A method for unmanned aerial vehicle to plan a directional flight path for inspecting a tower pole of a power transmission line in a mountainous area

Abstract

The application provides a method for unmanned aerial vehicle (UAV) inspection of mountainous power transmission line tower and pole directional path planning, which solves the problem that the path planning formed by initial inspection coordinate points cannot accurately give the accurate position of the power transmission line tower and pole, and makes the error between the tower and pole coordinate points given by the initial inspection coordinate points and the actual position coordinate points of the tower and pole larger and the efficiency lower when the UAV inspects. The application collects the latitude and longitude position of the target tower and pole for coordinate system transformation; obtains n coordinate points of the target tower path planning by frequency calculation and processing according to the extended Kalman filtering algorithm, acquires real-time position data of the UAV by the extended Kalman filtering, and transmits the real-time position data of the UAV and the n coordinate points of the target tower path planning to the target UAV position control loop for comparison; the workload of the ground staff can be greatly reduced, the required inspection tower and pole are guaranteed to be covered, the effectiveness of the inspection data is guaranteed, and the effective execution of the system is guaranteed. The application is used in the field of power transmission line inspection.

Description

Technical Field

[0001] This invention belongs to the field of power transmission line inspection, and specifically relates to a method for planning the directional flight path of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas. Background Technology

[0002] With the rapid development of drone technology and its application across various fields, it is gradually demonstrating its unique advantages in all industries. As the number of users on China's power grid increases year by year, more and more high-voltage lines are put into operation, and power companies spend a significant amount of manpower and resources annually on the operation, maintenance, and repair of transmission towers. Because my country's transmission lines are widely distributed throughout the country, their geographical locations and surrounding environments are often very complex and harsh. To ensure the safe operation of the national power system, regular inspections of power facilities are necessary. Compared to transmission lines in plains areas, mountainous transmission lines are more susceptible to damage from natural factors, thus requiring even more frequent inspections by relevant personnel. However, this presents a considerable challenge for inspection personnel.

[0003] Traditional inspection methods involve personnel driving to designated routes for scheduled inspections. This method not only consumes significant manpower and resources but also has low efficiency and safety. Furthermore, it is time-consuming, subject to terrain limitations, and the accuracy of data is difficult to guarantee. More importantly, the complex operations in mountainous areas or the climbing process in harsh weather conditions pose safety hazards to the workers.

[0004] With the continuous development of drone technology, drone inspection technology has been widely applied. Drones, equipped with onboard high-definition cameras, can conduct comprehensive inspections of power lines, ensuring the validity of inspection data. Their high efficiency, low cost, and high safety advantages have better replaced traditional manual inspection methods. For drone inspections in complex environments, the only considerations are the drone's own control system, navigation, and trajectory planning. However, during drone flight, environmental factors can affect the accuracy of power transmission lines. In mountainous areas, the trajectory planning based on initial inspection coordinates may not accurately pinpoint the location of the transmission towers. This results in a significant discrepancy between the tower coordinates provided by the drone and the actual tower location, impacting inspection efficiency, reducing the accuracy of the inspected area, and decreasing the validity of the data. Summary of the Invention

[0005] This invention addresses the problem that the trajectory planning based on initial inspection coordinate points cannot accurately determine the precise location of power transmission line towers, resulting in a large error between the tower coordinates given by the initial inspection coordinate points and the actual tower location coordinates during UAV inspections. This affects the efficiency of UAV inspections, reduces the accuracy of the inspected location area, and degrades the validity of the data. The invention provides a method for directional trajectory planning of UAVs for inspecting power transmission line towers in mountainous areas. This method can accurately and stably complete the autonomous inspection of power towers. It requires obtaining high-precision tower inspection point coordinates and using them to plan the autonomous inspection trajectory.

[0006] This invention provides a method for planning the directional flight path of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas. The method includes the following steps:

[0007] Step 1: The data acquisition and processing module converts the pre-collected latitude and longitude coordinates of the target tower into the position coordinates of the target tower in the geodetic rectangular coordinate system, obtaining K position coordinates of the target tower, which are in the same coordinate system as the initial inspection coordinates.

[0008] Step 2: The trajectory planning module calculates the target tower function of the K-1 segment of the route formed by connecting the position coordinates of K target towers in sequence. The coordinates of n target towers are obtained by frequency calculation and processing using the extended Kalman filter algorithm. The coordinates of the n target towers are the n coordinates of the target tower trajectory planning.

