Unmanned aerial vehicle flight control method based on path constraint

Abstract

The invention discloses an unmanned aerial vehicle flight control method based on path constraint, and belongs to the technical field of unmanned aerial vehicle inspection, and the method comprises the steps: obtaining an inspection task and an initial flight path; monitoring a flight environment and a task state; dynamically adjusting a flight path according to the monitored data; generating an emergency flight path conforming to the current flight environment and the task state; the unmanned aerial vehicle is controlled to fly according to the emergency flight path, and the flight environment and the task state are continuously monitored during flight; the inspection task is completed and returned; according to the unmanned aerial vehicle flight control method based on the path constraint, the flight environment and the task state are monitored in real time, and the flight path is dynamically adjusted, so that the unmanned aerial vehicle can quickly respond in an emergency condition, smooth proceeding of the inspection task is ensured, and the emergency response capability is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle inspection, and in particular relates to a flight control method of an unmanned aerial vehicle based on path constraints. Background Art

[0002] In recent years, drone inspection technology has been widely used in the power industry, mainly due to the rapid development of aviation, remote sensing and information processing technologies. Drone inspection has the advantages of flexible methods and low cost, and can find defects that are difficult to find by manual inspection, becoming one of the key directions of line inspection technology development.

[0003] Existing drone inspection systems often rely on preset routes for automatic inspections. However, during the actual inspection process, unexpected situations may occur, such as weather changes, equipment failures, emergency task changes, etc., requiring drones to be able to quickly adjust their flight paths to ensure the smooth completion of inspection tasks.

[0004] Defects and shortcomings of the existing technology:

[0005] 1. Drones fail to respond in time: Traditional drone inspection path planning methods often fail to respond quickly enough to these emergencies and fail to fully consider various constraints, resulting in reduced inspection efficiency or even mission failure.

[0006] 2. Greatly affected by the environment: Under sudden severe weather conditions, such as strong winds and heavy rain, the flight performance and stability of the drone will be greatly affected, and it may even fail to fly normally;

[0007] 3. Lack of intelligence: Autonomous inspection by UAVs is still in the early stages of development, and the entire product system needs to be continuously improved; therefore, it is necessary to develop a new path-constrained UAV flight control method to solve existing problems. Summary of the invention

[0008] The purpose of the present invention is to provide a UAV flight control method based on path constraints to solve the problem that the flight path cannot be dynamically adjusted during inspection.

[0009] To achieve the above object, the present invention provides the following technical solution: a flight control method of an unmanned aerial vehicle based on path constraints, comprising:

[0010] Obtain inspection tasks and initial flight paths;

[0011] Monitor flight environment and mission status;

[0012] Dynamically adjust the flight path based on the monitored data;

[0013] Generate emergency flight paths that meet the current flight environment and mission status;

[0014] Control the UAV to fly along the emergency flight path and continue to monitor the flight environment and mission status during flight;

[0015] Complete the inspection task and return.

[0016] Preferably, the initial flight path is pre-planned according to the terrain and tower locations of the inspection area.

[0017] Preferably, the flight environment includes: wind speed, wind direction and visibility;

[0018] The task status includes: remaining power and device health status.

[0019] Preferably, the dynamically adjusting the flight path includes: dynamically adjusting the flight path by a path constraint algorithm; the constraint conditions of the path constraint algorithm include:

[0020] Environmental constraints: The actual flight speed V of the drone actual The calculation formula is: Among them, V actual Represents the actual flight speed of the drone, V uav Represents the windless flight speed of the drone, V wind represents the wind speed of the drone, θ represents the angle between the wind direction and the flight direction of the drone; the calculation formula for the flight time T is as follows: Where L represents the path length;

[0021] Wind direction constraint: Set the maximum allowable deviation angle θ_max. If the angle θ is greater than the maximum allowable deviation angle θ_max, the path is not feasible. If the wind direction is opposite to the flight direction, the angle θ is 180°.

[0022] Visibility constraint: Set the minimum allowed visibility V min If the visibility at any point on the flight path is lower than the minimum permissible visibility V min , then the path is not feasible;

[0023] Task constraints: including inspection point coverage and inspection sequence;

[0024] Flight performance constraints: including the maximum speed, maximum climb, maximum dive angle and minimum turning radius of the drone;

[0025] Power constraint: Ensure that the drone has sufficient power during the inspection process.

[0026] Preferably, the task constraint processing method includes:

[0027] Path coding: sort and mark the key points on the inspection path to form an orderly coding sequence;

[0028] Discretize the inspection area into a grid map, where each grid represents a possible inspection point;

[0029] Encode the flight path of the drone as a sequence of grid points, each grid point corresponding to a position on the path;

[0030] A preset number of path codes are randomly generated as the initial population, each code representing a possible flight path;

[0031] The fitness of each path code in the population is evaluated based on the length, safety, and inspection efficiency of the flight path;

[0032] Through the selection, crossover and mutation operations of the genetic algorithm, the population is continuously optimized iteratively to generate new path codes;

[0033] When the preset number of iterations is reached or the fitness meets the set conditions, the algorithm is terminated and the optimal or approximately optimal flight path code is output;

[0034] The flight path of the UAV is encoded into a chromosome in the genetic algorithm through the genetic algorithm. One flight path plan is a chromosome.

[0035] Conduct fitness assessment on each flight path within the population to evaluate its pros and cons;

[0036] The selection methods of the genetic algorithm include: evaluating the pros and cons of each path according to the genetic algorithm fitness function, using roulette wheel selection or tournament selection method, selecting the path with high fitness from the current population as the parent generation to increase the chance of it being inherited to the next generation;

[0037] The crossover method of the genetic algorithm is: perform a crossover operation on the selected parent path, including single-point crossover or two-point crossover, and generate a new path solution by exchanging path segments;

[0038] The mutation methods of genetic algorithms include: making random changes to the set path, including changing the position of one or more nodes on the path;

[0039] The fitness function of the genetic algorithm includes: rewards for inspection point coverage or penalties for uncovered inspection points, and combining multiple constraints into a single evaluation index in the form of weighted summation to obtain a total fitness value;

[0040] Fitness = ω1*coverage reward + ω2*non-coverage penalty + ω3*other constraints 1 + ... + ω n *Other constraints n;

[0041] Among them, ω1, ω2, ..., ω n is the weight of each constraint;

[0042] Where: Coverage rate reward = actual number of inspection points covered / target number of inspection points covered;

[0043] Non-coverage penalty = number of uncovered inspection points * penalty coefficient;

[0044] Fitness = ω1*(actual number of inspection points covered / target number of inspection points covered)+ω2*(number of uncovered inspection points*penalty coefficient)+ω3*other constraints 1+...+ω n *Other constraints n.

[0045] Preferably, the flight performance constraints include: the flight performance constraints include the maximum speed, maximum climb, maximum dive angle, minimum turning radius, etc. of the UAV. In the genetic algorithm, the flight performance constraints can be processed by the following method:

[0046] Path smoothing: Clothoid curves are used to smooth the path. After the path is generated, it is smoothed to make the path meet the minimum turning radius requirement of the drone. Genetic algorithm and path smoothing are complementary. Genetic algorithm searches for the optimal path segment combination, and Clothoid curves are used for path smoothing, which can find the optimal and smoothest path in a complex environment.

