A method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios
By constructing a rotorcraft path planning method based on enhanced particle swarm optimization, and combining rotor aerodynamics and terrain perception, the problems of distorted energy consumption assessment and unexecutable paths in urban low-altitude environments are solved, generating safe, economical and smooth flight paths.
Patent Information
- Application Number
- CN202611454157.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-18
- Publication Date
- 2026-11-17
AI Technical Summary
Existing path planning methods suffer from problems such as distorted energy consumption assessment, premature convergence, and path unexecutability in urban low-altitude dense building clusters, making it difficult to generate safe, economical, and smooth aircraft inspection paths.
A nonlinear power consumption model is constructed by using a particle swarm optimization method combined with rotor aerodynamics principles. An initial solution generation strategy for overhead power transmission channels and airspace hazard gradient responses is designed. Adaptive adjustment of dual closed-loop control parameters is achieved by integrating terrain perception and energy efficiency feedback. Path optimization is performed using cubic uniform B-spline curves.
It achieves accurate energy consumption estimation, rapid convergence, and path smoothness, ensuring that the aircraft can safely and efficiently plan executable inspection paths in urban low-altitude environments, avoiding energy consumption deviations and path point entanglement, and meeting the control characteristics requirements of rotorcraft.
Smart Images

Figure CN123113933A_ABST
Abstract
Description
Technical Field
[0001] This invention pertains to autonomous navigation technology for low-altitude aircraft and intelligent inspection technology for power facilities. Specifically, it relates to a path generation method and system for safe obstacle avoidance and efficient inspection of aircraft in urban low-altitude dense building cluster environments. Background Technology
[0002] With the development of smart grids, aircraft have gradually replaced traditional manual line inspections, becoming a key means of ensuring the stable operation of power networks. Unlike high-voltage transmission network inspections in open areas, the maintenance of urban distribution networks requires aircraft to penetrate deep into densely built-up near-ground spaces such as old neighborhoods or industrial areas to perform multi-directional, close-range observations of targets such as transformer nodes and insulator strings. In these application scenarios, the complex network of buildings, crisscrossing exposed cables, and communication towers significantly reduces the effective safe airspace, posing extremely stringent safety and energy consumption challenges to the aircraft's trajectory planning and control system.
[0003] To address the aforementioned problems, existing path planning methods are mainly divided into three categories: First, discrete space search methods (such as A...). (and its derivatives): The state space needs to be meshed. When facing heterogeneous obstacles with complex boundaries in a real three-dimensional continuous space, the storage and computational costs explode exponentially. Second, random sampling (such as RRT): Although it can handle high-dimensional spaces, the sampling success rate in narrow channels is extremely low, and the optimality of the obtained path cannot be measured. Third, population evolution (such as standard PSO, GWO): It is widely used because it does not require environmental gradients and has strong global search capabilities.
[0004] However, the above methods still have three major common limitations. First, physical energy consumption assessment is distorted. Existing technologies often use the linear superposition of spatial distance and altitude difference to roughly estimate power consumption, completely ignoring the nonlinear aerodynamic attenuation of rotorcraft during obstacle avoidance maneuvers and vertical takeoff phases. This easily leads to the "theoretically optimal route" output by the system being depleted of power due to power system overload during actual flight. Second, optimization in complex spaces is prone to premature convergence. Solvers, represented by basic swarm intelligence algorithms, often use blind point initialization across the entire domain when dealing with narrow corridors, resulting in wasted computational resources inside the building. Moreover, their optimization step size is mostly a rigid fixed value, which easily loses population activity when facing dense obstacles, gets trapped in local extrema, and causes path point deadlock. Third, underlying kinematics are incompatible. The discrete spatial polygonal lines calculated by traditional algorithms contain acute angle changes at nodes that violate the physical constraints of the aircraft, requiring the aircraft to have the ability to instantaneously change the momentum direction, which is absolutely impossible to directly track and execute in the underlying flight control system.
