Power transmission line adaptive inspection method, unmanned aerial vehicle, medium, equipment and product

Abstract

The invention discloses a power transmission line adaptive inspection method, an unmanned aerial vehicle, a medium, equipment and a product, and relates to the technical field of power transmission line inspection, and the method comprises the steps: controlling the unmanned aerial vehicle to carry out level flight cruise along a power transmission line, and shooting an inspection image for preliminary detection; when the power transmission line fault is preliminarily detected, the attitude of the unmanned aerial vehicle is switched to a hovering state with the height unchanged; local abnormal region extraction and abnormal region space positioning are carried out in a hovering state; taking the abnormal area space positioning as a target, keeping the relative position between the unmanned aerial vehicle and the target and the angle of aligning the optical axis of the unmanned aerial vehicle to the target unchanged, performing fixed-point detection on the target, and recovering the level flight cruise after the fixed-point detection is finished. According to the method, the robustness, the parameter adaptability and the transition state control precision of a flight control algorithm in a high-disturbance environment are improved.

Description

Technical Field

[0001] This invention relates to the field of power transmission line inspection technology, and in particular to adaptive inspection methods, drones, media, equipment and products for power transmission lines. Background Technology

[0002] Transmission lines, as crucial infrastructure for long-distance energy transmission in power systems, are typically located in areas with complex terrain and variable climates. Their operational safety directly impacts the reliability and economy of the power grid. Due to prolonged exposure to the natural environment, transmission lines and their components (such as conductors, insulators, hardware, and clamps) are susceptible to mechanical damage or electrical hazards caused by wind vibration, icing, lightning strikes, external impacts, and aging corrosion. Timely and accurate line inspections are essential for ensuring the safe operation of the power system.

[0003] Traditional power transmission line inspections rely heavily on manual ground operations or helicopter inspections, which suffer from low efficiency, high costs, and significant safety risks. In recent years, with the development of drone technology, drone-based power transmission line inspections have gradually become the mainstream method. Drones offer advantages such as high mobility, wide coverage, and high data acquisition efficiency, enabling them to collect images and detect potential hazards on power transmission lines without interrupting power, providing fundamental data support for intelligent operation and maintenance.

[0004] However, power transmission line inspection operations are characterized by significant scene complexity. On the one hand, power transmission lines are often accompanied by uncertain environmental factors such as sudden changes in wind speed, airflow disturbances, and undulating terrain. On the other hand, inspection tasks require UAVs to have long endurance and high-speed cruising capabilities to quickly scan the line section, while also requiring them to quickly switch to low-speed or hovering modes after detecting suspected faults for fixed-point photography and detailed observation. Although fixed-wing UAVs have high cruising efficiency, they are difficult to hover stably in the air; multi-rotor UAVs can hover at fixed points, but their endurance is short and their coverage distance is limited. Therefore, how to achieve composite flight control that combines the characteristics of fixed-wing and rotary-wing UAVs, and enable UAVs to adaptively switch flight modes according to the inspection task and environmental conditions, has become an important technical issue in the field of intelligent inspection of power transmission channels.

[0005] Existing flight control systems for fixed-wing UAVs are mostly based on linear PID controllers or gain-adaptive PID controllers. These methods are simple in structure and have a fast response, making them suitable for attitude maintenance under stable airflow conditions. However, when UAVs operate in highly disturbed environments such as power transmission corridors, their fixed gain parameters struggle to adapt to sudden wind changes, airflow swirling, and electromagnetic interference, often leading to attitude overshoot, oscillations, and track deviations. To improve adaptability, some studies have introduced fuzzy logic self-tuning or neural network-based gain adjustment methods. However, these methods generally require a large number of training samples or expert rules, and are prone to high computational load and response lag when the flight environment changes drastically.

[0006] For hybrid unmanned aerial vehicles (UAVs) with hovering capabilities, some studies employ nonlinear control strategies such as nonlinear state feedback linearization control, sliding mode control, and adaptive backstepping control to achieve transition control from fixed-wing to vertical attitude. While these algorithms theoretically improve system robustness and accuracy, their modeling relies on complete flight mechanics parameters and accurate aerodynamic characteristics. However, under complex airflow and multi-source disturbance conditions, these parameters are difficult to acquire or update in real time. Furthermore, most linearization methods assume a constant flight envelope and lack stability control mechanisms for the transition state (quasi-stall region), leading to attitude drift or instability during the transition between lateral flight and hovering.

[0007] In summary, existing flight control algorithms still have significant shortcomings in terms of robustness, parameter adaptability, and transient control accuracy under high disturbance environments, making it difficult to meet the requirements of long-distance, high-precision inspection of power transmission lines. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings of existing flight control algorithms in terms of robustness, parameter adaptability, and transient control accuracy under high-disturbance environments. A method for adaptive inspection of transmission lines is proposed, comprising the following steps: S1. Control the drone to fly level along the power transmission line and take inspection images for preliminary testing; S2. When a power line fault is initially detected, switch the drone's attitude to a hovering state with constant altitude. S3. Extract and spatially locate local anomalies while hovering. S4. With the abnormal area spatial positioning as the target, keep the relative position between the UAV and the target and the angle of the UAV's optical axis aligned with the target unchanged, perform fixed-point detection on the target, and resume level flight cruise after the fixed-point detection is completed.

