzed to identify obstacles, including steep slopes, cliffs and gullies. For example, the location of steep slopes and cliffs is determined by analyzing the height change rate in the terrain data; obstacles are marked as dangerous areas based on their shape and location information. Based on the complexity of the terrain, the distribution of dangerous areas and the flight performance limitations of the drone, such as the minimum turning radius, maximum climbing angle or maximum descent angle, the Dijkstra algorithm is used to search in three-dimensional space for feasible paths from the current position of the drone to the target area, such as the area to be inspected on the slope of a reservoir. During the search process, different paths are evaluated, and the path cost is calculated based on the path length, the risk level of passing through the dangerous area, and the energy consumption of the flight. For example, a path passing through a dangerous area will be assigned a higher cost, and a longer path will also have a corresponding increase in cost. By searching and comparing the costs of different paths, the path with the lowest cost is found. The lowest path is taken as the optimal path; and during the flight of the UAV, the environmental changes are continuously monitored, such as terrain changes, the emergence of new obstacles and the status of the UAV itself, such as position, speed, and power. If it is found that the currently planned flight path is no longer feasible or is no longer the optimal path, the path planning is immediately re-performed until the optimal flight path is obtained. For example, when the UAV encounters a sudden strong wind during flight, causing it to deviate from the predetermined path, the radio navigation module will recalculate the optimal flight path according to the new position and environmental information, adjust the flight direction and altitude of the UAV, and ensure that it can continue to perform the inspection task safely and efficiently. At the same time, the new path planning information is transmitted to the flight control system of the UAV in a timely manner to realize real-time adjustment and dynamic planning of the UAV flight path. Through the above process, the radio navigation module can accurately obtain navigation data and plan the best flight path suitable for the UAV to fly in the complex environment of the reservoir slope, ensuring the smooth progress of the UAV inspection task.

[0056] It should be further explained that, in the specific implementation process, the multi-path evaluation algorithm is combined with lidar data to generate the best flight path: by dynamically identifying and evaluating the quality of multiple paths, the system can adaptively avoid obstacles and steep terrain, generate the safest and energy-saving flight path, and ensure the safety and effectiveness of the inspection mission; by using the Dijkstra algorithm for path optimization, the UAV can find the path with the lowest risk and lowest energy consumption in three-dimensional space, thereby extending the UAV's flight time and improving inspection efficiency.

[0057] The path planning module is used to generate the dynamic inspection route of the UAV through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module; during the generation of the dynamic inspection route, the path planning module uses a prediction algorithm to predict the complexity of the terrain to be encountered based on the current position, flight speed and terrain structure of the UAV, and then adjusts the flight parameters of the UAV in real time to obtain a dynamic inspection route, ensuring the stability and efficiency of the UAV inspection route and preventing inspection interruptions due to terrain changes.

[0058] The process of generating a dynamic inspection route for a UAV is as follows: the path planning module obtains orientation data from the radio orientation module. This data clarifies the position and orientation information of the UAV relative to the radio base station, so that the path planning module can accurately grasp the orientation of the UAV in space. At the same time, it receives navigation data from the radio navigation module, which contains the precise position, speed, flight direction and surrounding environment information of the UAV. These data provide comprehensive basic information for subsequent path planning. The current position of the UAV is determined using the orientation and navigation data obtained, and the terrain structure features around the position are extracted in combination with the terrain data provided by the lidar sensor. The terrain structure features include the degree of terrain undulation, such as slope and height difference, terrain type, such as mountain, plain, water edge, and whether there are obstacles, such as trees, buildings and cliffs;

[0059] Using a prediction algorithm, based on the current position, flight speed and direction of the drone, as well as the analyzed current terrain structure characteristics, the terrain complexity of the area that the drone will fly through in the future is predicted, and a prediction result is obtained. For example, by sampling and analyzing the terrain data on the flight trajectory of the drone, it is estimated whether there will be steep slopes, gullies or complex obstacle areas ahead, as well as the scope and severity of these areas;

[0060] According to the predicted results of terrain complexity, the flight parameters of the drone are adjusted in real time. If a steep slope or complex terrain is predicted ahead, the flight speed of the drone is reduced to ensure flight safety and stability. If it is necessary to cross a narrow valley or bypass a large obstacle, the flight altitude and direction of the drone are adjusted to enable it to pass smoothly. At the same time, based on the drone's endurance, the battery power is reasonably allocated according to the terrain and mission requirements. Energy consumption is planned in advance in areas that require climbing or acceleration to ensure that the drone has enough power to complete the task during the entire inspection process and avoid interruptions due to insufficient power.