[0047] The Clothoid curve is approximated by the following formula:

[0048]

[0049] Where: X and Y represent the horizontal and vertical coordinates of the point on the curve, A is a proportional constant, and t is the arc length from the starting point of the curve to the current point;

[0050] Path smoothing fitness function limitations:

[0051] The path smoothing fitness function gives penalties when the maximum slope and maximum speed change exceed the limit;

[0052] The flight state adaptability calculation formula is F(h,v)=αh+βv, where F represents the path smoothing fitness function, h represents the flight altitude, v represents the flight speed, and α and β represent weight coefficients, which are used to adjust the importance of altitude and speed in the path smoothing fitness function.

[0053] Preferably, the power constraint includes:

[0054] Power model: Establish a power consumption model for the drone, taking into account the impact of flight speed, flight altitude, and load factors on power consumption;

[0055] The structure of the UAV power consumption model is mainly built based on the Thevenin model;

[0056] Thevenin model consists of a voltage source (Thevenin voltage, V TH ) and a series resistor (Thevenin resistor, R TH ) can be equivalent to a complex circuit network; the formula of Thevenin model can be expressed as:

[0057] V TH =V OC ;

[0058] R TH =R int ;

[0059] V out =V TH -I out *R TH ;

[0060] P=V out *I out ;

[0061] in:

[0062] V TH It is the Thevenin voltage, which is equal to the open circuit voltage of the circuit network, that is, the voltage when both ends are open;

[0063] V OC is the open circuit voltage, that is, the voltage of the circuit network when there is no load;

[0064] R TH It is the Thevenin resistance, which is equal to the input resistance of the circuit network, that is, the internal resistance of the circuit network, and the resistance seen from the port when all independent power supplies are removed or set to zero;

[0065] R int is the internal resistance, i.e. the internal resistance of the circuit network;

[0066] V out is the output voltage, i.e., under load R L The voltage on

[0067] I out is the output current, that is, the current flowing through the load R L The current;

[0068] P is the power consumed by the drone in flight;

[0069] Path evaluation: When evaluating a path, the path length is calculated, and the amount of power required to complete the path length is calculated based on the power model to determine whether it is within the power limit of the drone. Paths that exceed the power limit are assigned a lower fitness value.

[0070] Preferably, the monitored data includes: obtaining real-time inspection scene information: including inspection route information and inspection meteorological information;

[0071] Inspection line information includes: line grade, pole tower and conductor model split number;

[0072] Inspection meteorological information includes: coordinate information, altitude, temperature, wind speed, wind direction and atmospheric pressure;

[0073] A weighted multi-objective optimization function is established to optimize the position, velocity, acceleration and time allocation of the trajectory of discrete path points by combining flight smoothness, dynamic characteristics and flight time performance indicators; discrete path points are isolated point sets and points where the path is discontinuous;

[0074] Flight smoothness: Use the curvature of the trajectory and the rate of change of acceleration to quantify flight smoothness. Calculation of trajectory curvature: The curvature of the trajectory can reflect the curvature of the flight trajectory. For a given flight trajectory, its smoothness is measured by calculating the curvature of each point on the trajectory; the calculation of trajectory curvature uses the derivative of the trajectory coordinates. For a plane curve, the curvature K is defined as the rate of change of the unit tangent vector with the arc length, and its calculation formula is:

[0075] Where: K represents the curvature;

[0076] and Respectively represent the speed of the curve in the X and Y directions, i.e., the first derivative;

[0077] and Respectively represent the acceleration of the curve in the X and Y directions, i.e., the second derivative;

[0078] Jerk: The rate of change of acceleration, also known as jerk, is a measure of the rate at which acceleration changes over time. Jerk is the time derivative of acceleration and is calculated as:

[0079] Where d represents the differential operator, which means to take the derivative of a physical quantity, a represents acceleration, and t represents time;

[0080] For two-dimensional or three-dimensional motion, jerk is a vector whose components are the vector sums of the rates of change of acceleration in the corresponding directions: in is the acceleration vector;

[0081] If the acceleration vector Represented by its components in the x, y, and z directions, the jerk vector The components can be expressed as:

[0082] Construct a performance indicator function: You can construct a performance indicator function J 平滑性 , performance indicator function

[0083] J 平滑性 is the weighted sum of the curvature K function and the jerk J function, J 平滑性 =ω k .J K +ω J .J J Among them, J K is a term based on the curvature K, J J is a term based on the jerk J, ω k and ω J is the corresponding weight coefficient, which is used to balance the importance of the two indicators in the performance indicator function;

[0084] Dynamic characteristics: including the flight mechanical behavior of the UAV, lift, drag, stability and maneuverability;

[0085] Lift: The lift performance index function is expressed as:

[0086]

[0087] Where: Cl represents the lift coefficient, Cl_max represents the maximum lift coefficient, dCl / dα represents the slope of the lift line, α0 represents the zero lift angle of attack, Induced Drag represents the lift-induced drag, ω Cl ,ω Cl_max ,ω dCl / dα , ω Induced Drag is the weight factor;

[0088] The lift coefficient Cl calculation formula is: Where L is the actual lift, ρ is the air density, V is the flight speed, and S is the wingspan area;

[0089] Resistance: The resistance performance index function is expressed as:

[0090] J D =ω Cd ·Cd-ω DragatZero Lift Drag at Zero Lift+ω Induced Drag ·Induced Drag+ω Wave Drag Wave Drag-ω Parasite Drag ·Parasite Drag

[0091] Where: Cd ,ω Drag at Zero Lift ,ω Induced Drag ,ω Wave Drag ,ω Parasite Drag is the weighting factor; Cd represents the drag coefficient, Drag at Zero Lift represents the zero lift drag, Induced Drag represents the induced drag, Wave Drag represents the wave drag, Parasite Drag represents the parasite drag, and Total Drag represents the total drag;

[0092] Stability: The stability performance index function includes two aspects: static stability and dynamic stability; the stability performance index function is expressed as:

[0093] J s =ω static ·S static +ω dynamic ·S dynamic

[0094] in:

[0095] ω static ,ω dynamic is the weight factor;

[0096] S static It represents the static stability index, which is evaluated by the positional relationship between the center of gravity and the center of lift. The basic evaluation formula for static stability is: in:

[0097] is the rate at which the lift coefficient Cl changes with the angle of attack α;

[0098] is the rate at which the pitch moment coefficient Cm changes with the angle of attack α;

[0099] The pitching moment coefficient itself is expressed as:

[0100] Cm 0 is the pitching moment coefficient at zero angle of attack;

[0101] Cm α α represents the rate of change of the pitch moment coefficient caused by the change in the angle of attack;

[0102] Cm q It represents the rate of change of the pitch moment coefficient caused by the change of the pitch angular velocity;

[0103] It expresses the rate of change of the pitch moment coefficient caused by the change of the elevator deflection angle;

[0104] α represents the angle of attack;

[0105] q represents the pitch angular velocity;

[0106] δ e Indicates the elevator deflection angle;

[0107] S dynamic Indicates the dynamic stability index, which is determined by analyzing the dynamic characteristics of the aircraft; damping ratio and the natural frequency ω n It is the key parameter to judge the dynamic stability of the system. The dynamic stability of the system is judged according to the characteristic value S.