[0009] Furthermore, during level flight cruise, the roll and pitch angles are controlled as follows: The PID control equation for the roll angle is:

[0010] in, This indicates the output of the roll angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the expected roll angle and the actual roll angle. express The first derivative; The roll angle limiting function is:

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017] in, Limiting the roll angle, To control the torque, For rolling inertia, This represents the maximum value of the second derivative of the actual roll angle. Indicates the control moment of the control surface. Indicates the equivalent rolling torque. The rolling moment coefficient represents the variation of the aileron deflection angle. q represents the aileron deflection angle, q represents the dynamic pressure, S represents the wing area, and b represents the wingspan. This indicates the distance from the propeller to the fuselage's center of mass. The lever arm of direction, The direction is perpendicular to the plane of symmetry of the fuselage and points towards the right wing. This indicates the thrust difference between the left and right propellers. This represents a constant obtained from calibration based on the wing mounting position and the motor thrust curve. The dynamic weights obtained from fuzzy scheduling This represents an input consisting of three types of disturbances. Jointly determine the weight The nonlinear mapping relationship This represents three types of disturbances. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor. The first derivative represents the lateral deviation of the UAV from its flight path; The PID control equation for pitch angle is:

[0018] in, This represents the output of the pitch angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the desired pitch angle and the actual pitch angle. express The first derivative; right After applying feedforward compensation, the final pitch angle control value is obtained as follows:

[0019]

[0020] in, This indicates the final pitch angle control value. This is the pitch angle feedforward compensation amount. This represents the pitch angle control feedforward coefficient. This represents the feedforward compensation function based on wind disturbance and electromagnetic disturbance. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor.

[0021] Furthermore, , and according to The values ​​are set in segments, represented as follows:

[0022]

[0023]

[0024] in, Represents the nth segment , and The set value, express The threshold of the 1st, 2nd, ..., nth segment. This represents the updated segment threshold after fuzzy inference. This represents the baseline threshold for the i-th error segment. This indicates that based on three types of disturbance inputs The adjustment function for adaptively correcting the segmented control threshold. This represents the threshold correction amount for the i-th segment. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor. The first derivative represents the lateral deviation of the UAV from its flight path.

[0025] Furthermore, a transitional linear decoupling control algorithm is used to switch the UAV attitude to a hovering state, as detailed below: The actual equations of motion for the organism are:

[0026] in, Let x be the first derivative of the control object in the transition state. , These represent the actual roll angle, pitch angle, and yaw angle, respectively. These represent the angular velocities along the three axes of the body coordinate system. This refers to the portion of the drone's attitude that evolves naturally over time when there is no control input. A function that describes how control inputs affect the system state. This represents the input for body attitude control. Indicates the deflection angle of the aileron. This indicates the deflection angle of the elevator. Indicates the deflection angle of the rudder. This indicates the thrust difference between the left and right propellers; The mapping from u to the body torque is: M=Bu in, Indicates the torque of the machine body. These represent the roll, pitch, and yaw moments, respectively.

[0027] Where B is the torque distribution matrix, Let B be the diagonal principal component. The coupling components that need to be compensated for in B; The decoupled u is represented as:

[0028]

[0029]

[0030]

[0031]

[0032] in, This represents the ideal input for body attitude control. These represent the target moments for roll, pitch, and yaw, respectively. Indicates the expected roll angle. Indicates the desired pitch angle. Indicates the desired yaw angle. , and These are the proportional gains for roll, pitch, and yaw angles, respectively. , and These represent the differential gains with respect to roll, pitch, and yaw angles, respectively. The height balance equation is:

[0033] Where T represents the total thrust, m is the mass of the UAV, and g is the acceleration due to gravity. This represents a correction amount.

[0034] Furthermore, S3 specifically refers to: Extract the main axis of the power transmission line from the inspection images; Perform a projection flattening transformation on the image coordinates along the principal axis; On the flattened strip image, a pre-stored standard component image template is called, and the matching degree between the strip image and the standard component image template is calculated using normalized cross-correlation. For matching regions where the matching degree exceeds a threshold, calculate the local brightness gradient and structural symmetry index to determine whether the matching region is abnormal. For each matching region, calculate the local brightness variance and edge direction divergence to define an anomaly score, expressed as:

[0035] Where Q represents the anomaly score, Represents the weight parameters. Indicates the local brightness variance. This represents the local brightness reference variance. Indicates the divergence along the edge direction. Indicates the reference divergence in the edge direction; Based on the UAV attitude matrix and camera intrinsic and extrinsic parameters, outliers in the image coordinate system are mapped to the body coordinate system to achieve spatial positioning.

[0036] Furthermore, the S4 implementation method is as follows: obtain the attitude correction amount, thrust correction amount and safety displacement command of the UAV, and realize the UAV hovering control based on the attitude correction amount, thrust correction amount and safety displacement command of the UAV; The attitude correction for the UAV is expressed as:

[0037]

[0038]

[0039]

[0040]

[0041] in, This indicates the attitude correction amount for the drone. Let J represent the attitude correction coefficient of the UAV, and let J represent the objective function for attitude correction of the UAV. Indicates the attitude of the drone. They represent and The weight, and Let represent the costs of the optical axis angle constraint and the position constraint, respectively, and d represent the distance between the UAV and the target. This indicates the expected safe distance between the drone and the target. Indicates the current optical axis direction of the camera. Indicates the direction of the target. This indicates the position of the target point in the machine's coordinate system. The thrust correction for the UAV is:

[0042] in, This indicates the thrust correction amount for the drone. This indicates the thrust correction factor for the drone; The safe displacement command for the drone is:

[0043] in, Indicates the safe displacement command for the drone. A unit vector representing the direction of a power transmission line. This represents the distance error adjustment gain coefficient along the direction of the conductor.