[0009] Step 3: The tower orientation processing module obtains the real-time position data of the target UAV by processing the position data of the UAV during its flight process through an algorithm. The real-time position data of the target UAV and the target tower trajectory planning of n coordinate points are then transmitted to the target UAV flight control position control loop.

[0010] Preferably, in step 1, the data acquisition and processing module converts the pre-acquired latitude and longitude coordinates of the target tower into the position coordinates of the target tower in a geodetic rectangular coordinate system. The specific steps are as follows:

[0011] Step 201: Transform the latitude and longitude coordinates P(L,B,H) of the target tower in the geodetic coordinate system to the geodetic space rectangular coordinate system to obtain the coordinates P(X,Y,Z) of the target tower; obtain the coordinates (x,y) of the latitude B value of point P in the coordinate system O-xy of the meridian two-dimensional plane:

[0012] x = |PK|cos B = (N+H)cos B

[0013] y = |PQ|sin B = [N(1-e 2 )+H]sin B

[0014] Wherein, PK and PQ represent the line segment lengths from point P to point K and from point P to point Q, respectively;

[0015] Step 202 involves obtaining the two-dimensional plane coordinates (x, y) of the meridian in the geodetic rectangular coordinate system (X, Y, Z):

[0016] X=xcosβ, Y=xsinβ, Z=y;

[0017] Where β is the angle between the x-axis of the meridian two-dimensional plane and the x-axis of the geodetic rectangular coordinate system; x, y are the coordinates of the latitude and longitude of the desired point P in the meridian two-dimensional plane;

[0018] Step 203 yields the coordinates (X, Y, Z) of the geodetic coordinate system P(L, B, H) in the geodetic rectangular coordinate system.

[0019]

[0020] Where N represents the radius of curvature of the ellipse: e 2 Represents the eccentricity of the ellipsoid: a is the major semi-axis, and b is the minor semi-axis.

[0021] Preferably, the coordinates of the n target towers in step 2 are specifically defined as the tower points traversed between the previous inspection point P1(X1,Y1) and the next inspection point P2(X2,Y2) as Pi(Xi,Yi) and Pj(Xj,Yj); and the projection points on the inspection line of the tower coordinates as (Xi',Yi') and (Xj',Yj').

[0022] When the target drone travels to the section from (X1,Y1) to (X2,Y2), the original inspection route is replaced, changing (X1,Y1) to (X2,Y2) to (X1,Y1) to (Xi,Yi), (Xi,Yi) to (Xj,Yj), and (Xj,Yj) to (X2,Y2).

[0023] The formula for solving the equation of the straight line of the track from (X1,Y1) to (Xi,Yi) is:

[0024] (X-X1) / (Xi-X1)=(Y-Y1) / (Yi-Y1)

[0025] The formula for solving the equation of the straight line of the track from (Xi,Yi) to (Xj,Yj) is:

[0026] (X-Xi) / (Xj-Xi)=(Y-Yi) / (Yj-Yi)

[0027] The formula for solving the equation of the straight line of the trajectory from (Xj,Yj) to (X2,Y2) is:

[0028] (X-Xj) / (X2-Xj)=(Y-Yj) / (Y2-Yj)

[0029] Where i and j are both natural numbers, i < n, j < n, n > 1;

[0030] After obtaining the radius of curvature N of the ellipse, the (X,Y,Z) coordinates of all target towers on the required inspection route are obtained.

[0031] Preferably, the inspection line is calculated according to the formula for solving the straight line equation of the track, and so on. According to the frequency of the extended Kalman filter algorithm, each inspection line is set as the nth coordinate point of the target inspection tower track at a preset distance of d meters. Starting from the initial value of the x value of the nth target coordinate point, the y value of the nth target coordinate point is calculated every time the preset value d meters is added to x. The nth (x, y) value is then saved as the nth coordinate point of the target inspection tower track planning.

[0032] Connecting the n target coordinate points forms the target inspection tower trajectory plan, where r is: r = {{x1,y1}, {x2,y2}, ..., {x...} n ,y n}}.

[0033] Preferably, in step 3, the tower orientation processing module obtains the real-time position data of the target UAV by processing the position data of the UAV during its flight process using an algorithm. The real-time position data of the target UAV and the n coordinate points of the target tower trajectory planning are then transmitted to the position control loop of the target UAV flight control system. Specifically, the n coordinate points of the target inspection tower trajectory planning and the real-time position data of the target UAV obtained by the tower orientation processing module through the extended Kalman filter algorithm based on the position data of the UAV during its flight process are transmitted to the position control loop in the target UAV flight control system, providing relevant control input parameters for the UAV to achieve flight status control.