[0108] if represents overdamping, the eigenvalues ​​are all negative real numbers, and the system is dynamically stable;

[0109] if Represents critical damping, the eigenvalue is a pair of negative real numbers, the system is dynamically stable, but the response speed is slow;

[0110] if represents underdamping, the eigenvalue is the complex conjugate root, the real part is negative, the system is dynamically stable, and there will be an oscillatory damped response;

[0111] if It represents no damping, the eigenvalue is a pair of pure imaginary numbers, the system is not dynamically stable, and there will be continuous oscillation;

[0112] Maneuverability: The maneuverability performance index function is as follows:

[0113]

[0114] Where: controlsurface ,ω frequency ,ω steady-state ,ω controlforce ,ω sensitivity ,ω delay ,ω handlingquality is a weight factor used to balance the importance of different indicators in the performance function;

[0115] E controlsurface represents the efficiency of the rudder surface,

[0116] R frequency represents the frequency response,

[0117] S steady-state represents the steady-state error,

[0118] F controlforce Represents control.

[0119] S sensitivity Represents control sensitivity,

[0120] D delay represents control lag,

[0121] Q handlingquality It represents the manipulative quality;

[0122] Flight time: The total time required for the drone to complete the mission, including the flight time from the starting point to the end point and possible mission execution time;

[0123] Combining flight smoothness, dynamic characteristics and flight time, assigning a weight to each indicator, the formula for optimizing the objective function is:

[0124] J=W 平滑性 ·J 平滑性 +W 动力学 ·J 动力学 +W 时间 ·J 时间 ;

[0125] Among them, J represents the weighted multi-objective optimization function, J 平滑性 represents the flight smoothness performance index function, J 动力学 Represents the performance index function of dynamic characteristics, J 时间 Represents the flight time performance indicator function, W 平滑性 , W 动力学 and W 时间 are the corresponding weight coefficients respectively;

[0126] The selection of weight coefficients includes: the mission requirements of the UAV and the flight environment: if the mission requires high stability of flight, increase W 平滑性 If the task requires a fast response, increase W 动力学 If the task is time-sensitive, increase W 时间 The value of

[0127] By solving the weighted multi-objective optimization function, a set of optimized discrete path points is obtained. The discrete path points meet all performance indicators and achieve the optimization of the overall performance.

[0128] Preferably, the position, velocity, acceleration and time allocation of the discrete path points are adjusted to minimize the optimization objective function J, and the convex optimization algorithm of gradient descent is used for optimization processing to generate a time-domain continuous trajectory represented by a piecewise polynomial, that is, an emergency flight path that meets the current flight environment and mission status.

[0129] In the convex optimization algorithm of gradient descent, the objective function is a convex function, and J represents a weighted multi-objective optimization function, which maps an input variable or a set of variables to an output value; specifically, J:R n →R is a real number space from n dimensions R n to the real space R, that is, for any two points x and y, that is, they are R n For any vector in , and any real number λ∈[0,1], J(λx+(1-λ)y)≤λJ(x)+(1-λ)J(y); the gradient descent convex optimization algorithm iteratively updates the parameters along the negative gradient direction of the objective function to find the minimum point of the objective function.

[0130] The main steps of the gradient descent convex optimization algorithm include:

[0131] Initialization: Select the initial point X0, set the learning rate α and the threshold η in the stopping criterion;

[0132] Calculate the gradient: at the current point X k At , calculate the gradient of the weighted multi-objective optimization function J(x) The gradient is the steepest rising direction of the weighted multi-objective optimization function at the current point;

[0133] Update parameters: Update parameters according to the gradient direction and learning rate α Update the parameters along the negative direction of the gradient to find the minimum value of the function;

[0134] Check convergence: Repeat the steps of calculating gradients and updating parameters until the stopping criteria are met, which include: the gradient norm is less than a threshold, The gradient of the current point changes very little, the function value changes little, and is close to the local minimum; the change in the function value is less than the threshold: |J(X k+1 )-J(X k )∣≤∈, where ∈ is another preset small positive number, indicating the tolerance of function value changes;

[0135] Termination condition: If any of the stopping criteria is met, the algorithm terminates and outputs the current point Xk as an approximation of the optimal solution; if the stopping criteria is not met, continue to execute the gradient calculation step.

[0136] The technical effects and advantages of the present invention are as follows: the UAV flight control method based on path constraints monitors the flight environment and mission status in real time and dynamically adjusts the flight path. The UAV can respond quickly in emergency situations, ensure the smooth progress of inspection tasks, and improve emergency response capabilities; the inspection route path constraint algorithm comprehensively considers multiple constraints to generate the optimal emergency flight path, reduce invalid flight time, and improve inspection efficiency; by considering flight performance constraints and power constraints, it ensures that the UAV always remains in a safe flight state during the inspection process to avoid the occurrence of flight accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0137] Figure 1 It is a schematic diagram of a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0138] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0139] The present invention provides Figure 1 A UAV flight control method based on path constraints as shown in the figure includes the following steps:

[0140] S10: Obtaining the inspection task and initial flight path. The UAV receives the inspection task and obtains the initial flight path corresponding to the task. The initial flight path is pre-planned based on factors such as the terrain of the inspection area and the location of the tower.

[0141] S11: Real-time monitoring of flight environment and mission status. During the flight, the drone monitors the flight environment and mission status in real time through the sensors it carries.

[0142] Flight environment, such as wind speed, wind direction, visibility, etc.;

[0143] Task status, such as remaining battery power, device health, etc.

[0144] S12: Inspection route path constraint algorithm, based on real-time monitoring data, the inspection route path constraint algorithm is used to dynamically adjust the flight path;

[0145] S13: Generate an emergency flight path, based on the path constraint algorithm of the multi-objective optimization function, to generate an emergency flight path that meets the current flight environment and mission status;

[0146] S14: The UAV flies according to the generated emergency flight path, and continues to monitor the flight environment and mission status during the flight, and further adjusts the flight path;

[0147] S15: After completing the inspection mission and returning to the base, the drone automatically plans the path back to the base and lands safely after completing all inspection missions according to the emergency flight path.

[0148] In step S10, .

[0149] Determining the initial flight path involves a detailed analysis of the inspection area, including the ups and downs of the terrain, the specific location of the towers, and any obstacles that may affect the flight path. Based on the above information, the drone management system will plan an optimal flight path to ensure that the drone can complete the inspection task efficiently and safely. During the flight, the drone will also adjust the path according to actual conditions to deal with emergencies or optimize flight efficiency.

[0150] In step S11,

[0151] During the flight, the drone uses the sensors it carries to monitor the flight environment and mission status in real time. The sensors can sense various parameters of the surrounding environment, such as wind speed, wind direction, visibility, etc., to help the drone adjust its flight attitude and path to ensure flight safety. At the same time, the drone will also monitor its own mission status, including remaining power, equipment health, etc., so as to promptly discover and resolve potential problems and prevent accidents.

[0152] In step S12, the flight path is dynamically adjusted using the inspection route path constraint algorithm based on the real-time monitoring data;

[0153] During the flight, the drone will use the path constraint algorithm to dynamically adjust the flight path according to the real-time monitoring data. The drone uses a variety of sensors, such as gas detectors and PM2.5 detectors, to monitor environmental factors such as the atmosphere, water quality, and soil in real time. At the same time, the drone system inspection will also conduct comprehensive inspection and analysis of various parts of the drone through sensor detection, data analysis, and manual inspection to ensure that the drone can operate normally during the flight and avoid accidents. Based on the data of real-time monitoring and analysis, the drone will use the path constraint algorithm to dynamically adjust its flight path to adapt to environmental changes and mission requirements.