[0044] This invention also proposes an unmanned aerial vehicle (UAV) for implementing the above-mentioned adaptive inspection method for power transmission lines, comprising: Airframe structure, twin front propeller drive system, fixed-wing lift system, flight control system, image acquisition module, and adaptive inspection and processing module; The flight control system is configured to perform two operating modes: horizontal flight and hovering. In horizontal flight mode, the fixed-wing lift system provides lift, while the twin-head propeller drive system provides forward thrust. In hovering mode, the twin-head propeller drive system generates lift. The adaptive inspection processing module is used to generate control signals for the flight mode based on the inspection images and environmental status signals, and the control signals control the flight control system.

[0045] The present invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described adaptive inspection method for power transmission lines.

[0046] The present invention also proposes an electronic device, including a processor and a memory, wherein the processor and the memory are interconnected, wherein the memory is used to store a computer program, the computer program including computer-readable instructions, and the processor is configured to invoke the computer-readable instructions to execute the above-described adaptive inspection method for power transmission lines.

[0047] The present invention also proposes a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-described adaptive inspection method for power transmission lines.

[0048] The beneficial effects of the technical solution provided by this invention are: (1) When the UAV is flying horizontally, a PID controller is used to control the roll angle and pitch angle. A fuzzy scheduler with the lateral wind shear, the electromagnetic disturbance intensity estimate and the lateral trajectory deviation rate as input is used to construct an adaptive gain tuning mechanism based on the disturbance source, improve the robustness of the flight control algorithm in a high disturbance environment, and ensure the heading maintenance capability and image acquisition stability in the horizontal flight state.

[0049] (2) By defining a safe attitude range for the UAV, the nonlinear attitude dynamics are linearized in this range, and a set of torque distribution matrices are used to cancel the coupling between attitude axes. At the same time, the total thrust and attitude angle are subject to the cooperative constraint of gravity direction force balance, thus completing the attitude switch from level flight cruise to hovering and improving the transition state control accuracy. Attached Figure Description

[0050] Figure 1 This is a flowchart of the adaptive inspection method for power transmission lines according to an embodiment of the present invention; Figure 2 This is a block diagram of an electronic device according to an exemplary embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0052] The flowchart of the adaptive inspection method for transmission lines according to an embodiment of the present invention is as follows: Figure 1Specifically, it includes the following steps: S1. Control the drone to fly level along the power transmission line and take inspection images for preliminary testing.

[0053] Preferably, the present invention employs an adaptive multi-segment gain tuning control algorithm based on transmission channel disturbance compensation to control the roll angle and pitch angle.

[0054] First, establish a common ground machine coordinate system. The origin O is located at the center of mass. The wing chord, parallel to the fuselage axis or plane of symmetry, points forward (towards the nose). Perpendicular to the plane of symmetry, pointing to the right wing; Within the plane of symmetry of the aircraft, perpendicular to Pointing downwards. This conforms to the rules for establishing a right-handed system.

[0055] The expected roll angle for horizontal flight is The actual roll angle is Then the roll angle error for:

[0056] Desired pitch angle is The actual pitch angle is Then the pitch angle error for:

[0057] To characterize the lateral trajectory drift of a UAV in a long power transmission line corridor scenario, the reference centerline of the power transmission line is assumed to be a curve Γ( in the geographic coordinate system). ), projecting the drone's current position onto Γ( The nearest point is obtained on ) Then the lateral deviation It can be defined as:

[0058] in, This represents the spatial position vector of the UAV at a certain moment.

[0059] Since there is often lateral wind shear along the inter-tower direction near transmission lines, this disturbance can be approximated as acting on... Equivalent disturbance force in direction Let the crosswind component estimated jointly by the airborne pitot tube and IMU be Then, a simple first-order approximate model of the perturbation to the attitude angle can be established as follows:

[0060] in, for The first derivative, This represents the equivalent wind disturbance gain determined by the wing aspect ratio, the position of the forward propeller, and the current angle of attack.

[0061] To enable the controller to adapt to environmental disturbances, the attitude error is divided into multiple error intervals, and a set of PID gains is configured for each interval. Let the segmented threshold for the roll angle error be... ,satisfy Then the reference gain table for the roll controller can be written as:

[0062] Unlike conventional piecewise PID controllers, the thresholds for each segment here are not fixed constants, but adjustable parameters influenced by the characteristics of power transmission channel disturbances. Therefore, this invention designs a fuzzy scheduler with "lateral wind shear," "electromagnetic disturbance intensity estimate," and "lateral trajectory deviation rate" as inputs. Let the fuzzy input quantities be... for:

[0063] The output of fuzzy inference is the segmented threshold correction amount. and differential thrust compensation coefficient , can be represented as:

[0064] The actual segmentation threshold after fuzzy inference is updated as follows:

[0065] in, Represents the nth segment , and The set value, express The threshold of the 1st, 2nd, ..., nth segment. This represents the updated segment threshold after fuzzy inference. This represents the baseline threshold for the i-th error segment, which is set as needed. This indicates that based on three types of disturbance inputs An adaptive adjustment function for piecewise control thresholds is implemented based on fuzzy inference. The fuzzy inference system uses... The input variable is the perturbation, and the output variable is the threshold value of each segment. This represents the threshold correction amount for the i-th segment. These represent three types of disturbances. The thrust difference compensation weight is a dynamic weight obtained from fuzzy scheduling. This represents an input consisting of three types of disturbances. Joint decision The nonlinear mapping relationship is realized based on fuzzy inference. The fuzzy inference system uses... As the perturbation input variable, with The dynamic value is the output variable; The lateral wind component can be estimated jointly by the airborne pitot tube and the IMU. The estimated electromagnetic disturbance intensity of the conductor can be obtained from the variance of the attitude disturbance when the airborne magnetic compass approaches the conductor. The first derivative represents the lateral deviation of the UAV from its flight path.