[0034] Preferably, the real-time position data of the target UAV is obtained by filtering and fusing IMU data and GPS data acquired by the target UAV during flight using an extended Kalman filter algorithm to obtain the actual position coordinates of the target aircraft.

[0035] The Kalman filter algorithm filters and fuses IMU data with GPS data. Position data collected during the target aircraft's flight is processed by the extended Kalman filter algorithm to continuously update the UAV's position and attitude data. The extended Kalman filter algorithm equations include time prediction updates and measurement correction updates. The steps of the extended Kalman filter algorithm are as follows:

[0036] Step 301: Estimate the current state based on the state transition equation and the posterior estimate from the previous time step, and obtain the predicted prior state estimate.

[0037]

[0038] Where k is the loop number. Let u be the posterior estimate of the (k-1)th iteration. k-1 This is the control matrix for the k-th cycle. ω is a nonlinear function estimated a priori; k-1 It is an ι×1 dimensional system noise vector, where ι is the dimension of the state parameters;

[0039] T denotes the transpose of a matrix;

[0040] Step 302: Predict the covariance of the current estimation error based on the covariance of the estimation error at the previous time step.

[0041] Prediction prior error covariance P k∣k-1 :

[0042]

[0043] Among them, F k-1 Let P be the state transition matrix for the (k-1)th cycle. k-1∣k-1 G is the posterior estimate of the covariance for the (k-1)th iteration. k-1 Let Q be the control input matrix for the (k-1)th cycle. k-1 Let be the covariance matrix of the (k-1)th cycle, assuming it follows a multivariate zero-mean normal distribution. This is the transpose of the control input matrix for the (k-1)th iteration;

[0044] Step 303 calculates the Kalman gain K from the error covariance and the observation matrix. k :

[0045]

[0046] Among them, H k Let Q be the observation matrix, and let both Q and R follow a normal distribution. k :ω k ~N(0,Q) k ), R k :v k ~N(0,R k ).

[0047] Let R′ be the measurement matrix for the k-th cycle. k The observation noise is the zero-mean Gaussian covariance of the k-th cycle;

[0048] Step 304 uses the Kalman filter gain to compensate for the observations in the estimate from step 301, thus obtaining the updated state estimate.

[0049]

[0050] in, For prior state estimation, and ∈ state variables m, z′ k Let z′ be the current state measurement value in the k-th cycle. k|k-1 This is the state measurement value for the (k-1)th cycle;

[0051] state variables

[0052] q is the attitude quaternion:

[0053] V NED This represents the machine's velocity in the NED coordinate system.

[0054] P NED This indicates the machine's position in the NED coordinate system:

[0055] Δθ b This represents the incremental angular deviation (bias) in the XYZ body coordinate system:

[0056] ΔV n This represents the incremental velocity deviation in the XYZ body coordinate system:

[0057] Step 305 corrects the estimation error covariance from step 302: This yields the updated posterior estimation error covariance matrix.

[0058] P k∣k =(I n -K k H k )P k∣k-1 (5)

[0059] Among them, i n For an n-dimensional identity matrix, H k For the observation matrix, K k For Kalman gain, P k∣k-1 Let be the prior error covariance.

[0060] Preferably, step 301 estimates the current state based on the state transition equation and the estimated state at the previous moment, and obtains the predicted prior state estimate: the IMU data and the attitude quaternion q are converted. The IMU data is read by two sensors, an accelerometer and a gyroscope. The gyroscope reads the real-time three-axis angular velocity ω value of the target UAV, the accelerometer obtains the real-time three-axis acceleration value in the body coordinate system of the target UAV, and the GPS data obtains the actual measured position and velocity values ​​of the aircraft.

[0061] The gyroscope reads the three-axis angular velocity ω, and the three-axis angle change Δang_meas is obtained by integrating the three-axis angular velocity. Then, the gyroscope offset Δang_bias is subtracted to obtain the target UAV angle change Δang_truth.

[0062]

[0063] Δ ang_truth =Δ ang_meas -Δ ang_bias

[0064] The change Δ of the quaternion q of the attitude quaternion quat The change in the angle of the target UAV is then used to represent it.

[0065]

[0066] Using quaternion multiplication to represent the rotation of the state from time K to time K+1, we obtain the quaternion state transition equation, and the multiplication is represented in matrix form as follows:

[0067]

[0068] Where K≥1.