[0061] Based on the adjusted flight parameters, combined with radio orientation and navigation data and terrain information, the dynamic inspection route of the UAV is planned. During the route planning process, multiple factors will be comprehensively considered, such as ensuring the integrity of the inspection coverage to avoid missing important areas; try to choose a path with relatively flat terrain and fewer obstacles to improve flight safety and efficiency; at the same time, to meet the inspection requirements of the inspection task for specific areas or targets, such as focusing on the key parts of the reservoir slope, the sampling-based rapid exploration random tree algorithm searches for feasible paths from the current position of the UAV to the target inspection area in three-dimensional space, randomly generates a series of path points, and screens and optimizes them according to the constraints to construct a dynamic inspection route that meets the requirements. During the planning period, the route is continuously adjusted and optimized according to the new data obtained in real time, such as terrain changes and obstacle movements, to ensure that the route always adapts to actual environmental changes. Among them, the constraints include terrain, obstacles and flight parameters. During the flight of the UAV along the dynamic inspection route, the path planning module continuously monitors the flight status of the UAV and changes in the surrounding environment. Among them, the flight status includes position, speed, and speed. The module can monitor the surrounding environment and the flight status of the drone, and obtain the latest information in time. Once the surrounding environment changes or the flight status of the drone changes, resulting in the current inspection route no longer being optimal or feasible, the path planning module immediately re-plans the path. For example, if sudden severe weather such as strong winds and heavy rains are encountered during the flight, affecting the flight safety of the drone, or new obstacles such as suddenly collapsed trees are found, blocking the original route, the path planning module will quickly recalculate the flight parameters according to the new situation, generate a new dynamic inspection route, and transmit the adjusted route information to the flight control system of the drone, so that the drone can change the flight trajectory in time and continue to perform the inspection task safely and efficiently. It should be further explained that in the specific implementation process, through the above process, the path planning module can generate the dynamic inspection route of the drone according to the changing environment and task requirements, ensure the stability, efficiency and safety of the drone during the inspection of the reservoir bank slope, and realize the comprehensive and accurate monitoring of the reservoir bank slope.

[0062] It should be further explained that, in the specific implementation process, by analyzing the terrain structure characteristics in real time during the inspection process, predicting complex terrain and adjusting the flight parameters to achieve high-precision inspection, the ability of the UAV to adapt to terrain changes is effectively improved; by dynamically adjusting the flight parameters of the UAV through the predictive algorithm, the UAV can make advance adjustments based on the characteristics of the terrain to be encountered, ensuring the smoothness and safety of the inspection process, and effectively reducing the difficulty of operation in complex terrain.

[0063] The process of planning your energy expenditure in advance in areas where you need to climb or accelerate is as follows:

[0064] The path planning module obtains terrain data, including the slope, height difference, and distance of the area that needs to be climbed or accelerated, through data interaction with the lidar sensor and the radio navigation module, and clarifies the task requirements of the inspection task for these areas, such as whether a detailed inspection is required at a specific height. For example, through lidar scanning, the slope of the hillside ahead is 30 degrees, the climbing height needs to reach 100 meters, and the horizontal distance is 500 meters. According to the task requirements, high-definition images of the slope surface need to be collected during the climbing process. According to the technical specifications of the drone, the energy consumption characteristic parameters during climbing and acceleration are obtained to obtain the performance parameters of the drone. Among them, the energy consumption characteristic parameters include the power consumption curve of the drone at different speeds and climbing rates. It should be further explained that in the specific implementation process, the power consumption is 500W when the level flight speed is 10m / s, and the power consumption increases to 800W when the climbing rate is 2m / s; at the same time, the capacity and discharge efficiency of the drone battery are considered, such as the battery capacity is 10000mAh and the discharge efficiency is 80%;