[0066] To achieve rapid assistance in attitude control from the fore-wing twin propellers, this invention divides the final control quantity into control surface control quantities. With thrust difference control quantity Δ The system consists of two parts. The control quantity for the control surface is mainly generated by a segmented PID controller, while the thrust difference control quantity is obtained by adjusting the attitude error and its rate of change through fuzzy gain.

[0067] Let the classic PID output of roll control be:

[0068] in, This indicates the output of the roll angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the expected roll angle and the actual roll angle. express The first derivative.

[0069] Considering the roll disturbance caused by lateral wind shear, an additional thrust difference feedforward compensation term based on the error change rate needs to be introduced. Let the thrust of the left and right propellers be respectively... and The thrust difference is defined as:

[0070] Based on the requirements for airframe roll control, the thrust difference is set as follows:

[0071] in, This indicates the thrust difference between the left and right propellers. This represents a constant obtained by calibration based on the wing mounting position and the motor thrust curve.

[0072] Since the left and right propellers are mounted on the leading edge of the wing, the thrust difference ΔT will be felt on the fuselage. shaft and A torque is generated simultaneously on the shaft, which needs to be mapped to the roll control channel. Let its equivalent roll torque be... Then we have the following formula:

[0073] in, This indicates the distance from the propeller to the fuselage's center of mass. The lever arm of direction. The final roll control is achieved jointly by the control surface control torque and the thrust differential control torque, and can be written as:

[0074] in, To control the torque, Indicates the control moment of the control surface. This represents the equivalent rolling torque.

[0075] The control moment of the control surface can be expressed using the conventional control surface efficiency model as follows:

[0076] in, The rolling moment coefficient, which represents the change in aileron deflection angle, is an aerodynamic derivative function. with aileron deflection angle They exhibit an approximately linear relationship. q represents the aileron deflection angle, q represents the dynamic pressure, S represents the wing area, and b represents the wingspan.

[0077] To suppress control overshoot and frequent switching that may occur under large disturbance conditions, this invention further incorporates a safety limiter at the control output. Let the upper limit of the allowable roll acceleration be... max Then, based on the current rolling inertia and control torque Relationship:

[0078] The following limiting function can be constructed:

[0079] in, Limiting the roll angle, To control the torque, For rolling inertia, This is the maximum value of the second derivative of the actual roll angle.

[0080] Similarly, to ensure the stability of the inspection images, a similar piecewise adaptive PID structure can be established in the pitch control channel, assuming the pitch control output is:

[0081] in, This represents the output of the pitch angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the desired pitch angle and the actual pitch angle. express The first derivative.

[0082] When encountering gusts of wind along the route or airflow swirling caused by towers or mountains, the pitch channel can also introduce a feedforward compensation quantity output by fuzzy scheduling. The final pitch control quantity is then:

[0083]

[0084] in, This indicates the final pitch angle control value. This is the pitch angle feedforward compensation amount. This represents the pitch angle control feedforward coefficient. This represents a feedforward compensation function based on wind and electromagnetic disturbances, used to counteract the impact of disturbances on the system. Since the disturbances in this invention can be estimated, the feedforward compensation function can be established based on known disturbances using existing technologies, which will not be elaborated here. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor.

[0085] The level flight cruise algorithm of this invention uses the roll angle, pitch angle, trajectory deviation of the UAV, and environmental disturbances estimated by the onboard sensors as the main control variables. It adopts an integrated control law structure of "error segmentation + fuzzy scheduling + thrust difference compensation + safety limit". While maintaining the speed and ease of implementation of the traditional PID controller, it introduces an adaptive gain tuning mechanism for the specific disturbance sources of the power transmission channel. Thus, without changing the overall aerodynamic shape of the UAV, it improves the heading maintenance capability and image acquisition stability in the horizontal flight state.

[0086] Preliminary inspection of captured inspection images involves real-time detection of any anomalies. This preliminary inspection aims to quickly identify suspected anomalies in real time, without needing to output the location or precise information of the anomalies; only a yes / no result is required. For example, existing lightweight models can be used for real-time airborne processing to achieve preliminary detection. If a suspected anomaly is identified, subsequent hovering operations are immediately executed for more precise detection.

[0087] To achieve seamless integration with subsequent hovering / stationary control algorithms, the controller's current segment state, disturbance intensity estimate, and thrust difference weight are incorporated. The data is also output to the mode manager so that when a target anomaly is detected and a vertical attitude needs to be entered, a more conservative transition envelope can be selected based on the current disturbance level. By segmenting the attitude error, fuzzy quantizing the disturbances specific to the power transmission corridor, jointly allocating the thrust difference of the front twin propellers, and applying a safety limit at the output, the fixed-wing inspection UAV can maintain small roll and pitch fluctuations during long-distance traverse flight, thereby ensuring that the airborne vision system can acquire clear, continuous, and low-jitter images of the power transmission line inspection.

[0088] S2. When a power line fault is initially detected, switch the drone's attitude to a hovering state with constant altitude.