[0069] Preferably, the three-axis acceleration values ​​in the body coordinate system during the real-time flight of the target UAV are obtained from the accelerometer, and the steps are as follows:

[0070] Step 401: Read the velocity change from the accelerometer in the IMU data. The velocity change is in the body coordinate system. Then, subtract the velocity offset to obtain the velocity change Δ of the target UAV in the body coordinate system. vel_truth :

[0071]

[0072] Δ vel_truth =Δ vel_meas -Δ vel_bias

[0073] Step 402 calculates the change in velocity of the target UAV in the body coordinate system based on step 401, and obtains the aircraft velocity value in the ground coordinate system. Since the accelerometer itself removes the influence of gravitational acceleration, the influence of gravitational acceleration needs to be added when calculating the velocity at time K+1, resulting in the quaternion-based transfer matrix. The quaternion-based transfer matrix is ​​as follows:

[0074]

[0075] in, It is the attitude rotation matrix, the attitude rotation matrix VN, VE, and VD are the calculated aircraft velocity values ​​in three directions in the ground coordinate system.

[0076]

[0077] Step 403 calculates the target UAV's velocity in the ground coordinate system based on step 402, and calculates the aircraft position state variables, which will also be used as the prediction values ​​for the extended Kalman filter. The prior position estimate required for the extended Kalman filter is obtained from the displacement formula x = x0 + vt in physics. This prior position estimate is the predicted value of the target UAV's flight position state. Specifically:

[0078]

[0079] Where P represents position, NED is the abbreviation for the direction name of the three axes of the geodetic coordinate system (northeast, east, and west), and PN, PE, and PD represent the position coordinates of the three directions.

[0080] Preferably, the actual position data of the target UAV is estimated using formula (4).

[0081] The process is iterated, and in each iteration, the state quantity m of the target UAV at the current time K is calculated: The current values ​​in the table;

[0082] The error covariance matrix P is obtained through formula (5). k∣k :P k∣k =(I n -K k H k )P k∣k-1 The error covariance matrix is ​​updated in each iteration by calculating the value of the state variable m at the next moment. The real-time position data of the target UAV is obtained by iterating through each iteration of formulas (4) and (5).

[0083] Preferably, based on the n coordinate points (x, y, z) in the inspection tower trajectory planning... n ,y n The data is transmitted along with the real-time location data of the target UAV to the target UAV's position control loop, providing input parameters for the target UAV's position control loop to perform position comparison.

[0084] This invention provides a method for directional flight path planning for UAV inspection of power transmission line towers in mountainous areas. It solves the problem that the flight path planning formed by the initial inspection coordinate points cannot accurately give the precise location of the power transmission line towers, resulting in a large error between the tower coordinates given by the UAV based on the initial inspection coordinates and the actual position coordinates of the towers. This affects the efficiency of UAV inspection, reduces the accuracy of the inspected location area, and deteriorates the validity of the data. This invention acquires the latitude and longitude positions of the target tower and performs coordinate system transformation; it places the coordinates of the target tower and the initial inspection coordinates in the same coordinate system, and calculates and processes them using an extended Kalman filter algorithm to obtain n coordinate points for the target tower's trajectory planning. The invention then uses the extended Kalman filter to obtain the real-time position data of the UAV, and transmits the UAV's real-time position data and the n coordinate points of the target tower's trajectory planning to the target UAV's position control loop for comparison and input parameters. This significantly reduces the workload of ground personnel, ensures that all required inspection towers are covered and that the inspection data is valid, enables wireless monitoring of the operational status of mountain circuits within the inspection range, helps inspection personnel accurately and efficiently record the inspection results of mountain lines without climbing, provides effective and accurate target trajectory coordinates, and ensures the effective execution of the system. Attached Figure Description

[0085] Figure 1 A schematic diagram of a method for planning the directional flight path of a drone for inspecting power transmission line towers in mountainous areas;

[0086] Figure 2 A schematic diagram of a module processing method for directional flight path planning of unmanned aerial vehicle (UAV) inspection of power transmission line towers in mountainous areas;

[0087] Figure 3 A schematic diagram of an arbitrary tower point P in the geodetic coordinate system, illustrating a method for directional flight path planning of UAVs for inspecting power transmission line towers in mountainous areas.

[0088] Figure 4 A schematic diagram of a method for planning the directional flight path of an unmanned aerial vehicle (UAV) for inspecting power transmission line towers in mountainous areas, showing a point P on any tower in its two-dimensional meridian coordinate system.