[0005] Therefore, there is an urgent need for an aircraft path planning method that can systematically solve the three major problems of energy consumption assessment distortion, premature convergence, and path non-executability. Summary of the Invention
[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios. This method replaces the traditional geometrically weighted energy consumption estimation method by constructing a high-fidelity power consumption model. It overcomes the problem of premature algorithm stalling by designing an initial solution generation strategy combining overhead power channels and airspace hazard gradient responses, and an adaptive adjustment mechanism for dual closed-loop control parameters. Finally, it outputs a continuous path that satisfies the aircraft's handling characteristics through uniform B-spline curve smoothing technology, ultimately achieving optimal overall performance in terms of safety, economy, smoothness, and rapid response. To achieve the above objectives, this invention proposes a method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, comprising the following steps: S1: Construct a heterogeneous geometric three-dimensional spatial model for complex operational airspace in urban low-altitude power line inspection application scenarios. S2: Construct a nonlinear power loss evaluation system based on rotor aerodynamics principles; S3: Construct the optimization objective cost function for multi-dimensional coupling of packages; S4: Initial solution generation based on overhead power transmission channels and airspace hazard gradient response; S5: Adaptive adjustment of dual closed-loop control parameters that integrate terrain perception and energy efficiency feedback; S6: Use cubic uniform B-splines to optimize the continuity of the planned path and output an executable smooth path.
[0007] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S1 is as follows: A hybrid geometric obstacle representation method is adopted: for ground structures with regular edge contours, such as commercial buildings and residential buildings, which are common in urban environments, a three-dimensional orthogonal hexahedron is used for enclosure; for tall pole-like facilities such as power transmission towers and communication base stations, a vertical cylindrical bounding box is used for enclosure. In terms of spatial layout, the two types of geometric shapes are arranged in a pseudo-random staggered manner to simulate the spatial characteristics of dense buildings and narrow passageways in a real urban canyon environment.
[0008] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S2, is as follows: A nonlinear power consumption model based on rotor aerodynamics is constructed. The instantaneous total power of the aircraft during flight between adjacent path points is calculated. The expression is: (1) in and These represent the shape drag power and induced power of a rotorcraft in hovering mode, respectively. The horizontal forward speed of a rotorcraft. The linear velocity at the rotor blade tip. The average induced airflow velocity during hovering. The drag coefficient of the fuselage. Atmospheric density, For rotor realism, For the area swept by the propeller disk, The total mass of the aircraft and its payload. This is the acceleration due to gravity.
[0009] The instantaneous total power consumption in formula (1) consists of the following four physical components: (a) Type drag power: used to overcome the surface friction drag generated when the rotor blades rotate in the air, and its value is proportional to the square of the horizontal speed of the aircraft; (b) Induced power: used to accelerate the airflow downward to generate lift against gravity, and its value decreases monotonically with increasing horizontal velocity; (c) Exhaust drag power: used to overcome the air resistance experienced by the aircraft fuselage and onboard equipment during horizontal motion, and its value is proportional to the cube of the horizontal velocity; (d) Climbing power: This component is used to overcome gravity and achieve a certain height. The condition-triggered method is defined as follows: (2) in This refers to the vertical ascent rate.
[0010] Global mission cumulative energy consumption By performing time-discretized accumulation of the instantaneous power for each flight segment: (3) in The flight time between the two waypoints. It is the total number of flight segments. This is the flight segment index number.
[0011] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S3 is as follows: To achieve the optimal balance among multiple objectives, the optimization objective cost function is constructed as follows: (4) The design incorporates a comprehensive evaluation index system with four independent cost components: (a) Cost of reduced battery life ( ): Mapped to the total power consumption calculated in S2 ; (b) Cost of Spatial Transition ( ): This evaluates the curvature abrupt changes in the path point sequence through vector dot product, used to filter out frequent yaws; it is used to suppress unnecessary direction abrupt changes, and its calculation method is as follows: (5) in Indicates by the first The path point points to the first The direction vectors of the path points Indicates by the first The path point points to the first The direction vectors of the path points This represents the total number of path points.
[0012] (c) Cost of extreme collision avoidance The method employs a dense sampling approach combined with an exponential repulsive field to penalize the distance between each sampling point and the surface of the nearest obstacle. If a sampling point enters the interior of an obstacle, a maximum constant penalty is applied. The calculation method is as follows: (6) in As the baseline intensity factor for obstacle avoidance penalty, The distance is the tangential distance from the sampling point to the surface of the nearest obstacle. As a safety buffer threshold, This is a very large penalty value.
[0013] (d) Cost of spillover from the safe corridor ): This restricts aircraft to fly within compliant altitudes. Penalties are imposed when aircraft exceed the upper or lower limits of the designated area. The calculation method is as follows: (7) in The height of the sampling point. and This is the height boundary of the safe corridor for the flight area.