[0065] According to the terrain data and the performance parameters of the UAV, the energy consumption in the area where climbing or acceleration is required is calculated: For the climbing process, the formula E is used. climb = mgh / P climb Calculate the required energy, where m is the mass of the drone, g is the acceleration due to gravity, h is the climbing height, and P climb It is the climbing power. It should be further explained that in the specific implementation process, the mass of the UAV is 5kg and the gravity acceleration is 9.8m / s 2 , the climbing height is 100 meters, the climbing power is 800W, then the energy required for climbing is E climb =5×9.8×100 / 800=6.125Wh; for the acceleration process, according to the kinetic formula Calculate the energy consumption, where m is the mass of the drone, v is the speed after acceleration, and P accel is the acceleration power. Assuming that the UAV accelerates from an initial speed of 5 m / s to 10 m / s, the acceleration power is 600 W. The energy required for acceleration can be calculated as

[0066]

[0067] After calculating the basic climbing and acceleration energy consumption, combined with the impact of the flight path and mission requirements on energy consumption, if additional operations are required in the climbing or acceleration area, including hovering shooting and turning, the corresponding energy consumption calculation is increased. It should be further explained that in the specific implementation process, it is necessary to hover for 10 seconds for image acquisition during the climbing process. Assuming the hovering power is 400W, the hovering energy consumption is E hover=400×10 / 3600=1.11Wh, where the time is converted into hours for calculation. At the same time, the energy is allocated based on the priority and time requirements of the inspection task and the total energy consumption does not exceed the available energy of the drone battery to obtain the total energy consumption. If there are higher task requirements in a certain area, the energy consumption needs to be increased to ensure the quality of task completion. It should be further explained that in the specific implementation process, in a key monitoring area, in order to obtain clearer images, it is decided to increase the shooting frequency during the climb, which will lead to increased energy consumption, but the overall energy consumption can be balanced by optimizing the flight speed and operation in other areas.

[0068] According to the calculated total energy consumption, optimize the flight parameters of the drone in the climbing or acceleration area. If the energy consumption exceeds expectations, reduce the climbing speed or acceleration rate to reduce power consumption. For example, reduce the climbing speed from 2m / s to 1.5m / s, recalculate the climbing power and energy consumption, and find a balance point that can meet the mission requirements and reasonably control energy consumption. During the entire inspection process, continuously monitor the battery power of the drone. Before approaching the area that needs to climb or accelerate, re-evaluate the feasibility of the flight plan based on the remaining power and the calculated energy consumption. If the remaining power is not enough to support the energy consumption demand of the area, adjust the inspection route, bypass the area or choose an alternative path with lower energy consumption, or return to the charging station. The mission will continue after charging at the power station. It needs to be further explained that, in the specific implementation process, the remaining battery power is 4000mAh, equivalent to 4Wh, and the calculated total energy consumption in the climbing and acceleration areas is 6Wh, including the climbing, acceleration and hovering energy consumption calculated above. It is necessary to re-plan the path or take energy-saving measures, such as lowering the flight altitude, reducing unnecessary operations, etc., to ensure that the UAV can safely complete the inspection mission and complete the formulation of the energy management strategy; it needs to be further explained that, in the specific implementation process, through the above process, the energy consumption can be reasonably planned in advance in the area where climbing or acceleration is required, ensuring that the UAV has sufficient energy support during the inspection mission, while optimizing the flight performance and the quality of mission completion.

[0069] It should be further explained that, in the specific implementation process, through the energy management strategy in the climbing or acceleration area and the optimization of energy consumption, the power consumption of the UAV in the climbing or acceleration area is effectively allocated, thereby reducing energy waste and improving the UAV's endurance and the completion rate of inspection tasks; and by constructing a collision prediction model, obstacles can be effectively identified during the UAV inspection, and the flight attitude and speed can be adjusted in time, thereby enhancing the UAV's autonomous obstacle avoidance capability in complex terrain.