[0089] This step involves the inspection drone cruising along the power transmission channel in a level flight, already in a high-efficiency fixed-wing flight state. Once the front-end identification module issues a command to "slow down and observe a specific component," the flight control system smoothly transitions the flight attitude, which is primarily driven by the lifting surface, to a "quasi-hovering attitude" primarily driven by the front propeller thrust. During this transition, the system maintains the balance of forces on the aircraft, preventing altitude loss or stall due to excessive angle of attack or roll, or deviation towards the power line due to incorrect thrust direction. A safe transition range is first defined, within which the state feedback is linearized and decoupled. Simultaneously, thrust distribution, control surface actions, and the attitude angles themselves are bound to a "cooperative constraint," ensuring that the thrust direction and gravity direction are roughly aligned.

[0090] For a fixed-wing composite configuration with twin front propellers, during the transition to hovering, the vertical component of the thrust is equal to the aircraft's weight, as shown below:

[0091] Where T represents the total thrust, m is the mass of the UAV, and g is the gravitational acceleration.

[0092] To avoid significant nonlinear coupling around this equation, it is necessary to first determine the allowable transition attitude ranges for the aircraft's pitch and roll angles:

[0093] in, These represent the maximum and minimum pitch angles, respectively. These represent the maximum and minimum roll angles, respectively.

[0094] Once the flight attitude enters this envelope, it can be considered that what needs to be controlled subsequently are "transitional states" rather than ordinary cruise states, and linearization only becomes meaningful at this point. The core objects of transitional state control are the attitude angles themselves and the three-axis angular velocities, which are arranged into a state vector x:

[0095] in, These represent the actual roll angle, pitch angle, and yaw angle, respectively. These represent the angular velocities of the three axes of the body coordinate system.

[0096] Actual organism attitude dynamics are nonlinear and can be expressed in an abstract form as follows:

[0097] in, Let x be the first derivative of the control object in the transition state. , These represent the actual roll angle, pitch angle, and yaw angle, respectively. These represent the angular velocities along the three axes of the body coordinate system. This refers to the portion of the drone's attitude that evolves naturally over time when there is no control input. A function that describes how control inputs affect the system state. This represents the input for aircraft attitude control. This indicates the deflection angle of the ailerons, which are located at the trailing edge of the left and right wings and are used to control the aircraft's rotation around the longitudinal axis of the fuselage, that is, to control the roll attitude (left and right tilt). This indicates the deflection angle of the elevator, which is located at the trailing edge of the horizontal stabilizer and is used to control the aircraft's rotation around its lateral axis, that is, to control its pitch attitude (the nose of the aircraft rising or falling). This indicates the deflection angle of the rudder, which is located at the trailing edge of the vertical tail and is used to control the aircraft's rotation around its vertical axis, that is, to control the yaw attitude (the nose yaws to the left or right). This indicates the thrust difference between the left and right propellers.

[0098] The decoupling approach is to introduce a desired output, such as the desired attitude angle tracking. Then, by selecting a suitable feedback The original nonlinear component is canceled out, making the closed loop become The rate of change of attitude angle is transmitted through a new input. Let's determine it directly.

[0099] The mapping from control input to machine torque is as follows:

[0100] in, These represent the roll, pitch, and yaw moments, respectively, and B is the moment distribution matrix. Indicates the deflection angle of the aileron. This indicates the deflection angle of the elevator. This matrix represents the rudder deflection angle, and ΔT is the thrust difference between the left and right propellers, used for rapid attitude correction. Since the front propellers are typically installed in a left-right configuration, this matrix... In reality, it's not diagonal; differential thrust can affect both roll and yaw simultaneously. Decompose B into diagonal principal components. and the coupling components that need to be compensated , means as follows:

[0101] It can control the three attitude torques independently, first press Find an ideal control variable, expressed as:

[0102] The three target torques are calculated based on the attitude error and are expressed as follows:

[0103]

[0104]

[0105] in, This represents the ideal input for body attitude control. These represent the target moments for roll, pitch, and yaw, respectively. Indicates the expected roll angle. Indicates the desired pitch angle. Indicates the desired yaw angle. , and These are the proportional gains for roll, pitch, and yaw angles, respectively. , and These represent the differential gains with respect to roll, pitch, and yaw angles, respectively.

[0106] Then use a compensation control to This part of the effect is offset, which is represented as:

[0107] In this way, the coupling of differential thrust to yaw, rudder, and roll is canceled out by feedforward, and their relationship is linearized and decoupled. Next, based on the constraints of thrust and attitude trim, the altitude is kept constant during the transition phase, resulting in the altitude balance equation:

[0108] Where T represents the total thrust, m is the mass of the UAV, and g is the acceleration due to gravity. This represents a correction amount used to offset a momentary drop or rise.

[0109] To avoid airframe vibration caused by frequent control surface movements near power lines, one control surface can be locked to a fixed small trim angle after entering the transition envelope. Fine-tuning can then be performed using only the remaining control surfaces and differential thrust. For example, the rudder can be locked, leaving only the ailerons, elevator, and differential thrust for control. The trigger condition for locking the rudder is:

[0110] in, This indicates the maximum allowable error range (tolerance band) when the pitch and roll angles approach the target attitude. It indicates the real-time deflection angle of the rudder at the current moment and the fixed deflection angle that the rudder is locked after entering a quasi-hover steady state.

[0111] The advantages of this approach are: once the drone has raised its nose to a suitable near-hovering attitude, the controller prevents the rudder from shaking back and forth, and all high-frequency attitude fine-tuning tasks rely on differential thrust, avoiding image jitter. Finally, to ensure the entire transition process is one of "slowing down first and then raising the attitude," rather than rapidly raising the nose while flying at high speed, a certain threshold can be maintained when the flight speed V is below a certain threshold. Only when this decoupling control law is entered is it permitted to proceed.