[0014] These are the weights of each loss cost. The above four costs are weighted and summed using their respective weighting coefficients to form a comprehensive evaluation index.
[0015] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S4, is as follows: The Standard Particle Swarm Optimization (SSO) system adjusts its state based on information about the individual's historical best position and the population's best position. The mathematical model is as follows: (8) (9) in As the inertia weighting factor, and Individual and social learning factors, respectively Let the position vector of the particle be... Let be the position vector of the particle in the next instant. Let be the particle's velocity vector. For the individual's historical best position, The optimal position for the global population. and It is a random number between (0 and 1).
[0016] Instead of globally random point placement, a three-dimensional straight-line reference axis is established between the beginning and end nodes of the task. This spatial straight line connecting the start and end points of the task is translated to the intermediate value of the height constraint, serving as the central reference line for the overhead power transmission channel. For any point in space, its local environmental pressure is defined. for: (10) Three-dimensional Gaussian perturbations are applied at each node of the central baseline to generate the initial population. The perturbation amplitudes in the horizontal and vertical directions are exponentially adjusted by the local environmental pressure at that location, as follows: (11) (12) (13) in and These are the horizontal and vertical unit vectors orthogonal to the line of sight, respectively. Environmental pressure sensitivity coefficient Let the standard deviation be the Gaussian perturbation in the horizontal direction. Let the standard deviation be the Gaussian perturbation in the vertical direction. and These are the preset standard deviations of the maximum horizontal and vertical disturbances, respectively. The initial population particle space coordinates are used. The coordinates of the nodes on the central baseline. With a mean of 0 and a variance of The Gaussian normal distribution is used to generate random perturbations.
[0017] This mechanism allows the initial population to stay close to the safety axis in densely built-up areas and autonomously expand its detection range in open areas.
[0018] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S5, is as follows: During the population's iterative evolution, the following parameter adjustment operations are performed independently for each particle, with terrain-driven adaptive adjustment applied to the inertia weight, and the adjustment rate being: (14) in For adaptive inertia weights, and Preset lower and upper limits for inertia weights. For terrain risk response gain, This represents the highest risk penalty value for the current population. This strategy transforms time-driven processes into space-driven processes suitable for electric scenarios, enabling each particle to possess independent spatial awareness, giving the aircraft spatial situational awareness, and achieving "automatic deceleration and fine-tuning upon encountering obstacles, and high-speed leapfrogging in open areas."
[0019] The learning factor is adaptively adjusted using energy-efficient closed-loop feedback, with an adjustment rate of... for: (15) in The energy consumption of the current particle. This serves as the baseline energy consumption for the globally optimal trajectory. Based on this state feedback, and The dynamic regulation law is designed as follows: (16) (17) in As the basic learning factor, it ensures the algorithm's most basic convergence capability. and These are the self-feedback gain and the social feedback gain, respectively.
[0020] The ratio of the current particle's flight energy consumption to that of the current globally optimal particle is used as an energy consumption state feedback indicator to drive nonlinear linkage adjustment between individual cognitive factors and social sharing factors. When the energy consumption of a certain particle is significantly higher than the global optimal level, the system weakens the particle's dependence on its own experience while enhancing its learning intensity of the group's optimal experience, forcing it to move towards an energy-saving path. When the energy consumption of a certain particle has reached or exceeded the global optimal level, the system strengthens its maintenance of its own excellent experience.
[0021] The above-described method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, specifically the operation of S6, is as follows: The optimal discrete sequence coordinates of the output after S5 optimization are extracted, and a cubic uniform B-spline curve is introduced for continuity optimization. At the same time, combined with the safety buffer boundary reserved by the collision avoidance cost in S3, the inherent inward contraction of the B-spline curve at narrow curves is effectively suppressed, avoiding the smoothed path from passing through obstacles. Finally, a high-quality flight path command that meets the continuity requirement of the second derivative of the flight controller is output.