[0089] Figure 5 This is a schematic diagram of an embodiment of a method for planning the directional flight path of a drone for inspecting power transmission line towers in mountainous areas. Detailed Implementation

[0090] This invention provides a method for directional flight path planning for unmanned aerial vehicle (UAV) inspection of power transmission line towers in mountainous areas, as shown in the embodiments of the present invention. Figure 1 As shown, the method includes the following steps:

[0091] Step 1: The data acquisition and processing module converts the pre-collected latitude and longitude coordinates of the target tower into the position coordinates of the target tower in the geodetic rectangular coordinate system, obtaining K position coordinates of the target tower, which are in the same coordinate system as the initial inspection coordinates.

[0092] Step 2: The trajectory planning module calculates the target tower function of the K-1 segment of the route formed by connecting the position coordinates of K target towers in sequence. The coordinates of n target towers are obtained by frequency calculation and processing using the extended Kalman filter algorithm. The coordinates of the n target towers are the n coordinates of the target tower trajectory planning.

[0093] Step 3: The tower orientation processing module obtains the real-time position data of the target UAV by processing the position data of the UAV during its flight process through an algorithm. The real-time position data of the target UAV and the target tower trajectory planning of n coordinate points are then transmitted to the target UAV flight control position control loop.

[0094] In one embodiment, the data acquisition and processing module in step 1 converts the pre-acquired latitude and longitude coordinates of the target tower into the position coordinates of the target tower in a geodetic rectangular coordinate system. The specific steps are as follows:

[0095] Step 201: Transform the latitude and longitude coordinates P(L, B, H) of the target tower in the geodetic coordinate system to the geodetic space rectangular coordinate system to obtain the coordinates P(X, Y, Z) of the target tower; obtain the coordinates (x, y) of the meridian two-dimensional plane O-xy from the latitude B value of point P:

[0096] x = |PK|cos B = (N+H)cos B

[0097] y = |PQ|sin B = [N(1-e 2 )+H]sin B

[0098] Wherein, PK and PQ represent the line segment lengths from point P to point K and from point P to point Q, respectively;

[0099] Step 202 involves obtaining the two-dimensional plane coordinates (x, y) of the meridian in the geodetic rectangular coordinate system (X, Y, Z):

[0100] X=xcosβ, Y=xsinβ, Z=y;

[0101] Where β is the angle between the x-axis of the meridian two-dimensional plane and the x-axis of the geodetic rectangular coordinate system; x, y are the coordinates of the latitude and longitude of the desired point P in the meridian two-dimensional plane;

[0102] Step 203 obtains the coordinates (X, Y, Z) of the geodetic coordinate system P(L, B, H) in the geodetic space rectangular coordinate system.

[0103]

[0104] Where N represents the radius of curvature of the ellipse: e 2 Represents the eccentricity of the ellipsoid: a is the major semi-axis, and b is the minor semi-axis.

[0105] In one embodiment, the coordinate points of the n target towers mentioned in step 2 are specifically the tower points that Pi(Xi,Yi) and Pj(Xj,Yj) that are passed along the route between the previous inspection point P1(X1,Y1) and the next inspection point P2(X2,Y2); and the projection points on the inspection route of the tower coordinate points are (Xi',Yi') and (Xj',Yj').

[0106] When the target drone travels to the section from (X1,Y1) to (X2,Y2), the original inspection route is replaced, changing (X1,Y1) to (X2,Y2) to (X1,Y1) to (Xi,Yi), (Xi,Yi) to (Xj,Yj), and (Xj,Yj) to (X2,Y2).

[0107] The formula for solving the equation of the straight line of the track from (X1,Y1) to (Xi,Yi) is:

[0108] (X-X1) / (Xi-X1)=(Y-Y1) / (Yi-Y1)

[0109] The formula for solving the equation of the straight line of the track from (Xi,Yi) to (Xj,Yj) is:

[0110] (X-Xi) / (Xj-Xi)=(Y-Yi) / (Yj-Yi)

[0111] The formula for solving the equation of the straight line of the trajectory from (Xj,Yj) to (X2,Y2) is:

[0112] (X-Xj) / (X2-Xj)=(Y-Yj) / (Y2-Yj)

[0113] Where i and j are both natural numbers, i < n, j < n, n > 1;

[0114] After obtaining the radius of curvature N of the ellipse, the (X,Y,Z) coordinates of all target towers on the required inspection route are obtained.