[0112] This step first defines a safe attitude range over the power transmission channel. Then, the original nonlinear attitude dynamics are linearized within this range. Next, a set of torque distribution matrices specifically designed for the propeller and control surfaces is used to eliminate the coupling between the attitude axes. Simultaneously, the total thrust and attitude angle are made to comply with a cooperative constraint that "supports the aircraft." Finally, when the speed is low enough and the attitude is close enough to the target, some control surfaces are locked, and high-frequency control is handed over to differential thrust. In this way, the fixed-wing aircraft can complete the attitude transition from level flight cruise to fixed-point observation of the power transmission components without adding an additional tilt mechanism, and the entire control law structure can still be implemented in real time on the flight control system.

[0113] S3. Extract and spatially locate local anomalies while hovering.

[0114] Preferably, this invention utilizes the stable structural feature of the conductor's geometric orientation, combined with component template matching and anomaly analysis of local brightness and texture, to achieve rapid identification and three-dimensional position calculation of damaged areas in components such as insulators, clamps, and power fittings. Its core idea is to first reconstruct a standardized principal axis reference system along the conductor's direction from the image coordinate system, and then extract local anomalies within this reference system.

[0115] First, the principal axis of the power transmission line is extracted from the image. Specifically, edge detection is performed on the original grayscale image to obtain a set of edge points. ( , Then, the principal direction is fitted using the least squares method to form a linear expression:

[0116] Where a and b are fitting coefficients.

[0117] This straight line is the geometric principal axis of the conductor. When multiple conductors run parallel, the one with the smallest angle to the horizon is chosen as the principal axis. Next, the entire image coordinate system is projected and flattened along this principal axis, making the conductor approximately horizontal in the new reference frame. This transformation can be expressed as:

[0118] in,( ', ') is the result of the flattening transformation. , ), R( () represents the two-dimensional transformation matrix based on the tilt angle of the conductor. After performing a projection and flattening transformation on the image, subsequent matching and anomaly extraction only slide along the horizontal direction, improving computational efficiency and robustness.

[0119] On the flattened strip image, a pre-stored standard component image template is invoked, and the matching degree between the image and the standard component image template is calculated using normalized cross-correlation:

[0120] in, This represents the matching degree between the image and the i-th standard component image template, where u represents the translation of the template in the horizontal direction (x-direction of the reference frame), and v represents the translation of the template in the vertical direction (y-direction of the reference frame). This represents the image after the flattening transformation. This represents the pixels of the flattened strip image. This represents the average grayscale value of the local window of the image currently participating in the matching. Indicates the first A standard component template in grayscale value at that location Indicates the first The average grayscale value of a standard component template.

[0121] When the matching degree exceeds a threshold, the region is considered to contain the corresponding standard component. Subsequently, the local brightness gradient and structural symmetry index are calculated for this region to determine the presence of anomalies. For each matching region, the local brightness variance and edge direction divergence are calculated to define an anomaly score, expressed as:

[0122] Where Q represents the anomaly score, Represents the weight parameters. Indicates the local brightness variance. This represents the local brightness reference variance. Indicates the divergence along the edge direction. This indicates the reference divergence in the edge direction.

[0123] when Points exceeding a threshold are marked as suspected damage points. Finally, based on the UAV attitude matrix and camera intrinsic and extrinsic parameters, outliers in the image coordinate system are identified. ′, Mapping this to the body coordinate system enables spatial positioning, represented as:

[0124] in, Indicates the location of anomalies in the body coordinate system. denoted by , represents the mapping matrix from image coordinate system to coordinate system , K represents the camera intrinsic parameter matrix, (x',y') represents the coordinates of the outlier point in the image coordinate system, and d represents the estimated spatial distance along the line of sight between the UAV and the fault point of the power transmission line.

[0125] This step, through geometric principal axis flattening, template consistency matching, and structural feature deviation analysis, enables rapid detection and spatial localization of local anomalies in power transmission components without relying on deep learning.

[0126] S4. With the abnormal area spatial positioning as the target, keep the relative position between the UAV and the target and the angle of the UAV's optical axis aligned with the target unchanged, perform fixed-point detection on the target, and resume level flight cruise after the fixed-point detection is completed.

[0127] The key feature of this step is that the controlled object is no longer the "absolute attitude of the drone" itself, but rather the "relative position and relative line of sight between the drone and the target".

[0128] Preferably, this step is as follows: First, represent the position of the target point in the machine's coordinate system, denoted as:

[0129] Because power transmission lines can sway slightly due to environmental influences, drones are also subject to environmental disturbances while hovering. Therefore, it is necessary to... The next position can be predicted. The relative position and relative velocity can be combined into a single state:

[0130] The next position of the target can be described using a common linear prediction model, expressed as:

[0131] in, Let x represent the x-coordinate of the target point in the body coordinate system at time k+1, and let A represent the state transition matrix, which is used to describe the evolution of the position and velocity state of the target point of the power transmission line relative to the UAV between adjacent sampling times, so as to characterize the slow drift characteristics of the target under wind load and attitude disturbance. This represents the x-coordinate of the target point in the body coordinate system at time k. This refers to the external disturbance at time k. Image measurements can obtain the target's relative position observation at this time:

[0132] in, H represents the observation of the target's relative position at time k, and H represents the observation matrix, which describes the correspondence between the target point's true relative motion state and the image measurement results. It maps the position components in the state vector into observations that can be obtained by the airborne vision system and the ranging unit. The variable representing the adaptive change of image quality at time k can be set as an adaptive observation noise, expressed as:

[0133] in, express The observation noise covariance matrix at time 1. This represents the covariance of the basic observation noise. This represents the adaptive adjustment coefficient for observation noise, used to control the degree to which image quality indicators affect the adjustment magnitude of observation noise covariance. This indicates the jitter or sharpness metric of the current frame.