[0022] Compared with the prior art, the beneficial effects of the present invention are: 1) This invention constructs a high-fidelity energy consumption model, which replaces the geometric weighted estimation method lacking physical basis in the traditional method with a power consumption model based on rotor aerodynamics theory. It can accurately characterize the nonlinear power response characteristics of the aircraft during low-speed obstacle avoidance and high-altitude climb, thus eliminating the problem of energy consumption estimation bias from the source. 2) This invention enhances particle swarm optimization with multiple strategies to achieve rapid convergence and high-precision optimization. The initial solution generation strategy, which combines overhead power channels with spatial dangerous gradient responses, actively constrains the effective search range of the initial swarm to the vicinity of feasible channels, completely eliminating the path point entanglement and initial collision problems commonly found in traditional methods. The dual closed-loop parameter adaptive adjustment mechanism endows particles with the ability to respond in real time to local spatial features and their own energy states, significantly accelerating the optimization and convergence process. 3) This invention ensures that the output path has second-order geometric continuity by using cubic uniform B-spline smoothing, which satisfies the yaw rate and curvature continuity constraints of the rotorcraft. 4) The energy consumption state feedback mechanism of this invention guides particles to actively choose macroscopic spatial channels with lower power loss during ascent / descent, rather than simply pursuing the shortest projection path in a geometric sense; 5) This invention designs multiple safety assurance mechanisms. The exponential risk expansion model in the extreme collision avoidance cost function is combined with the hard constraint of electrical safety distance to ensure that the aircraft always maintains a safe distance from various obstacles; the safety corridor overflow cost constraint limits the flight altitude to a preset range, eliminating the dual risks of ultra-low-altitude collisions and encountering severe weather at excessively high altitudes. Attached Figure Description
[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0024] Figure 1 This is a flowchart illustrating the enhanced particle swarm vehicle inspection path planning in an embodiment.
[0025] Figure 2 The diagram shows the generation of the initial solution for the enhanced particle swarm in this embodiment.
[0026] Figure 3 This is a 3D main view of the enhanced particle swarm inspection path planning in the embodiment.
[0027] Figure 4 This is a three-dimensional side view of the enhanced particle swarm inspection path planning in the embodiment.
[0028] Figure 5 This is a two-dimensional view of the enhanced particle swarm inspection path planning in the embodiment.
[0029] Figure 6 The image shows the enhanced particle swarm convergence curve in the example. Detailed Implementation
[0030] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, embodiments of the present invention are described below in detail with reference to the accompanying drawings. This application provides a method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios; its framework diagram is shown below. Figure 1 This includes the following steps: S1: Construct a heterogeneous geometric three-dimensional spatial model for complex operational airspace in urban low-altitude power line inspection application scenarios. S2: Construct a nonlinear power loss evaluation system based on rotor aerodynamics principles; S3: Construct the optimization objective cost function for multi-dimensional coupling of packages; S4: Initial solution generation based on overhead power transmission channels and airspace hazard gradient response; S5: Adaptive adjustment of dual closed-loop control parameters that integrate terrain perception and energy efficiency feedback; S6: Use cubic uniform B-splines to optimize the continuity of the planned path and output an executable smooth path.
[0031] S1: For urban low-altitude power line inspection applications, a heterogeneous geometric 3D spatial model of the complex operational airspace is constructed. The specific steps are as follows: Step 1-1: For entities with regular orthogonal geometric features that exist in large numbers within the operational airspace, such as residential complexes and commercial buildings, a three-dimensional hexahedron (i.e., a rectangular outer envelope box) is used for voxel abstraction. The physical spatial form and absolute positional attributes of this hexahedron are uniquely determined by four core parameters: the two-dimensional projection coordinates corresponding to the centroid of its base, the length and width dimensions of the base, and the elevation of the facade. For vertical structures with significant height-to-diameter ratios, such as high-voltage transmission towers and signal relay antenna base stations, a vertical cylindrical envelope model is configured for physical feature mapping. The spatial boundary of such facilities is defined by the coordinates of its geometric center on its base, the characteristic value of the radius of its base circle, and the elevation of its column top along the Z-axis.
[0032] S2: Construct a nonlinear power loss evaluation system based on rotor aerodynamics principles. The specific steps are as follows: Step 2-1: Construct a nonlinear power consumption model based on rotor aerodynamics principles. The instantaneous total power of the aircraft during flight between adjacent path points. The expression is: (18) in and These represent the shape drag power and induced power of a rotorcraft in hovering mode, respectively. The horizontal forward speed of a rotorcraft. The linear velocity at the rotor blade tip. The average induced airflow velocity during hovering. The drag coefficient of the fuselage. Atmospheric density, For rotor realism, For the area swept by the propeller disk, The total mass of the aircraft and its payload. This is the acceleration due to gravity.