[0070] The obstacle avoidance module includes a collision prediction unit and an obstacle avoidance execution unit; the collision prediction unit is used to analyze the terrain around the drone in real time based on the terrain data of the radio navigation module and the lidar sensor, build a collision prediction model, identify potential obstacles or areas of rapid change in the terrain, and generate collision prediction results; the obstacle avoidance execution unit adjusts the drone's flight attitude and flight speed based on the collision prediction results to achieve safe terrain tracking and automatic avoidance;

[0071] The process of building a collision prediction model is as follows: obtaining the navigation data of the UAV from the radio navigation module and obtaining the terrain data from the lidar sensor. The terrain data includes detailed information of the surrounding terrain and obstacles, including the height, slope, location, shape and size of the terrain. The collected data is preprocessed to remove noise points and outliers. The accuracy and reliability of the data are improved. For example, for lidar point cloud data, a filtering algorithm can be used to remove isolated points caused by measurement errors or environmental interference. At the same time, the data is coordinate transformed and normalized to be unified to the same coordinate system and in a suitable numerical range for subsequent calculation and analysis. Based on the preprocessed data, features related to collision prediction are extracted, including the distance, relative speed, relative direction, and geometric shape features of the obstacles between the UAV and the surrounding obstacles, such as length, width, height, area, volume, and slope change rate of the terrain. Through correlation analysis in the feature selection algorithm, the most valuable feature subset for collision prediction is determined to reduce the data dimension and improve the calculation efficiency and accuracy of the model. For example, the distance between the UAV and the obstacle is found. Distance and relative speed are the two features that have the greatest impact on collision prediction, while some complex terrain features have little impact on collision prediction in the short term. Therefore, distance and relative speed are selected as the main features for model construction, and a decision tree is selected to build a collision prediction model. The extracted feature subsets are divided into training sets and test sets. The model is trained using the training set. Based on the node splitting rule of the decision tree, the relationship between the model learning features and the possibility of collision is ensured. During the training process, the model is verified using the validation set to prevent overfitting and ensure that the model has good generalization ability. When the drone flies in different environments and working conditions, new sample data is collected, added to the training set, and the model is retrained so that the model can adapt to more diverse scenarios and conditions. The accuracy, recall rate, and F1 value of the model are regularly evaluated to obtain evaluation results, and the model is improved based on the evaluation results to obtain a collision prediction model. For example, if it is found that the prediction accuracy of the model is low under certain specific terrains or obstacle types, the reasons are analyzed and the feature extraction method or algorithm parameters of the model are adjusted in a targeted manner to improve the performance of the model in these situations.

[0072] The process of generating collision prediction results: During the flight of the drone, the collision prediction unit preprocesses and extracts features from the navigation data and lidar data obtained in real time to obtain the feature vector at the current moment, where the navigation data includes the drone's position, speed, and direction; the feature vector at the current moment is input into the constructed collision prediction model. The collision prediction model calculates the probability of the drone colliding with the surrounding terrain or obstacles in the future based on the input feature vector. The probability value is between 0 and 1, where 0 means that a collision is almost impossible and 1 means that a collision is very likely to occur. According to the set collision threshold: 0.5, it can be adjusted according to actual needs and model performance to determine whether the drone has a collision risk. If the calculated collision probability is greater than or equal to the collision threshold, it is considered that there is a collision risk; otherwise, the current flight state is considered to be relatively safe. For example, the model calculates that the probability of the drone colliding with the obstacle ahead in the next 3 seconds is 0.6 If the threshold is set to 0.5, the UAV is judged to have a collision risk. If it is judged to have a collision risk, the collision prediction unit further generates a collision prediction result, which includes the expected time and location of the collision, the location is relative to the UAV or geographic coordinates, the collision object, such as the specific obstacle type or terrain features, and the estimated severity of the collision, such as a slight collision, a moderate collision, or a severe collision; that is, it can be evaluated based on factors such as speed, relative position, and obstacle characteristics. For example, the prediction result may show that the UAV will have a moderate collision with a large tree 10 meters ahead (relative to the current position of the UAV) in 5 seconds. The collision may cause the UAV's flight attitude to lose control, and it is necessary to take obstacle avoidance measures in time. These detailed prediction results will provide important decision-making basis for the obstacle avoidance execution unit, so as to adopt corresponding obstacle avoidance strategies to ensure the flight safety of the UAV. At the same time, the prediction results are recorded for subsequent flight data analysis and model optimization;