[0115] In one embodiment, the formula for solving the straight line equation of the track is used to calculate each inspection line segment. According to the frequency of the extended Kalman filter algorithm, each inspection line segment is set as the nth coordinate point of the target inspection tower track at a preset distance of d meters. Starting from the initial value of the x value of the nth target coordinate point, the preset value d meters is added to x each time, and the y value of the nth target coordinate point is calculated. The nth (x, y) value is then saved as the nth coordinate point of the target inspection tower track planning.

[0116] Connecting the n target coordinate points forms the target inspection tower trajectory plan, where r is: r = {{x1,y1}, {x2,y2}, ..., {x...} n ,y n}}.

[0117] In one embodiment, step 3, the tower orientation processing module obtains the real-time position data of the target UAV based on the position data of the UAV during flight through algorithm processing. The real-time position data of the target UAV and the target tower trajectory planning n coordinate points are transmitted to the target UAV flight control position control loop. Specifically, the target inspection tower trajectory planning n coordinate points and the real-time position data of the target UAV obtained by the tower orientation processing module based on the position data of the UAV during flight through extended Kalman filtering algorithm are transmitted to the position control loop in the target UAV flight control, providing relevant control input parameters for the UAV to achieve flight state control.

[0118] In one embodiment, the real-time position data of the target UAV is obtained by filtering and fusing IMU data and GPS data acquired by the target UAV during flight using an extended Kalman filter algorithm to obtain the actual position coordinates of the target aircraft; the IMU mentioned here is the Inertial Measurement Unit.

[0119] The Kalman filter algorithm filters and fuses IMU data with GPS data. Position data collected during the target aircraft's flight is processed using the Extended Kalman Filter (EKF) algorithm to continuously update the UAV's position and attitude data. The EKF algorithm equations include time prediction updates and measurement correction updates. The steps of the EKF algorithm are as follows:

[0120] Step 301: Estimate the current state based on the state transition equation and the posterior estimate from the previous time step, and obtain the predicted prior state estimate.

[0121]

[0122] Where k is the loop number. Let u be the posterior estimate of the (k-1)th iteration. k-1 This is the control matrix for the k-th cycle. For the prior estimate of the nonlinear function; the nonlinear function f:R n ×R m ×R n →R n and h:R n ×R m →R m ;, m and n are parameters of the nonlinear function, referring to the dimension of its internal terms; ω k-1 It is an ι×1 dimensional system noise vector, where ι is the dimension of the state parameters;

[0123] T denotes the transpose of a matrix;

[0124] Step 302: Predict the covariance of the current estimation error based on the covariance of the estimation error at the previous time step.

[0125] Prediction prior error covariance P k∣k-1 :

[0126]

[0127] Among them, F k-1 Let P be the state transition matrix for the (k-1)th cycle. k-1∣k-1 G is the posterior estimate of the covariance for the (k-1)th iteration. k-1 Let Q be the control input matrix for the (k-1)th cycle. k-1 Let be the covariance matrix of the (k-1)th cycle, assuming it follows a multivariate zero-mean normal distribution. This is the transpose of the control input matrix for the (k-1)th iteration;

[0128] Step 303 calculates the Kalman gain K from the error covariance and the observation matrix. k :

[0129]

[0130] Among them, H k Let Q be the observation matrix, and let both Q and R follow a normal distribution. k :ω k ~N(0,Q) k ), R k :v k ~N(0,R k ).

[0131] Let R′ be the measurement matrix for the k-th cycle. kThe observation noise is the zero-mean Gaussian covariance of the k-th cycle;

[0132] Step 304 uses the Kalman filter gain to compensate for the observations in the estimate from step 301, thus obtaining the updated state estimate.

[0133]

[0134] in, For prior state estimation, and ∈ state variables m, z′ k Let z′ be the current state measurement value in the k-th cycle. k|k-1 This is the state measurement value for the (k-1)th cycle;

[0135] state variables

[0136] q is the attitude quaternion:

[0137] V NED This represents the machine's velocity in the NED coordinate system.

[0138] P NED This indicates the machine's position in the NED coordinate system:

[0139] Δθ b This represents the incremental angular deviation (bias) in the XYZ body coordinate system:

[0140] ΔV b This represents the incremental velocity deviation in the XYZ body coordinate system:

[0141] Step 305 corrects the estimation error covariance from step 302: This yields the updated posterior estimation error covariance matrix.

[0142] P k∣k =(I n -K k H k )P k∣k-1 (5)

[0143] Among them, I n For an n-dimensional identity matrix, H k For the observation matrix, K k For Kalman gain, P k∣k-1 Let be the prior error covariance.