[0033] The instantaneous total power consumption in formula (18) consists of the following four physical components: (a) Type drag power: used to overcome the surface friction drag generated when the rotor blades rotate in the air, and its value is proportional to the square of the horizontal speed of the aircraft; (b) Induced power: used to accelerate the airflow downward to generate lift against gravity, and its value decreases monotonically with increasing horizontal velocity; (c) Exhaust drag power: used to overcome the air resistance experienced by the aircraft fuselage and onboard equipment during horizontal motion, and its value is proportional to the cube of the horizontal velocity; (d) Climbing power: This component is used to overcome gravity and achieve a certain height. The condition-triggered method is defined as follows: (19) in This refers to the vertical ascent rate.
[0034] Step 2-2: Calculate the global cumulative energy consumption of the energy consumption model constructed above. The calculation involves time-discretizing and accumulating the instantaneous power for each flight segment. (20) in The flight time between the two waypoints. It is the total number of flight segments. This is the flight segment index number.
[0035] S3: Construct the optimization objective cost function for multi-dimensional coupling of the package. The specific steps are as follows: Step 3-1: To achieve the cooperative optimality of multiple objectives, the optimization objective cost function is constructed as follows: (twenty one) The design incorporates a comprehensive evaluation index system with four independent cost components: (a) Cost of reduced battery life ( ): Mapped to the total power consumption calculated in S2 ; (b) Cost of Spatial Transition ( ): This evaluates the curvature abrupt changes in the path point sequence through vector dot product, used to filter out frequent yaws; it is used to suppress unnecessary direction abrupt changes, and its calculation method is as follows: (twenty two) in Indicates by the first The path point points to the first The direction vectors of the path points Indicates by the first The path point points to the first The direction vectors of the path points This represents the total number of path points.
[0036] (c) Cost of extreme collision avoidance The method employs a dense sampling approach combined with an exponential repulsive field to penalize the distance between each sampling point and the surface of the nearest obstacle. If a sampling point enters the interior of an obstacle, a maximum constant penalty is applied. The calculation method is as follows: (twenty three) in As the baseline intensity factor for obstacle avoidance penalty, The distance is the tangential distance from the sampling point to the surface of the nearest obstacle. As a safety buffer threshold, This is a very large penalty value.
[0037] (d) Cost of spillover from the safe corridor ): This restricts aircraft to fly within compliant altitudes. Penalties are imposed when aircraft exceed the upper or lower limits of the designated area. The calculation method is as follows: (twenty four) in The height of the sampling point. and This is the height boundary of the safe corridor for the flight area.
[0038] These are the weights of each loss cost. The above four costs are weighted and summed using their respective weighting coefficients to form a comprehensive evaluation index.
[0039] S4: Initial solution generation based on overhead power transmission channels and airspace hazard gradient response, the specific steps are as follows: Step 4-1: The standard particle swarm adjusts its state based on the individual's historical best position and the population's best position. The mathematical model is as follows: The Standard Particle Swarm Optimization (SSO) system adjusts its state based on information about the individual's historical best position and the population's best position. The mathematical model is as follows: (25) (26) in As the inertia weighting factor, and Individual and social learning factors, respectively Let the position vector of the particle be... Let be the position vector of the particle in the next instant. Let be the particle's velocity vector. For the individual's historical best position, The optimal position for the global population. and It is a random number between (0 and 1).
[0040] Step 4-2: Construct a spatial straight line connecting the start and end points of the task, and translate it to the intermediate value of the height constraint as the center baseline of the overhead power transmission channel. For any point in space, define its local environmental pressure. for: (27) Step 4-3: Apply three-dimensional Gaussian perturbations at each node location along the central baseline to generate the initial population. The perturbation amplitudes in the horizontal and vertical directions are exponentially adjusted by the local environmental pressure at that location, as follows: (28) (29) (30) in and These are the horizontal and vertical unit vectors orthogonal to the line of sight, respectively. Environmental pressure sensitivity coefficient Let the standard deviation be the Gaussian perturbation in the horizontal direction. Let the standard deviation be the Gaussian perturbation in the vertical direction. and These are the preset standard deviations of the maximum horizontal and vertical disturbances, respectively. The initial population particle space coordinates are used. The coordinates of the nodes on the central baseline. With a mean of 0 and a variance of The Gaussian normal distribution is used to generate random perturbations.
[0041] This mechanism allows the initial swarm to stay close to the safety axis in densely built-up areas and autonomously expand its detection field of view in open areas. The particle swarm initialization experiment diagram provided in the embodiment is as follows: Figure 2 As shown.