[0134] Following the standard filtering approach, we first predict and then update to obtain an estimate of the target's relative position at the next moment, expressed as:

[0135]

[0136]

[0137] in, Let A represent the estimated predicted state at the current moment, and let A represent the state transition matrix. This represents the optimal state estimate at the previous moment. This represents the Kalman gain at time k. H represents the prediction error covariance matrix, and H represents the observation matrix. Let the observed noise covariance matrix at time k be denoted as . This represents the optimal fusion state estimate at the current moment. Let k represent the observation at time k.

[0138] After obtaining the predicted relative position, let the current optical axis direction of the camera be... Target direction Represented as:

[0139] Require and To minimize the square of the difference between the two, the optical axis angle constraint is:

[0140] Simultaneously, the distance between the drone and the target must be controlled, with the desired safe distance set as follows: The cost of position constraints can be written as:

[0141]

[0142] Where d represents the distance between the drone and the target, This indicates the expected safe distance between the drone and the target.

[0143] These two costs combined constitute the objective function to be optimized:

[0144] in, They represent and The weight.

[0145] The controller adjusts its settings according to the latest data each cycle. ^ | Calculate this cost and determine the required attitude and thrust corrections. The attitude corrections can be written as:

[0146] in, This indicates the attitude correction amount for the drone. Let J represent the attitude correction coefficient of the UAV, and let J represent the objective function for attitude correction of the UAV. This indicates the drone's attitude.

[0147] The thrust correction is used to change the distance to the target, and the thrust correction is expressed as:

[0148] in, This indicates the thrust correction amount for the drone. This represents the thrust correction factor for the drone.

[0149] The safe displacement command can be written as:

[0150] in, Indicates the safe displacement command for the drone. A unit vector representing the direction of a power transmission line. This represents the distance error adjustment gain coefficient along the conductor direction (tangential direction).

[0151] Finally, the attitude correction, thrust correction, and safety displacement commands are sent to the attitude-thrust co-controller in step 2 above, and then the differential thrust is used to complete the high-frequency small-amplitude attitude adjustment. In this way, even when wind disturbance, wire sway, and camera micro-shake are present, it can still maintain the optical axis of the fault point and maintain a safe distance from the wire.

[0152] This invention also proposes an unmanned aerial vehicle (UAV) for implementing the aforementioned adaptive inspection method for power transmission lines, comprising: Airframe structure, twin front propeller drive system, fixed-wing lift system, flight control system, image acquisition module, and adaptive inspection and processing module; The flight control system is configured to perform two operating modes: horizontal flight and hovering. In horizontal flight mode, the fixed-wing lift system provides lift, while the twin-head propeller drive system provides forward thrust. In hovering mode, the twin-head propeller drive system generates lift. The adaptive inspection processing module is used to generate control signals for the flight mode based on the inspection images and environmental status signals, and the control signals control the flight control system.

[0153] In one exemplary embodiment, a computer-readable storage medium is included, which stores a computer program that, when executed by a processor, implements the above-described adaptive inspection method for power transmission lines.

[0154] Please see Figure 2 In one exemplary embodiment, the device further includes an electronic device including at least one processor, at least one memory, and at least one communication bus.

[0155] The memory stores a computer program, which includes computer-readable instructions. The processor calls the computer-readable instructions stored in the memory through the communication bus to execute the aforementioned adaptive inspection method for power transmission lines.

[0156] In one exemplary embodiment, a computer program product is proposed, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described adaptive inspection method for power transmission lines.

[0157] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An adaptive inspection method for transmission lines, characterized in that, Includes the following steps: S1. Control the drone to fly level along the power transmission line and take inspection images for preliminary testing; S2. When a power line fault is initially detected, switch the drone's attitude to a hovering state with constant altitude. S3. Extract and spatially locate local anomalies while hovering. S4. With the abnormal area spatial positioning as the target, keep the relative position between the UAV and the target and the angle of the UAV's optical axis aligned with the target unchanged, perform fixed-point detection on the target, and resume level flight cruise after the fixed-point detection is completed.

2. The adaptive inspection method for transmission lines according to claim 1, characterized in that, During level flight cruise, roll and pitch angles are controlled as follows: The PID control equation for roll angle is: in, This indicates the output of the roll angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the expected roll angle and the actual roll angle. express The first derivative; The roll angle limiting function is: in, Limiting the roll angle, To control the torque, For rolling inertia, This represents the maximum value of the second derivative of the actual roll angle. Indicates the control moment of the control surface. Indicates the equivalent rolling torque. The rolling moment coefficient represents the variation of the aileron deflection angle. q represents the aileron deflection angle, q represents the dynamic pressure, S represents the wing area, and b represents the wingspan. This indicates the distance from the propeller to the fuselage's center of mass. The lever arm of direction, The direction is perpendicular to the plane of symmetry of the fuselage and points towards the right wing. This indicates the thrust difference between the left and right propellers. This represents a constant obtained from calibration based on the wing mounting position and the motor thrust curve. The dynamic weights obtained from fuzzy scheduling This represents an input consisting of three types of disturbances. Jointly determine the weight The nonlinear mapping relationship This represents three types of disturbances. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor. The first derivative represents the lateral deviation of the UAV from its flight path; The PID control equation for pitch angle is: in, This represents the output of the pitch angle PID control. They represent respectively by The proportional, integral, and differential gains are determined. This represents the difference between the desired pitch angle and the actual pitch angle. express The first derivative; right After applying feedforward compensation, the final pitch angle control value is obtained as follows: in, This indicates the final pitch angle control value. This is the pitch angle feedforward compensation amount. This represents the pitch angle control feedforward coefficient. This represents the feedforward compensation function based on wind disturbance and electromagnetic disturbance. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor.