[0073] When using it, first ensure that the drone is in normal working condition, start the lidar, connect to the ground GNSS base station on the shore, calibrate the drone's position and communication signal, and set the inspection area boundary and flight parameters; use the navigation data provided by the shore base station and the terrain data obtained by the lidar to generate the drone's initial inspection path. During the path design, avoid dangerous areas to ensure that the entire slope is covered;

[0074] Then, the aircraft begins patrolling according to the planned route, dynamically positioning itself based on radio signals and terrain data, and continuously collecting three-dimensional terrain data of the slope during flight to update its position and flight path in real time.

[0075] Later, during the inspection process, the drone identifies obstacles or dangerous terrain ahead based on terrain data acquired by the LiDAR, and automatically adjusts the flight altitude and direction. If it encounters complex terrain, it replans the route to ensure the continuity and safety of the inspection.

[0076] After the final inspection is completed, the system conducts a comprehensive analysis of all collected data, extracts the structural characteristics of the slope, and identifies high-risk areas prone to landslides and collapses, and finally generates a complete inspection report; after the drone completes the inspection and returns, the system resets to prepare for the next inspection, and stores the analysis data for further research or maintenance reference.

[0077] It should be further explained that, in the specific implementation process, by combining radio orientation and navigation technology with lidar sensors, the UAV's high-precision navigation and obstacle avoidance capabilities in complex reservoir terrain were achieved; the system works together through multiple modules, including path planning, and can achieve precise inspection tasks in complex reservoir slope environments, adapt to changes in signal environment under complex terrain conditions, and ensure the continuity, safety and efficiency of inspections.

[0078] It should be further explained that, in the specific implementation process, the collision prediction model based on the decision tree realizes the accurate obstacle avoidance of the UAV in different terrains and environments, effectively reduces the risk of collision, and improves the adaptability and safety of the UAV in various terrains; the collision prediction results are generated based on the collision prediction model, the collision probability is calculated in real time and detailed prediction results are generated, so that the system can accurately identify and handle potential collision risks, significantly improving the safety and accuracy of inspection tasks;

[0079] Through radio navigation and lidar terrain data, the best inspection route can be planned in real time, steep slopes and obstacles can be avoided, and the optimal path can be found in complex terrain to ensure the efficiency and safety of the drone inspection process.

[0080] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A reservoir slope inspection method based on UAV laser radar, characterized in that: The steps include: Step 1: Ensure that the UAV is in normal working condition, start the LiDAR, connect to the ground GNSS base station on the shore, calibrate the UAV position and communication signal, and set the inspection area boundary and flight parameters; use the navigation data provided by the shore base station and the terrain data obtained by the LiDAR to generate the initial inspection path of the UAV; Step 2: Start inspection according to the route planning, dynamically locate its own position and direction based on radio signals and terrain data, and continuously collect three-dimensional terrain data of the slope during flight; Step 3: During the inspection, the drone identifies obstacles or dangerous terrain ahead based on terrain data acquired by the LiDAR, automatically adjusts flight altitude and direction, and replans the route if it encounters complex terrain; Step 4: After the inspection is completed, the system conducts a comprehensive analysis of all collected data, extracts the structural characteristics of the slope, and identifies high-risk areas prone to landslides and collapses, and finally generates a complete inspection report; After the drone completes the inspection, it returns and the system is reset to prepare for the next inspection.