[0042] S5: Adaptive adjustment of dual closed-loop control parameters integrating terrain perception and energy efficiency feedback, the specific steps of which are as follows: Step 5-1: During the population iterative evolution process, the following parameter adjustment operations are performed independently for each particle. Terrain-driven adaptive adjustment is used for the inertia weight, with an adjustment rate of: (31) in For adaptive inertia weights, and Preset lower and upper limits for inertia weights. For terrain risk response gain, This represents the highest risk penalty value for the current population. This strategy transforms time-driven processes into space-driven processes suitable for electric scenarios, enabling each particle to possess independent spatial awareness, giving the aircraft spatial situational awareness, and achieving "automatic deceleration and fine-tuning upon encountering obstacles, and high-speed leapfrogging in open areas."
[0043] The collision avoidance cost of current particles is used as a quantitative indicator of the density of spatial obstacles. When this indicator increases, the inertial weight decreases exponentially to enhance the aircraft's ability to fine-tune its local position in dangerous areas. When this indicator is low, the inertial weight remains at a large value to maintain global search efficiency.
[0044] Step 5-2: Adaptive adjustment of the learning factor using energy-efficient closed-loop feedback, adjustment rate for: (32) in The energy consumption of the current particle. The baseline energy consumption for the globally optimal trajectory.
[0045] Step 5-3: Based on the status feedback in Step 5-2, and The dynamic regulation law is designed as follows: (33) (34) in As the basic learning factor, it ensures the algorithm's most basic convergence capability. and These are the self-feedback gain and the social feedback gain, respectively.
[0046] The ratio of the current particle's flight energy consumption to that of the current globally optimal particle is used as an energy consumption state feedback indicator to drive nonlinear linkage adjustment between individual cognitive factors and social sharing factors. When the energy consumption of a certain particle is significantly higher than the global optimal level, the system weakens the particle's dependence on its own experience while enhancing its learning intensity of the group's optimal experience, forcing it to move towards an energy-saving path. When the energy consumption of a certain particle has reached or exceeded the global optimal level, the system strengthens its maintenance of its own excellent experience.
[0047] S6: The planned path is continuously optimized using cubic uniform B-splines, outputting an executable smooth path. The specific steps are as follows: Step 6-1: For the optimal discrete path point sequence obtained after optimization in S5, a cubic uniform B-spline curve is introduced for continuity optimization. Simultaneously, combined with the safety margin reserved by the extreme collision avoidance cost in S3, the B-spline curve is effectively suppressed from inward contraction when turning in narrow channel regions, preventing the smoothed path from passing through obstacles, ultimately outputting a continuous and executable smooth path. The enhanced particle swarm inspection path planning 3D main view provided in the embodiment is as follows: Figure 3 As shown, the three-dimensional side view of the enhanced particle swarm inspection path planning is as follows: Figure 4 As shown, the enhanced particle swarm path planning two-dimensional view is as follows: Figure 5As shown, the convergence curve of the enhanced particle swarm is as follows: Figure 6 As shown.
[0048] The above results demonstrate that the proposed method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios can effectively plan a safe, feasible, and low-energy-consumption power line inspection path in dense urban building environments.
[0049] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.
Claims
1. A method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios, characterized in that, Includes the following steps: S1: Construct a heterogeneous geometric three-dimensional spatial model for complex operational airspace in urban low-altitude power line inspection application scenarios. S2: Construct a nonlinear power loss evaluation system based on rotor aerodynamics principles; S3: Construct the optimization objective cost function for multi-dimensional coupling of packages; S4: Initial solution generation based on overhead power transmission channels and airspace hazard gradient response; S5: Adaptive adjustment of dual closed-loop control parameters that integrate terrain perception and energy efficiency feedback; S6: Use cubic uniform B-splines to optimize the continuity of the planned path and output an executable smooth path; The operation of S1 is specifically as follows: A hybrid geometric obstacle representation method is adopted: for ground structures with regular edge contours, a three-dimensional orthogonal hexahedron is used for enclosure equivalent; for tall pole-like structures, a vertical cylindrical bounding box is used for enclosure equivalent; in terms of spatial layout, the above two types of geometric shapes are arranged in a pseudo-random staggered manner to simulate the spatial characteristics of dense buildings and narrow passable corridors in a real urban canyon environment. The specific operation of S2 is as follows: Construct a nonlinear power consumption model based on rotor aerodynamics principles to determine the instantaneous total power of the aircraft during flight between adjacent path points. The expression is: (1) in and These represent the shape drag power and induced power of a rotorcraft in hovering mode, respectively. The horizontal forward speed of a rotorcraft. The linear velocity at the rotor blade tip. The average induced airflow velocity during hovering. The drag coefficient of the fuselage. Atmospheric density, For rotor realism, For the area swept by the propeller disk, The total mass of the aircraft and its payload. It is the acceleration due to gravity; Rise Power The condition-triggered method is defined as follows: (2) in Vertical climb rate; Global mission cumulative energy consumption By performing time-discretized accumulation of the instantaneous power for each flight segment: (3) in The flight time between the two waypoints. It is the total number of flight segments. For flight segment index number; The specific operation of S3 is as follows: To achieve the optimal balance among multiple objectives, the optimization objective cost function is constructed as follows: (4) The design incorporates the cost of reduced battery life as a global factor. The cost of spatial transformation Extreme collision avoidance costs Safe corridor spillover cost A comprehensive evaluation index system consisting of four independent cost components. It is the weight of each loss cost, which is weighted and summed through the corresponding weight coefficients to form a comprehensive evaluation index.