3. The adaptive inspection method for transmission lines according to claim 2, characterized in that, , and according to The values ​​are set in segments, represented as follows: in, Represents the nth segment , and The set value, express The threshold of the 1st, 2nd, ..., nth segment. This represents the updated segment threshold after fuzzy inference. This represents the baseline threshold for the i-th error segment. This indicates that based on three types of disturbance inputs The adjustment function for adaptively correcting the segmented control threshold. This represents the threshold correction amount for the i-th segment. Indicates the lateral wind component. This represents the estimated intensity of electromagnetic disturbance to the conductor. The first derivative represents the lateral deviation of the UAV from its flight path.

4. The adaptive inspection method for transmission lines according to claim 1, characterized in that, The transitional linear decoupling control algorithm is used to switch the drone's attitude to hovering state, as detailed below: The actual equations of motion for the organism are: in, Let x be the first derivative of the control object in the transition state. , These represent the actual roll angle, pitch angle, and yaw angle, respectively. These represent the angular velocities along the three axes of the body coordinate system. This refers to the portion of the drone's attitude that evolves naturally over time when there is no control input. A function that describes how control inputs affect the system state. This represents the input for body attitude control. Indicates the deflection angle of the aileron. This indicates the deflection angle of the elevator. Indicates the deflection angle of the rudder. This indicates the thrust difference between the left and right propellers; The mapping from u to the body torque is: M=Bu in, Indicates the torque of the machine body. These represent the roll, pitch, and yaw moments, respectively. Where B is the torque distribution matrix, Let B be the diagonal principal component. The coupling components that need to be compensated for in B; The decoupled u is represented as: in, This represents the ideal input for body attitude control. These represent the target moments for roll, pitch, and yaw, respectively. Indicates the expected roll angle. Indicates the desired pitch angle. Indicates the desired yaw angle. , and These are the proportional gains for roll, pitch, and yaw angles, respectively. , and These represent the differential gains with respect to roll, pitch, and yaw angles, respectively. The height balance equation is: Where T represents the total thrust, m is the mass of the UAV, and g is the acceleration due to gravity. This represents a correction amount.

5. The adaptive inspection method for transmission lines according to claim 1, characterized in that, S3 specifically refers to: Extract the main axis of the power transmission line from the inspection images; Perform a projection flattening transformation on the image coordinates along the principal axis; On the flattened strip image, a pre-stored standard component image template is called, and the matching degree between the strip image and the standard component image template is calculated using normalized cross-correlation. For matching regions where the matching degree exceeds a threshold, calculate the local brightness gradient and structural symmetry index to determine whether the matching region is abnormal. For each matching region, calculate the local brightness variance and edge direction divergence to define an anomaly score, expressed as: Where Q represents the anomaly score, Represents the weight parameters. Indicates the local brightness variance. This represents the local brightness reference variance. Indicates the divergence along the edge direction. Indicates the reference divergence in the edge direction; Based on the UAV attitude matrix and camera intrinsic and extrinsic parameters, outliers in the image coordinate system are mapped to the body coordinate system to achieve spatial positioning.

6. The adaptive inspection method for transmission lines according to claim 1, characterized in that, The S4 implementation method is as follows: obtain the attitude correction amount, thrust correction amount and safety displacement command of the UAV, and realize the UAV hovering control based on the attitude correction amount, thrust correction amount and safety displacement command of the UAV; The attitude correction for the UAV is expressed as: in, This indicates the attitude correction amount for the drone. Let J represent the attitude correction coefficient of the UAV, and let J represent the objective function for attitude correction of the UAV. Indicates the attitude of the drone. They represent and The weight, and Let represent the costs of the optical axis angle constraint and the position constraint, respectively, and d represent the distance between the UAV and the target. This indicates the expected safe distance between the drone and the target. Indicates the current optical axis direction of the camera. Indicates the direction of the target. This indicates the position of the target point in the machine's coordinate system. The thrust correction for the UAV is: in, This indicates the thrust correction amount for the drone. This indicates the thrust correction factor for the drone; The safe displacement command for the drone is: in, Indicates the safe displacement command for the drone. A unit vector representing the direction of a power transmission line. This represents the distance error adjustment gain coefficient along the direction of the conductor.

7. A drone, characterized in that, An adaptive inspection method for transmission lines according to any one of claims 1-6 includes: Airframe structure, twin front propeller drive system, fixed-wing lift system, flight control system, image acquisition module, and adaptive inspection and processing module; The flight control system is configured to perform two operating modes: horizontal flight and hovering. In horizontal flight mode, the fixed-wing lift system provides lift, while the twin-head propeller drive system provides forward thrust. In hovering mode, the twin-head propeller drive system generates lift. The adaptive inspection processing module is used to generate control signals for the flight mode based on the inspection images and environmental status signals, and the control signals control the flight control system.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.

9. An electronic device, characterized in that, The device includes a processor and a memory, the processor being interconnected with the memory, wherein the memory is used to store a computer program, the computer program including computer-readable instructions, and the processor is configured to invoke the computer-readable instructions to perform the method as described in any one of claims 1-6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-6.

Images