2. A reservoir slope inspection system based on UAV laser radar, characterized by: It includes a UAV platform, a radio orientation module, a radio navigation module, a laser radar sensor, an obstacle avoidance module and a path planning module, wherein: the UAV platform is equipped with a radio orientation module and a radio navigation module, and performs position calibration by receiving radio signals sent by a radio reference station set up on the shore; the laser radar sensor is used to collect three-dimensional data of the reservoir bank slope to obtain terrain data; the obstacle avoidance module combines the radio navigation signal and the laser radar data to analyze the terrain around the UAV in real time; the path planning module is used to generate a dynamic inspection route for the UAV through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module; Among them, the radio orientation module is used to automatically identify the radio reference station set up on the shore, perform triangulation positioning calculation by measuring the radio signal strength, arrival time difference and frequency offset of different radio reference stations, calculate the position information of the UAV relative to the radio reference station, and obtain positioning data; the radio orientation module includes a signal enhancement unit and an environmental adaptation unit; the signal enhancement unit is used to analyze the influence of different working conditions on the accuracy of positioning data based on the UAV under different working conditions, and clarify the orientation accuracy of the UAV by correcting and optimizing the positioning data to obtain orientation data, wherein the working conditions include being far away from the reference station, signal interference, complex terrain and signal weak area; the environmental adaptation unit is used to identify changes in the signal environment based on the environmental adaptation algorithm, and adjust the communication frequency and transmission power of the radio orientation module.

3. A reservoir slope inspection system based on drone laser radar according to claim 2, characterized in that: The process of adjusting the communication frequency and transmission power of the radio orientation module is as follows: the radio orientation module continuously collects communication-related data, the communication-related data includes the signal strength, signal frequency, and signal arrival time received from the shore radio reference station, and records the current position, flight attitude, and some basic characteristics of the surrounding environment of the UAV, pre-processes the collected communication-related data, removes noise interference, and normalizes the data to obtain communication processing data, and uses an environmental adaptation algorithm to analyze the pre-processed communication processing data to identify changes in the signal environment. The environmental adaptation algorithm takes into account multiple factors, including the fluctuation of signal strength, the stability of signal frequency, and the change law of signal arrival time, to obtain the analysis result of the signal environment change; According to the analysis results, the communication frequency is adjusted: the frequency band list and preset threshold are pre-set, and they are sorted from small to large according to the interference degree. When the interference is detected to exceed the preset threshold, the next frequency band in the list is switched to communicate in turn; for the adjustment of the transmission power: according to the distance between the UAV and the radio base station and the signal strength, when the UAV is far away from the base station, the signal strength is weakened. In order to ensure reliable communication, the transmission power is increased. The increase in transmission power is calculated according to the distance, and the distance-based power adjustment formula P is used. d =P0+kd, where P d is the adjusted transmit power, P0 is the initial transmit power, k is the power adjustment coefficient, and d is the distance between the UAV and the base station.

4. A reservoir slope inspection system based on UAV laser radar according to claim 3, characterized in that: The radio navigation module combines the terrain data of the lidar sensor and adopts a multi-path evaluation algorithm to dynamically identify the signal paths sent by different radio reference stations, obtain navigation data, and plan the optimal flight path of the UAV.