2. The method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios according to claim 1, characterized in that, The specific operation of S4 is as follows: Construct a spatial straight line connecting the start and end points of the task, and translate it to the middle value of the height constraint as the central baseline of the overhead power channel. For any point in space, define its local environmental pressure as the reciprocal of the distance to the surface of the nearest obstacle. Three-dimensional Gaussian perturbations are applied at each node of the central baseline to generate the initial population. The perturbation amplitudes in the horizontal and vertical directions are exponentially adjusted by the local environmental pressure at that location, as follows: (5) (6) (7) in and These are the horizontal and vertical unit vectors orthogonal to the line of sight, respectively. Environmental pressure sensitivity coefficient For local environmental pressure, Let the standard deviation be the Gaussian perturbation in the horizontal direction. Let be the standard deviation of the Gaussian perturbation in the vertical direction. and These are the preset standard deviations of the maximum horizontal and vertical disturbances, respectively. The initial population particle space coordinates are used. The coordinates of the nodes on the central baseline. With a mean of 0 and a variance of The Gaussian normal distribution is used to generate random perturbations; This mechanism allows the initial population to stay close to the safety axis in densely built-up areas and autonomously expand its detection range in open areas. In densely built-up, high-risk areas, the disturbance amplitude is reduced to ensure the initial population stays close to the central baseline for safety; in open areas, the disturbance amplitude is amplified to enhance the global spatial coverage of the initial population. The specific operation of S5 is as follows: During the population's iterative evolution, the following parameter adjustment operations are performed independently for each particle, with terrain-driven adaptive adjustment applied to the inertia weight, and the adjustment rate being: (8) in For adaptive inertia weights, and Preset lower and upper limits for inertia weights. For terrain risk response gain, This represents the highest risk penalty value for the current population. Transforming time-driven operation into space-driven operation suitable for electric scenarios endows the aircraft with spatial situational awareness, enabling "automatic deceleration and fine-tuning upon obstacle encounter, and high-speed leapfrogging in open areas"; The learning factor is adaptively adjusted using energy-efficient closed-loop feedback, with an adjustment rate of... for: (9) in The energy consumption of the current particle. The baseline energy consumption for the globally optimal trajectory; Based on this state feedback, the individual learning factor Social learning factors The dynamic regulation law is designed as follows: (10) (11) in As the basic learning factor, it ensures the algorithm's most basic convergence capability. and These are the self-feedback gain and the social feedback gain, respectively. Using the ratio of particle energy consumption to global optimal energy consumption as a feedback indicator, the individual cognitive factor and social sharing factor are dynamically adjusted: when energy consumption is significantly high, self-experience dependence is weakened and learning from the group's optimal is enhanced; when energy consumption has reached or is better than the global optimal, self-experience retention is strengthened.
3. The method for power line inspection path planning of rotorcraft based on enhanced particle swarm optimization in complex low-altitude scenarios according to claim 1, characterized in that, The specific operation of S6 is as follows: For the optimal discrete path point sequence obtained after S5 optimization, a cubic uniform B-spline curve is introduced for continuity optimization. At the same time, combined with the safety margin reserved by the collision risk cost in S3, the inherent inward contraction of the B-spline curve when turning in narrow channel areas is effectively suppressed, and finally a smooth path that can be continuously executed is output.