[0144] In one embodiment, step 301 estimates the current state based on the state transition equation and the estimated state of the previous time step, and obtains the predicted prior state estimate: the IMU data and the attitude quaternion q are converted. The IMU data is read by two sensors, an accelerometer and a gyroscope. The gyroscope reads the real-time three-axis angular velocity ω value of the target UAV during flight, the accelerometer obtains the real-time three-axis acceleration value of the target UAV in the body coordinate system during flight, and the GPS data obtains the actual measured position and velocity values ​​of the aircraft.

[0145] The gyroscope reads the three-axis angular velocity ω, and the three-axis angle change Δang_meas is obtained by integrating the three-axis angular velocity. Then, the gyroscope offset Δang_bias is subtracted to obtain the target UAV angle change Δang_truth.

[0146]

[0147] Δ ang_truth =Δ ang_meas -Δ ang_bias

[0148] The change Δ of the quaternion q of the attitude quaternion quat The change in the angle of the target UAV is then used to represent it.

[0149]

[0150] Using quaternion multiplication to represent the rotation of the state from time K to time K+1, we obtain the quaternion state transition equation, and the multiplication is represented in matrix form as follows:

[0151]

[0152] Where K≥1.

[0153] In one embodiment, the three-axis acceleration values ​​in the body coordinate system during the real-time flight of the target UAV are obtained from the accelerometer, and the steps are as follows:

[0154] Step 401: Read the velocity change from the accelerometer in the IMU data. The velocity change is in the body coordinate system. Then, subtract the velocity offset to obtain the velocity change Δ of the target UAV in the body coordinate system. vel_truth :

[0155]

[0156] Δ vel_truth =Δ vel_meas -Δ vel_bias

[0157] Step 402 calculates the change in velocity of the target UAV in the body coordinate system based on step 401, and obtains the aircraft velocity value in the ground coordinate system. Since the accelerometer itself removes the influence of gravitational acceleration, the influence of gravitational acceleration needs to be added when calculating the velocity at time K+1, resulting in the quaternion-based transfer matrix. The quaternion-based transfer matrix is ​​as follows:

[0158]

[0159] in, It is the attitude rotation matrix, the attitude rotation matrix VN, VE, and VD are the calculated aircraft velocity values ​​in three directions in the ground coordinate system.

[0160]

[0161] Step 403 calculates the target UAV's velocity in the ground coordinate system based on step 402, and calculates the aircraft position state variables, which will also be used as the prediction values ​​for the extended Kalman filter. The prior position estimate required for the extended Kalman filter is obtained from the displacement formula x = x0 + vt in physics. This prior position estimate is the predicted value of the target UAV's flight position state. Specifically:

[0162]

[0163] Where P represents position, NED is the abbreviation for the direction name of the three axes of the geodetic coordinate system (northeast, east, and west), and PN, PE, and PD represent the position coordinates of the three directions.

[0164] Preferably, the actual position data of the target UAV is estimated using formula (4).

[0165] The process is iterated, and in each iteration, the state quantity m of the target UAV at the current time K is calculated: The current values ​​in the table;

[0166] The error covariance matrix P is obtained through formula (5). k∣k :P k∣k =(I n -K k H k )P k∣k-1 The error covariance matrix is ​​updated in each iteration by calculating the value of the state variable m at the next moment. The real-time position data of the target UAV is obtained by iterating through each iteration of formulas (4) and (5).

[0167] In one embodiment, based on n coordinate points (x, y, z) in the inspection tower trajectory planning... n ,y n The data is transmitted along with the real-time location data of the target UAV to the target UAV's position control loop, providing input parameters for the target UAV's position control loop to perform position comparison.

[0168] This invention provides a method for directional flight path planning for UAV inspection of power transmission line towers in mountainous areas. It solves the problem that the flight path planning formed by the initial inspection coordinate points cannot accurately give the precise location of the power transmission line towers, resulting in a large error between the tower coordinates given by the UAV based on the initial inspection coordinates and the actual position coordinates of the towers. This affects the efficiency of UAV inspection, reduces the accuracy of the inspected location area, and deteriorates the validity of the data. This invention acquires the latitude and longitude positions of the target tower and performs coordinate system transformation; it places the coordinates of the target tower and the initial inspection coordinates in the same coordinate system, and calculates and processes them using an extended Kalman filter algorithm to obtain n coordinate points for the target tower's trajectory planning. The invention then uses the extended Kalman filter to obtain the real-time position data of the UAV, and transmits the UAV's real-time position data and the n coordinate points of the target tower's trajectory planning to the target UAV's position control loop for comparison and input parameters. This significantly reduces the workload of ground personnel, ensures that all required inspection towers are covered and that the inspection data is valid, enables wireless monitoring of the operational status of mountain circuits within the inspection range, helps inspection personnel accurately and efficiently record the inspection results of mountain lines without climbing, provides effective and accurate target trajectory coordinates, and ensures the effective execution of the system.