[0154] The path constraint algorithm includes the following constraints:

[0155] 1. Environmental constraints: including the impact of environmental factors such as wind speed, wind direction, visibility, etc. on drone flight.

[0156] Wind speed mainly affects the flight speed and energy consumption of the UAV. In path planning, the impact of wind speed on flight time is considered as part of the path constraint. The actual flight speed V of the UAV actual is the windless flight speed V of the drone uav , wind speed V wind The actual flight speed of the UAV under the influence of wind speed is V, which is determined by the angle θ between the wind direction and the flight direction of the UAV. actual It can be calculated as: The flight time T can be calculated by the path length L and the actual flight speed Vactual:

[0157] Wind direction mainly affects the flight direction and energy consumption of the UAV. In path planning, it is necessary to ensure that the UAV can fly in the predetermined direction while considering the deviation of the flight trajectory caused by wind direction. The deviation of the UAV flight trajectory caused by wind direction can be evaluated by calculating the angle (θ) between the wind direction and the flight direction. If the wind direction is opposite to the flight direction and the angle is 180°, the UAV needs to consume more energy to maintain the flight direction. In the path constraint, a maximum allowable deviation angle θ_max can be set. If the calculated deviation angle is greater than θ_max, the path is considered infeasible.

[0158] Visibility mainly affects the visual navigation and obstacle avoidance capabilities of the drone. In path planning, it is necessary to ensure that the drone can maintain sufficient visibility during flight to ensure safety. Visibility constraints can be set by setting a minimum allowed visibility V min To achieve this, if the visibility at any point on the path is lower than V min , then the path is considered infeasible.

[0159] 2. Task constraints: Task constraints refer to the specific requirements of the inspection task, such as the coverage of inspection points, inspection sequence, etc. Task constraints are handled by the following methods:

[0160] Path coding: sort and mark the key points on the inspection path to form an orderly coding sequence;

[0161] Discretize the inspection area into a grid map, where each grid represents a possible inspection point;

[0162] Encode the flight path of the drone as a sequence of grid points, each grid point corresponding to a position on the path;

[0163] A preset number of path codes are randomly generated as the initial population, each code representing a possible flight path;

[0164] The fitness of each path code in the population is evaluated based on the length, safety, and inspection efficiency of the flight path;

[0165] Through the selection, crossover and mutation operations of the genetic algorithm, the population is continuously optimized iteratively to generate new path codes;

[0166] When the preset number of iterations is reached or the fitness meets the set conditions, the algorithm is terminated and the optimal or approximately optimal flight path code is output;

[0167] The flight path of the UAV is encoded into a chromosome in the genetic algorithm through the genetic algorithm. One flight path plan is a chromosome.

[0168] Conduct fitness assessment on each flight path within the population to evaluate its pros and cons;

[0169] The selection methods of the genetic algorithm include: evaluating the pros and cons of each path according to the genetic algorithm fitness function, using roulette wheel selection or tournament selection method, selecting the path with high fitness from the current population as the parent generation to increase the chance of it being inherited to the next generation;

[0170] The crossover method of the genetic algorithm is: perform a crossover operation on the selected parent path, including single-point crossover or two-point crossover, and generate a new path solution by exchanging path segments;

[0171] The mutation methods of genetic algorithms include: making random changes to the set path, including changing the position of one or more nodes on the path;

[0172] The fitness function of the genetic algorithm includes: rewards for inspection point coverage or penalties for uncovered inspection points, and combining multiple constraints into a single evaluation index in the form of weighted summation to obtain a total fitness value;

[0173] Fitness = ω1*coverage reward + ω2*non-coverage penalty + ω3*other constraints 1 + ... + ω n *Other constraints n;

[0174] Among them, ω1, ω2, ..., ω n is the weight of each constraint;

[0175] Where: Coverage rate reward = actual number of inspection points covered / target number of inspection points covered;

[0176] Non-coverage penalty = number of uncovered inspection points * penalty coefficient;

[0177] Fitness = ω1*(actual number of inspection points covered / target number of inspection points covered)+ω2*(number of uncovered inspection points*penalty coefficient)+ω3*other constraints 1+...+ω n *Other constraints n.

[0178] 3. Flight performance constraints: Consider the UAV’s maximum flight distance, maximum turning angle, maximum climb / descent rate and other flight performance parameters.

[0179] Flight performance constraints include the maximum speed of the drone, maximum climb / dive angle, minimum turning radius, etc. In genetic algorithms, flight performance constraints can be handled by the following methods:

[0180] Path smoothing: The Clothoid curve is used to smooth the path. After the path is generated, it is smoothed to make the path meet the minimum turning radius requirement of the drone. The genetic algorithm and path smoothing complement each other. The genetic algorithm searches for the optimal path segment combination, and the Clothoid curve is used for path smoothing, which can find the optimal and smoothest path in a complex environment.

[0181] The Clothoid curve is approximated by the following formula:

[0182]

[0183] Where: X' and Y' represent the horizontal and vertical coordinates of the point on the curve, A is a proportional constant, and t is the arc length from the starting point of the curve to the current point;

[0184] Path smoothing fitness function limitations:

[0185] The path smoothing fitness function gives penalties when the maximum slope and maximum speed change exceed the limit;

[0186] The flight state adaptability calculation formula is F(h,v)=αh+βv, where F represents the path smoothing fitness function, h represents the flight altitude, v represents the flight speed, and α and β represent weight coefficients, which are used to adjust the importance of altitude and speed in the path smoothing fitness function.

[0187] 4. Power Constraint: Ensure that the drone has sufficient power during the inspection process, and automatically plan the return route for battery replacement when necessary. The following methods are used to handle power constraints:

[0188] Power model: Establish a power consumption model for the drone, taking into account the impact of factors such as flight speed, flight altitude, and load on power.

[0189] The structure of the UAV power consumption model is mainly built based on the Thevenin model;

[0190] The basic Thevenin model consists of a voltage source, the Thevenin voltage, V TH and a series Thevenin resistor, R THThe model composed of the above two components is equivalent to a complex circuit network. The formula of the Thevenin model can be expressed as:

[0191] V TH =V OC ;

[0192] R TH =R int ;

[0193] V out =V TH -I out *R TH ;

[0194] P=V out *I out ;

[0195] in:

[0196] V TH It is the Thevenin voltage, which is equal to the open circuit voltage of the circuit network, that is, the voltage when both ends are open;

[0197] V OC is the open circuit voltage, that is, the voltage of the circuit network when there is no load;

[0198] R TH It is the Thevenin resistance, which is equal to the input resistance of the circuit network, that is, the internal resistance of the circuit network, and the resistance seen from the port when all independent power supplies are removed or set to zero;

[0199] R int is the internal resistance, i.e. the internal resistance of the circuit network;

[0200] V out is the output voltage, i.e., under load R L The voltage on

[0201] I out is the output current, that is, the current flowing through the load R L The current;

[0202] P is the power consumed by the drone in flight;

[0203] Path evaluation: When evaluating a path, in addition to considering the path length, the power required to complete the path should be calculated based on the power model, and whether it is within the power limit of the drone should be determined. Paths that exceed the power limit should be given a lower fitness value.

[0204] In step S13, based on the inspection route path constraint algorithm, an emergency flight path that meets the current flight environment and mission status is generated.