5. The reservoir slope inspection system based on drone laser radar according to claim 4 is characterized by: The process of the radio navigation module obtaining navigation data and planning the best flight path for the UAV is as follows: The radio navigation module continuously receives signals sent from different radio reference stations on the shore, including the location information of the reference stations and the time when the signals are sent. At the same time, the module records the signal strength, frequency and related characteristics during the signal propagation process. At the same time, the radio navigation module obtains the terrain data of the lidar sensor, which includes the height, slope, location and shape of the terrain, and pre-processes the terrain data, including removing noise points and classifying the point cloud data. A multipath evaluation algorithm is used to dynamically identify the signal paths sent by different radio reference stations. Based on the reflection, refraction and shielding encountered by the signal during the propagation process, the propagation path of the signal from the reference station to the UAV is analyzed to obtain the identification result. According to the identification result of the signal path, the triangulation principle is used to combine the position information of multiple reference stations and the signal arrival time difference to calculate the position of the UAV in three-dimensional space, and calculate the navigation data of the UAV, which includes the position, speed and direction of the UAV. Optimal flight path planning: Based on the terrain data provided by the lidar and the calculated navigation data, the environment around the drone is analyzed to identify obstacles, including steep slopes, cliffs and gullies. Based on the complexity of the terrain, the distribution of dangerous areas and the flight performance limitations of the drone, the Dijkstra algorithm is used to search from the current position of the drone to the target area in three-dimensional space. During the search process, different paths are evaluated, and the path cost is calculated based on the path length, the risk level of passing through dangerous areas, and the energy consumption of the flight. By searching and comparing the costs of different paths, the path with the lowest cost is found as the optimal path; and during the flight of the drone, the environmental changes and the drone's own status are continuously monitored. If it is found that the currently planned flight path is no longer feasible or is no longer the optimal path, the path planning is immediately re-performed until the optimal flight path is obtained.

6. The reservoir slope inspection system based on drone laser radar according to claim 5 is characterized by: The path planning module is used to generate a dynamic inspection route of the UAV through the orientation data obtained by the radio orientation module and the navigation data obtained by the radio navigation module; During the generation of dynamic inspection routes, the path planning module uses a prediction algorithm to predict the complexity of the terrain to be encountered based on the current position, flight speed and terrain structure of the drone, and then adjusts the flight parameters of the drone in real time to obtain a dynamic inspection route.

7. The reservoir slope inspection system based on drone laser radar according to claim 6 is characterized by: The process of generating a dynamic inspection route for a drone is as follows: The path planning module obtains orientation data from the radio orientation module and receives navigation data from the radio navigation module. The obtained orientation and navigation data are used to determine the current position of the UAV, and combined with the terrain data provided by the laser radar sensor, the terrain structure features around the position are extracted. The terrain structure features include the degree of undulation of the terrain, the type of terrain, and whether there are obstacles. A prediction algorithm is used to predict the terrain complexity of the area where the drone is about to fly based on the current position, flight speed and direction of the drone, as well as the analyzed current terrain structure characteristics, and obtain a prediction result. Based on the prediction result of the terrain complexity, the flight parameters of the drone are adjusted in real time. If a steep slope or complex terrain is predicted ahead, the flight speed of the drone is reduced; if it is necessary to cross a narrow valley or bypass a large obstacle, the flight altitude and direction of the drone are adjusted. At the same time, based on the endurance of the drone, the battery power is reasonably allocated according to the terrain and mission requirements, and energy consumption is planned in advance in areas that require climbing or acceleration; Based on the adjusted flight parameters, combined with radio orientation and navigation data and terrain information, the dynamic inspection route of the UAV is planned. The sampling-based rapid exploration random tree algorithm searches for feasible paths from the current position of the UAV to the target inspection area in three-dimensional space, randomly generates a series of path points, and screens and optimizes them according to the constraints to construct a dynamic inspection route that meets the requirements. During the planning period, the route is continuously adjusted and optimized according to the new data obtained in real time. The constraints include terrain, obstacles and flight parameters. When the UAV flies along the dynamic inspection route, the path planning module continuously monitors the flight status of the UAV and changes in the surrounding environment; by maintaining real-time data interaction with the radio navigation module and the lidar sensor, the latest information is obtained in a timely manner. Once the surrounding environment changes or the flight status of the UAV changes, resulting in the current inspection route being no longer optimal or no longer feasible, the path planning module immediately re-plans the path.