[0205] Obtain real-time inspection scene information: including inspection line information and inspection weather information, which provide basic data for route planning; inspection line information, such as line grade, tower, and conductor model split number;

[0206] Inspection meteorological information, such as coordinate information, altitude, temperature, wind speed, wind direction, and atmospheric pressure;

[0207] Coordinate information and altitude help determine the geographical location and altitude of the path, and are helpful in analyzing the geographical environment and terrain undulations of the path.

[0208] Meteorological factors such as temperature, wind speed, and wind direction affect the climatic conditions along the route. For example, high or low temperatures, strong winds, etc. may have different effects on vehicles and personnel, and the route planning needs to be adjusted accordingly.

[0209] Meteorological information such as atmospheric pressure can help predict weather changes along a route, thereby ensuring the safety of route selection. For example, when planning an aviation or maritime route, the impact of changes in air pressure on flight or navigation must be considered.

[0210] Establish a safe flight channel: Based on real-time data, establish a safe flight channel represented by a set of convex polyhedrons to ensure the safety of the flight path.

[0211] A weighted multi-objective optimization function is established to optimize the position, velocity, acceleration and time allocation of the trajectory of discrete path points, which are isolated point sets and points where the path is discontinuous, by combining flight smoothness, dynamic characteristics and flight time performance indicators.

[0212] It should be noted that the multi-objective optimization function is a method of weighing and optimizing multiple performance indicators. In the path planning of UAVs, a weighted multi-objective optimization function can be established by combining performance indicators such as flight smoothness, dynamic characteristics, and flight time to optimize the position, speed, acceleration, and time allocation of discrete path points.

[0213] First, clarify the specific meaning and calculation method of each performance indicator:

[0214] Flight smoothness: flight smoothness is quantified using trajectory curvature and acceleration change rate indicators;

[0215] Calculation of trajectory curvature: The curvature of the trajectory can reflect the curvature of the flight trajectory. For a given flight trajectory, its smoothness is measured by calculating the curvature of each point on the trajectory. The calculation of curvature usually involves the derivative of the trajectory coordinates. For a plane curve, the curvature K is defined as the rate of change of the unit tangent vector with the arc length. The calculation formula is:

[0216] Where: K represents the curvature;

[0217] and They are the speed of the curve in the X and Y directions, i.e. the first derivative;

[0218] and They are respectively the acceleration of the curve in the X and Y directions, i.e., the second derivative;

[0219] Acceleration rate of change: Acceleration rate of change, also known as jerk, is a measure of the rate at which acceleration changes over time. Jerk is the time derivative of acceleration and is calculated as:

[0220] Where d represents the differential operator, which means to take the derivative of a physical quantity, a is the acceleration, and t is the time;

[0221] For two-dimensional or three-dimensional motion, the jerk J is a vector whose components are the vector sums of the rates of change of acceleration in the corresponding directions: in is the acceleration vector;

[0222] If the acceleration vector Represented by its components in the x, y, and z directions, the jerk vector The components can be expressed as:

[0223] Constructing a performance indicator function: Constructing a smoothness performance indicator function J 平滑性 is the weighted sum of the curvature K function and the jerk J function, J 平滑性 =ω k .J K +ω J .J J Among them, J K is a term based on the curvature K, J J is a term based on the jerk J, ω k and ω J is the corresponding weight coefficient, which is used to balance the importance of the two indicators in the performance indicator function;

[0224] Dynamic characteristics: including the flight mechanical behavior of the UAV, lift, drag, stability and maneuverability;

[0225] Lift: The lift performance index function is usually used to measure the ability of an aircraft to generate lift under specific conditions. Lift is one of the key forces that allows an aircraft to fly in the air. It is generated by the relative movement of the wings or rotors in the air. The lift performance index function can include the following key parameters:

[0226] Lift coefficient, Cl,: The lift coefficient is a dimensionless parameter used to measure the ratio of actual lift to theoretical lift. The calculation formula is Where L is the actual lift, ρ is the air density, V is the flight speed, S is the wingspan area,

[0227] Maximum lift coefficient, Cl_max, is the maximum lift coefficient that an aircraft can generate under specific flight conditions, such as a specific angle of attack or flight speed;

[0228] Lift line slope dCl / dα: refers to the sensitivity of the lift coefficient to the change of the angle of attack, the rate of change of the lift coefficient when the angle of attack changes;

[0229] Zero lift angle of attack α0: The angle of attack of the aircraft when no lift is generated, that is, the angle of attack when the lift coefficient is zero;

[0230] Induced Drag: The additional drag induced by the generation of lift, usually proportional to the square of the lift coefficient;

[0231] The lift performance index function is expressed as:

[0232]

[0233] Where: Cl represents the lift coefficient, Cl_max represents the maximum lift coefficient, dCl / dα represents the slope of the lift line, α0 represents the zero lift angle of attack, Induced Drag represents the lift-induced drag, ω Cl ,ω Cl_max ,ω dCl / dα , ω Induced Drag is the weight factor;

[0234] Drag: The drag performance index function is an indicator used to measure the amount of drag encountered by an object when it moves in a fluid. In the field of aviation, drag performance is particularly important because it is directly related to the fuel efficiency and range of the aircraft. The drag performance index function can include the following key parameters:

[0235] Drag coefficient Cd: Drag coefficient is a dimensionless parameter used to measure the ratio of the resistance encountered by an object when it moves in a fluid to the product of the airflow pressure and the reference area. The smaller the drag coefficient, the better the aerodynamic design of the object and the smaller the resistance.

[0236] Zero Lift: The drag under the condition that no lift is generated, usually including shape drag and friction drag;

[0237] Induced Drag: Induced drag is generated by lift. When lift is generated, the airflow forms vortices on the upper and lower surfaces of the wing. The vortices form vortices at the wing tips, generating additional drag.

[0238] Wave Drag: When flying at supersonic speed, the front end of the aircraft will generate compression waves on the air, forming shock waves, which will cause wave drag.

[0239] Parasite Drag: includes shape drag and friction drag, which is the drag caused by the friction between the aircraft surface and the air and the shape of the aircraft;

[0240] Total Drag: It is the sum of all the drags experienced by the aircraft under specific flight conditions;

[0241] The resistance performance index function is expressed as:

[0242] J D =ω Cd ·Cd-ω Drag at Zero Lift Drag at Zero Lift+ω InducedDrag InducedDrag+ω Wave Drag Wave Drag-ω Parasite Drag ·Parasite Drag

[0243] Where: Cd ,ω Drag at Zero Lift ,ω Induced Drag ,ω Wave Drag ,ω Parasite Drag is the weight factor;

[0244] Stability: The stability performance index function usually includes two aspects: static stability and dynamic stability. Static stability mainly considers the return or deviation trend of the aircraft after being disturbed in a balanced state, while dynamic stability involves the dynamic response characteristics of the aircraft such as damping characteristics and natural frequency.

[0245] The stability performance index function is expressed as:

[0246] J s =ω static ·S static +ω dynamic ·S dynamic

[0247] in:

[0248] ω static ,ω dynamic is a weight factor used to balance the importance of different indicators in the performance function;

[0249] S static It represents the static stability index, which is evaluated by the positional relationship between the center of gravity and the center of lift; the evaluation formula for static stability is:

[0250] is the rate at which the lift coefficient (Cl) changes with the angle of attack (α);

[0251] is the rate at which the pitch moment coefficient (Cm) changes with the angle of attack;

[0252] The pitching moment coefficient is expressed as:

[0253] Cm 0 It is the pitch moment coefficient at zero angle of attack, including the sum of the moments generated by the fuselage, wings, tail and other components at zero angle of attack;

[0254] Cm α α is the rate of change of the pitching moment coefficient caused by the change in angle of attack;

[0255] Cm q is the rate of change of the pitch moment coefficient caused by the change in pitch angular velocity;

[0256] is the rate of change of the pitch moment coefficient caused by the change in the elevator deflection angle;

[0257] α is the angle of attack;

[0258] q is the pitch angular velocity;

[0259] δ e is the elevator deflection angle;

[0260] If the above inequality holds, the aircraft is considered statically stable, because an increase in the angle of attack will lead to an increase in lift, which will produce a moment that causes the aircraft's angle of attack to decrease and return to its original flight state;

[0261] S dynamic Indicates the dynamic stability index, which is determined by analyzing the dynamic characteristics of the aircraft; damping ratio and the natural frequency ω n It is the key parameter to judge the dynamic stability of the system. The dynamic stability of the system is judged according to the characteristic value S.

[0262] if indicates overdamping), the eigenvalues ​​are all negative real numbers, and the system is dynamically stable.

[0263] if represents critical damping), the eigenvalue is a pair of negative real numbers, the system is dynamically stable, but the response speed is slow.

[0264] if It indicates underdamping, the eigenvalue is a complex conjugate root, the real part is negative, the system is dynamically stable, and there will be an oscillatory damped response.

[0265] if It means there is no damping, the eigenvalue is a pair of pure imaginary numbers, the system is not dynamically stable, and there will be continuous oscillations.

[0266] Maneuverability: The maneuverability performance index function is a mathematical representation used to quantify the aircraft's responsiveness to control inputs, including considerations of the aircraft's control surface efficiency, frequency response, steady-state error, control force, control sensitivity, control lag, and control quality.

[0267] Control surface efficiency E controlsurface : It can be determined experimentally and also predicted by numerical simulation;

[0268] Frequency Response R frequency : Frequency response refers to the response of the system to periodic input;

[0269] Steady-state error S steady-state : Steady-state error refers to the difference between the expected output and the actual output when the system transitions from one steady state to another;

[0270] Control force F controlforce : The size of the control force directly affects the pilot's fatigue level in the air and the difficulty of controlling the aircraft;

[0271] Control sensitivity S sensitivity : Control sensitivity refers to the sensitivity of the system to control input;

[0272] Control hysteresis D delay : Control lag refers to the time delay between control input and system response;

[0273] Control Quality Q handlingquality : Handling quality refers to the pilot's subjective evaluation of the aircraft's handling performance;

[0274] Taking the above factors into consideration, a maneuverability performance index function is defined as follows:

[0275]

[0276] Where:controlsurface ,ω frequency ,ω steady-state ,ω controlforce ,ω sensitivity ,ω delay ,ω handlingquality is a weight factor used to balance the importance of different indicators in the performance function;

[0277] Define the dynamic performance index function J 动力学 =ω L .J L +ω D .J D +ω S .J S +ω C .J C ,

[0278] where ω L ,ω D ,ω S ,ω C Represent the corresponding weight coefficients respectively;

[0279] Flight time: The total time required for a drone to complete a mission, including the flight time from the starting point to the destination and possible mission execution time.

[0280] Combining flight smoothness, dynamic characteristics and flight time, assigning a weight to each indicator, the formula for optimizing the objective function is:

[0281] J=W 平滑性 ·J 平滑性 +W 动力学 ·J 动力学 +W 时间 ·J 时间 ;

[0282] Among them, J represents the weighted multi-objective optimization function, J 平滑性 represents the flight smoothness performance index function, J 动力学 Represents the performance index function of dynamic characteristics, J 时间 Represents the flight time performance indicator function, W 平滑性 , W 动力学 and W 时间 are the corresponding weight coefficients respectively;

[0283] The selection of weight coefficients includes: the mission requirements of the UAV and the flight environment: if the mission requires high stability of flight, increase W 平滑性 If the task requires a fast response, increase W 动力学 If the task is time-sensitive, increase W 时间 The value of

[0284] By solving the weighted multi-objective optimization function, a set of optimized discrete path points is obtained. The discrete path points meet all performance indicators and optimize the overall performance. This path planning method can enable drones to complete various tasks more efficiently and stably in complex environments.

[0285] Generate emergency flight path: A convex optimization algorithm based on gradient descent is used to generate a time-domain continuous trajectory represented by a piecewise polynomial, that is, an emergency flight path that conforms to the current flight environment and mission status.

[0286] The position, velocity, acceleration and time allocation of the discrete path points are adjusted to minimize the optimization objective function J. The convex optimization algorithm of gradient descent is used for optimization processing to generate a time-domain continuous trajectory represented by a piecewise polynomial, that is, an emergency flight path that conforms to the current flight environment and mission status.

[0287] In convex optimization problems, the objective function is a convex function, and J represents a weighted multi-objective optimization function that maps an input variable or a set of variables to an output value. Specifically, J:R n →R is a real number space from n dimensions R n to the real space R, that is, for any two points x and y, that is, they are R n For any vector in , and any real number λ∈[0,1], J(λx+(1-λ)y)≤λJ(x)+(1-λ)J(y); the gradient descent convex optimization algorithm iteratively updates the parameters along the negative gradient direction of the objective function to find the minimum point of the objective function.

[0288] The main steps of the gradient descent convex optimization algorithm include:

[0289] Initialization: Select the initial point X0, set the learning rate α and the threshold η in the stopping criterion;

[0290] Calculate the gradient: at the current point X k At , calculate the gradient of the weighted multi-objective optimization function J(x) The gradient is the steepest rising direction of the weighted multi-objective optimization function at the current point;

[0291] Update parameters: Update parameters according to the gradient direction and learning rate α Update the parameters along the negative direction of the gradient to find the minimum value of the function;

[0292] Check convergence: Repeat the steps of calculating gradients and updating parameters until the stopping criteria are met, which include: the gradient norm is less than a threshold, The gradient of the current point changes very little, the function value changes little, and is close to the local minimum; the change in the function value is less than the threshold: |J(X k+1)-J(X k )∣≤∈, where ∈ is another preset small positive number, indicating the tolerance of function value changes;

[0293] Termination condition: If any of the stopping criteria is met, the algorithm terminates and outputs the current point Xk as an approximation of the optimal solution; if the stopping criteria is not met, continue to execute the gradient calculation step.

[0294] In step S14, the UAV flies according to the generated emergency flight path and continues to monitor the flight environment and mission status during the flight so as to further adjust the flight path if necessary;

[0295] The drone flies according to the generated emergency flight path, which is calculated by using advanced path generation algorithms by comprehensively considering the flight environment, mission requirements and safety factors, aiming to ensure that the drone can complete the mission efficiently and safely. During the flight, the drone will continue to monitor the flight environment and mission status, while detecting the flight status and equipment health of the drone. Based on these real-time monitoring data, the drone can further adjust the flight path when necessary to adapt to environmental changes and mission requirements, ensuring flight safety and smooth mission execution.