8. The reservoir slope inspection system based on drone laser radar according to claim 7 is characterized by: The process of planning your energy expenditure in advance in areas where you need to climb or accelerate is as follows: The path planning module obtains terrain data, including the slope, height difference, and distance of the area that needs to be climbed or accelerated, by interacting with the laser radar sensor and the radio navigation module. At the same time, it clarifies the task requirements of the inspection task for these areas. According to the technical specifications of the UAV, it obtains its energy consumption characteristic parameters during climbing and acceleration, and obtains the performance parameters of the UAV. Among them, the energy consumption characteristic parameters include the power consumption curve of the UAV at different speeds and climbing rates; Based on terrain data and drone performance parameters, calculate the energy consumption in areas where climbing or acceleration is required: For the climbing process, using formula E climb = mgh / P climb Calculate the required energy, where m is the mass of the drone, g is the acceleration due to gravity, h is the climbing height, and P climb is the climbing power; For the acceleration process, according to the kinetic formula Calculate the energy consumption, where m is the mass of the drone, v is the speed after acceleration, and P accel is the acceleration power; After calculating the basic climbing and acceleration energy consumption, combined with the impact of the flight path and mission requirements on energy consumption, if additional operations are required in the climbing or acceleration area, including hovering shooting and turning, the corresponding energy consumption calculation is increased. At the same time, energy is allocated based on the priority and time requirements of the inspection task and based on the total energy consumption not exceeding the available energy of the drone battery, to obtain the total energy consumption; Based on the calculated total energy consumption, optimize the flight parameters of the drone in the climbing or acceleration area. If the energy consumption exceeds expectations, reduce the climbing speed or acceleration rate to reduce power consumption. During the entire inspection process, continuously monitor the battery power of the drone. Before approaching the area that needs to climb or accelerate, re-evaluate the feasibility of the flight plan based on the remaining power and the calculated energy consumption. If the remaining power is not enough to support the energy consumption requirements of the area, adjust the inspection route to bypass the area or choose an alternative path with lower energy consumption, or return to the charging station for charging before continuing the mission to complete the formulation of the energy management strategy.

9. The reservoir slope inspection system based on UAV laser radar according to claim 8 is characterized by: The obstacle avoidance module includes a collision prediction unit and an obstacle avoidance execution unit; the collision prediction unit is used to analyze the terrain around the UAV in real time according to the terrain data of the radio navigation module and the laser radar sensor, build a collision prediction model, identify potential obstacles or sharply changing areas in the terrain, and generate a collision prediction result; the obstacle avoidance execution unit adjusts the flight attitude and flight speed of the UAV according to the collision prediction result; The process of building a collision prediction model is as follows: obtaining the navigation data of the UAV from the radio navigation module and the terrain data from the lidar sensor, preprocessing the collected data to remove noise points and outliers, and extracting features related to collision prediction based on the preprocessed data, including the distance between the UAV and the surrounding obstacles, relative speed, relative direction, geometric shape characteristics of obstacles, and slope change rate of the terrain. The feature subset is determined through correlation analysis in the feature selection algorithm; A decision tree is selected to build a collision prediction model. The extracted feature subset is divided into a training set and a test set. The model is trained using the training set. The node splitting rule of the decision tree is used to ensure that the model learns the relationship between features and the possibility of collision. During the training process, the model is verified using a validation set. When the UAV flies in different environments and working conditions, new sample data is collected and added to the training set. The model is retrained and the accuracy, recall rate, and F1 value of the model are regularly evaluated to obtain evaluation results. The model is improved based on the evaluation results to obtain a collision prediction model.

10. The reservoir slope inspection system based on UAV laser radar according to claim 9 is characterized by: The process of generating collision prediction results: During the flight of the UAV, the collision prediction unit preprocesses and extracts features of the navigation data and lidar data obtained in real time to obtain the feature vector at the current moment, and inputs the feature vector at the current moment into the constructed collision prediction model. The collision prediction model calculates the probability value of the UAV colliding with the surrounding terrain or obstacles in the future based on the input feature vector. The probability value is between 0 and 1, where 0 indicates that a collision is almost impossible and 1 indicates that a collision is very likely to occur. According to the set collision threshold: 0.5, it is judged whether the UAV has a collision risk. If the calculated collision probability is greater than or equal to the collision threshold, it is considered that there is a collision risk; otherwise, it is considered that the current flight state is relatively safe. If it is determined that there is a collision risk, the collision prediction unit further generates a collision prediction result, which includes the expected time, location, collision object, and collision severity estimate of the collision.

Images