[0296] In step S15, after the drone completes all inspection tasks according to the emergency flight path, it automatically plans a path to return to the base and lands safely;

[0297] After completing all inspection tasks, the drone can automatically plan a path back to the base and land safely. In this process, the drone mainly relies on its advanced automatic return system. The system can map the flight environment in real time during the flight and record the flight trajectory. Once the aircraft remote control signal is interrupted or encounters an emergency, the flight control system will take over the control of the aircraft, refer to the original flight path to plan the route, and control the aircraft to return. During the return process, the drone can also intelligently choose to hover or bypass obstacles to ensure a safe return. In addition, the drone also has an intelligent low-battery return function, which can automatically plan the return path and land safely when the battery is low.

[0298] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A UAV flight control method based on path constraints, characterized in that: include: Obtain inspection tasks and initial flight paths; Monitor flight environment and mission status; Dynamically adjust the flight path based on the monitored data; Generate emergency flight paths that meet the current flight environment and mission status; Control the UAV to fly along the emergency flight path and continue to monitor the flight environment and mission status during flight; Complete the inspection task and return.

2. The method for controlling a UAV flight based on path constraints according to claim 1, characterized in that: The initial flight path is pre-planned according to the terrain and tower locations of the inspection area.

3. The method for controlling a UAV flight based on path constraints according to claim 1, characterized in that: The flight environment includes: wind speed, wind direction and visibility; The task status includes: remaining power and device health status.

4. A UAV flight control method based on path constraints according to any one of claims 1 to 3, characterized in that: The dynamically adjusting the flight path includes: dynamically adjusting the flight path by a path constraint algorithm; the constraint conditions of the path constraint algorithm include: Environmental constraints: The actual flight speed V of the drone actual The calculation formula is: Among them, V actual Represents the actual flight speed of the drone, V uav Represents the windless flight speed of the drone, V wind represents the wind speed of the drone, θ represents the angle between the wind direction and the flight direction of the drone; the calculation formula for the flight time T is as follows: Where L represents the path length; Wind direction constraint: Set the maximum allowable deviation angle θ_max. If the angle θ is greater than the maximum allowable deviation angle θ_max, the path is not feasible. If the wind direction is opposite to the flight direction, the angle θ is 180°. Visibility constraint: Set the minimum allowed visibility V min If the visibility at any point on the flight path is lower than the minimum permissible visibility V min , then the path is not feasible; Task constraints: including inspection point coverage and inspection sequence; Flight performance constraints: including the maximum speed, maximum climb, maximum dive angle and minimum turning radius of the drone; Power constraint: Ensure that the drone has sufficient power during the inspection process.

5. The method for controlling a UAV flight based on path constraints according to claim 4, characterized in that: The task constraint method includes: Path coding: sort and mark the key points on the inspection path to form an orderly coding sequence; Discretize the inspection area into a grid map, where each grid represents a possible inspection point; Encode the flight path of the drone as a sequence of grid points, each grid point corresponding to a position on the path; A preset number of path codes are randomly generated as the initial population, each code representing a possible flight path; The fitness of each path code in the population is evaluated based on the length, safety, and inspection efficiency of the flight path; Through the selection, crossover and mutation operations of the genetic algorithm, the population is continuously optimized iteratively to generate new path codes; When the preset number of iterations is reached or the fitness meets the set conditions, the algorithm is terminated and the optimal or approximately optimal flight path code is output; The flight path of the UAV is encoded into a chromosome in the genetic algorithm through the genetic algorithm. One flight path plan is a chromosome. Conduct fitness assessment on each flight path within the population to evaluate its pros and cons; The selection methods of the genetic algorithm include: evaluating the quality of each path according to the fitness function of the genetic algorithm, using the roulette wheel or tournament selection method, and selecting the path with high fitness from the current population as the parent to increase the chance of it being inherited to the next generation; The crossover method of the genetic algorithm is: perform a crossover operation on the selected parent path, including single-point crossover or two-point crossover, and exchange path segments to generate a new path solution; The mutation methods of genetic algorithms include: making random changes to the set path, including changing the position of one or more nodes on the path; The fitness function of the genetic algorithm includes: rewards for inspection point coverage or penalties for uncovered inspection points, and combining multiple constraints into a single evaluation index in the form of weighted summation to obtain a total fitness value; Fitness = ω1*coverage reward + ω2*non-coverage penalty + ω3*other constraints 1 + ... + ω n *Other constraints n; Among them, ω1, ω2, ..., ω n is the weight of each constraint; Where: Coverage rate reward = actual number of inspection points covered / target number of inspection points covered; Non-coverage penalty = number of uncovered inspection points * penalty coefficient; Fitness = ω1*(actual number of inspection points covered / target number of inspection points covered)+ω2*(number of uncovered inspection points*penalty coefficient)+ω3*other constraints 1+...+ω n *Other constraints n.

6. The method for controlling a UAV flight based on path constraints according to claim 4, characterized in that: The flight performance constraints include: The maximum speed, maximum climb, maximum dive angle and minimum turning radius of the drone; in the genetic algorithm, the flight performance constraints are handled by the following methods: Path smoothing: Use Clothoid curve to smooth the path. After the path is generated, smoothing is performed to make the path meet the minimum turning radius requirement of the drone. The Clothoid curve is approximated by the following formula: Among them: X' and Y 'represent the horizontal and vertical coordinates of the point on the curve, A is a proportional constant, and t is the arc length from the starting point of the curve to the current point; Path smoothing fitness function limitations: The path smoothing fitness function gives penalties when the maximum slope and maximum speed change exceed the limit; The flight state adaptability calculation formula is F(h,v)=αh+βv, where F represents the path smoothing fitness function, (h) represents the flight altitude, (v) represents the flight speed, and α and β represent weight coefficients, which are used to adjust the importance of altitude and speed in the path smoothing fitness function.

7. The method for controlling a UAV flight based on path constraints according to claim 4, characterized in that: The power constraints include: Power model: Establish the power consumption model of the UAV, including the impact of flight speed, flight altitude, and load factors on power; The power consumption model structure is built based on the Thevenin model; The formula of Thevenin model is expressed as: V TH =V OC ; R TH =R int ; V out =V TH -I out *R TH ; P=V out *I out ; Where: V TH It represents the Thevenin voltage, which is equal to the open circuit voltage of the circuit network; V OC Indicates the voltage when the circuit network is unloaded. ; R TH represents Thevenin resistance; R int It represents the internal resistance of the circuit network; V out Indicates the output voltage; I out Indicates the output current; P represents the power consumed by the drone during flight; Path evaluation: When evaluating a path, the path length is calculated, and the amount of power required to complete the path length is calculated based on the power model to determine whether it is within the power limit of the drone. Paths that exceed the power limit are assigned a lower fitness value.

8. The method for controlling a UAV flight based on path constraints according to claim 1, characterized in that: The monitored data includes: obtaining real-time inspection scene information: including inspection route information and inspection meteorological information; Inspection line information includes: line grade, pole tower and conductor model split number; Inspection meteorological information includes: coordinate information, altitude, temperature, wind speed, wind direction and atmospheric pressure; The generation of an emergency flight path that meets the current flight environment and mission status includes: establishing a weighted multi-objective optimization function, combining flight smoothness, dynamic characteristics, and flight time performance indicators to optimize the position, speed, acceleration, and time allocation of the trajectory of discrete path points; wherein the discrete path points are isolated point sets and points where the path is discontinuous; Flight smoothness: flight smoothness is quantified using trajectory curvature and acceleration change rate indicators; Calculation method of trajectory curvature: The formula is: Where: K represents the curvature; and Represent the speed of the curve in the X and Y directions respectively; and Represent the acceleration of the curve in the X and Y directions respectively; The formula for calculating the rate of change of acceleration is: Where d represents the differential operator, which means to take the derivative of a physical quantity, a represents acceleration, and t represents time; For two-dimensional or three-dimensional motion, the jerk J is a vector whose components are the vector sums of the rates of change of acceleration in the corresponding directions: in is the acceleration vector; If the acceleration vector Represented by its components in the x, y, and z directions, the jerk vector The components are expressed as: Constructing the flight smoothness performance index function includes: flight smoothness performance index function J 平滑性 is the weighted sum of the curvature K function and the jerk J function, J 平滑性 =ω k .J K +ω J .J J Among them, J K is a term based on the curvature K, J J is a term based on the jerk J, ω k and ω J is the corresponding weight coefficient; Dynamic characteristics: including the flight mechanical behavior of the UAV, lift, drag, stability and maneuverability; lift: the lift performance index function is expressed as: Where: Cl represents the lift coefficient, Cl_max represents the maximum lift coefficient, dCl / dα represents the slope of the lift line, α0 represents the zero lift angle of attack, Induced Drag represents the lift-induced drag, ω Cl ,ω Cl_max ,ω dCl / dα , ω Induced Drag is the weight factor; The lift coefficient Cl calculation formula is: Where L is the actual lift, ρ is the air density, V is the flight speed, and S is the wingspan area; Resistance: The resistance performance index function is expressed as: J D =ω Cd ·Cd-ω DragatZeroLift ·DragatZero Lift+ω InducedDrag · Induced Drag+ω Wave Drag ·Wave Drag-ω Parasite Drag ·Parasite Drag Where: Cd ,ω Drag atZero Lift ,ω Induced Drag ,ω Wave Drag ,ω Parasite Drag is the weight factor; Cd represents the drag coefficient, DragatZero Lift represents zero lift drag, Induced Drag represents induced drag, Wave Drag represents wave drag, Parasite Drag represents parasite drag, and Total Drag represents total drag; Stability: The stability performance index function includes two aspects: static stability and dynamic stability; the stability performance index function is expressed as: J s =ω static ·S static +oh dynamic ·S dynamic in: ω static ,ω dynamic is the weight factor; S static It represents the static stability index, which is evaluated by the positional relationship between the center of gravity and the center of lift. The basic evaluation formula for static stability is: in: is the rate at which the lift coefficient Cl changes with the angle of attack α; is the rate at which the pitch moment coefficient Cm changes with the angle of attack α; The pitching moment coefficient itself is expressed as: Cm 0 is the pitching moment coefficient at zero angle of attack; Cm α α represents the rate of change of the pitch moment coefficient caused by the change in the angle of attack; Cm q It represents the rate of change of the pitch moment coefficient caused by the change of the pitch angular velocity; It expresses the rate of change of the pitch moment coefficient caused by the change of the elevator deflection angle; α represents the angle of attack; q represents the pitch angular velocity; δ e Indicates the elevator deflection angle; S dynamic Indicates the dynamic stability index, which is determined by analyzing the dynamic characteristics of the aircraft; damping ratio and the natural frequency ω n It is the key parameter to judge the dynamic stability of the system. The dynamic stability of the system is judged according to the characteristic value S. if represents overdamping, the eigenvalues ​​are all negative real numbers, and the system is dynamically stable; if represents critical damping, the eigenvalue is a pair of negative real numbers, the system is dynamically stable, but the response speed is slow; if represents underdamping, the eigenvalue is the complex conjugate root, the real part is negative, the system is dynamically stable, and there will be an oscillatory damped response; if It represents no damping, the eigenvalue is a pair of pure imaginary numbers, the system is not dynamically stable, and there will be continuous oscillation; Maneuverability: The maneuverability performance index function is as follows: Where: ω controlsurface 、ω frequency 、ω steady-state 、ω controlforce 、ω sensitivity 、ω delay 、ω handlingquality are weighting factors; E controlsurface represents the efficiency of the rudder surface, R frequency represents the frequency response, S steady-state represents the steady-state error, F controlforce Represents control. S sensitivity Represents control sensitivity, D delay represents control lag, Q handlingquality It represents the manipulative quality; Dynamic performance index function J 动力学 =ω L .J L +ω D .J D +ω S .J S +ω C .J C , where ω L ,ω D ,ω S ,ω C Represent the corresponding weight coefficients respectively; Flight time: The total time required for a drone to complete a mission, including the flight time from the starting point to the destination and possible mission execution time. Combining flight smoothness, dynamic characteristics and flight time, assigning a weight to each indicator, the formula for optimizing the objective function is: J=W 平滑性 ·J 平滑性 +W 动力学 ·J 动力学 +W 时间 ·J 时间 ; Among them, J represents the weighted multi-objective optimization function, J 平滑性 represents the flight smoothness performance index function, J 动力学 Represents the performance index function of dynamic characteristics, J 时间 Represents the flight time performance indicator function, W 平滑性 , W 动力学 and W 时间 are the corresponding weight coefficients respectively; The selection of weight coefficients includes: the mission requirements of the UAV and the flight environment: if the mission requires high stability of flight, increase W 平滑性 If the task requires a fast response, increase W 动力学 If the task is time-sensitive, increase W 时间 The value of By solving the weighted multi-objective optimization function, a set of optimized discrete path points is obtained. The discrete path points meet all performance indicators and achieve the optimization of the overall performance.

9. The method for controlling a UAV flight based on path constraints according to claim 8, characterized in that: The method for optimizing the position, velocity, acceleration and time allocation of the trajectory of discrete path points includes: using a convex optimization algorithm of gradient descent to optimize and generate a time-domain continuous trajectory represented by a piecewise polynomial, that is, an emergency flight path that meets the current flight environment and mission status; In the convex optimization algorithm of gradient descent, the objective function is a convex function. J represents a weighted multi-objective optimization function, which maps an input variable or a set of variables to an output value. J:R n →R is a real number space from n dimensions R n to the real space R, that is, for any two points x and y, that is, they are R n For any vector in , and any real number λ∈[0,1], J(λx+(1-λ)y)≤λJ(x)+(1-λ)J(y); the gradient descent convex optimization algorithm iteratively updates the parameters along the negative gradient direction of the objective function to find the minimum point of the objective function.

10. The method for controlling a UAV flight based on path constraints according to claim 9, characterized in that: The main steps of the gradient descent convex optimization algorithm include: Initialization: Select the initial point X0, set the learning rate α and the threshold η in the stopping criterion; Calculate the gradient: at the current point X k At , calculate the gradient of the weighted multi-objective optimization function J(x) The gradient is the steepest rising direction of the weighted multi-objective optimization function at the current point; Update parameters: Update parameters according to the gradient direction and learning rate α Update the parameters along the negative direction of the gradient to find the minimum value of the function; Check convergence: Repeat the steps of calculating gradients and updating parameters until the stopping criteria are met, which include: the gradient norm is less than a threshold, The gradient change at the current point is close to the local minimum; the change in the function value is less than the threshold: |J(X k+1 )-J(X k )∣≤∈, where ∈ is another preset small positive number, indicating the tolerance of function value changes; Termination condition: If any of the stopping criteria is met, the algorithm terminates and outputs the current point Xk as an approximation of the optimal solution; if the stopping criteria is not met, continue to execute the gradient calculation step.

Images