[0169] Example 1

[0170] This invention provides a method for planning the directional flight path of a drone for inspecting power transmission line towers in mountainous areas, such as... Figure 2 As shown, this method includes a data acquisition and processing module, a target tower trajectory planning module, and a target tower orientation processing module;

[0171] The data acquisition and processing module, in the initial data acquisition and processing stage, needs to control the drone to collect images of several inspection points on the tower, as well as the pose of the drone and camera gimbal during shooting. Then, the acquired images are preprocessed to obtain data related to the inspection points, and the obtained coordinates are transformed into a coordinate system. Since the initial inspection coordinates for drone takeoff require manual setting via a ground station, the tower coordinates, after coordinate system transformation, are now unified with the initial inspection coordinates in the same coordinate system. This is for use in the target tower trajectory planning module's algorithm calculations.

[0172] The target tower trajectory planning module uses the initial inspection coordinates that were manually set before the inspection point calculation and the target tower coordinate data collected and processed by the module. Through the target tower trajectory planning module, it calculates the target tower function for the coordinates of several target tower inspection points. In other words, it connects the coordinates of K target towers in sequence with k-1 segments of the route, obtains the target function through calculation, and obtains the coordinates of n target towers by frequency calculation using the extended Kalman filter algorithm. These n coordinates of the target towers are the n coordinates of the target tower trajectory planning.

[0173] In one embodiment, it is necessary to solve the coordinate information of the three waypoints corresponding to each inspection point in the system coordinate system by their correspondence, including the three-dimensional coordinates of the waypoints and the attitude angle of the gimbal when shooting at that point;

[0174] All waypoints are used to generate inspection tracks for power poles according to the inspection operation requirements and the structural planning path of the poles. The coordinate information of each waypoint is then uploaded to the UAV flight control position control loop in the order of the inspection tracks.

[0175] The tower orientation processing module is used to process the coordinates of the target tower's planned trajectory when the UAV receives the coordinates of the target tower's actual position. The UAV's actual position coordinates need to be compared with the coordinates of the target tower's planned trajectory. The tower orientation processing module obtains the target UAV's real-time position data by processing the UAV's position data during flight through algorithms. The real-time position data of the target UAV and the n coordinates of the target tower's planned trajectory are then transmitted to the target UAV's flight control position control loop.

[0176] It should be noted that the target tower trajectory planning module and the tower orientation processing module are not sequential. For ease of explanation, the steps are numbered. The output results of both the target tower trajectory planning module and the tower orientation processing module are transmitted to the input end of the target UAV flight control position control loop as the input of the comparison parameters of the target UAV flight control position control loop.

[0177] This is the overall process of the inspection trajectory design framework. The method and process are explained as follows: Figure 2 As shown:

[0178] Step 1 Data Acquisition and Processing Module

[0179] 1. Data collection: Through preliminary research or satellite image data, the specific latitude and longitude coordinates of all towers on the required inspection route can be obtained. At this time, the coordinate system of the towers is marked with latitude and longitude values. First, convert them from the latitude and longitude coordinate system to the geodetic space rectangular coordinate system.

[0180] Latitude and longitude are a collective term for longitude and latitude, forming a coordinate system called the geographic coordinate system. It is a spherical coordinate system that uses a three-dimensional sphere to define space on Earth and can mark any location on Earth.

[0181] The commonly used latitude and longitude coordinate system is the WGS-84 geodetic coordinate system (World Geodetic System-1984 Coordinate System), which is a geocentric and geofixed coordinate system established by the U.S. Department of Defense. The Global Positioning System (GPS) is based on the WGS-84 coordinate system.

[0182] 2. Coordinate Transformation: Assuming a point P(L,B,H) exists in the geodetic coordinate system, it needs to be transformed to the coordinates (X,Y,Z) of the geodetic rectangular coordinate system. The specific transformation process is as follows... Figure 3 and Figure 4 As shown:

[0183] The following is the coordinate system transformation calculation process for the inspection tower coordinate points: Given a point P(X,Y,Z) on the Earth's surface, the normal line passing through this point is used. The angle between this normal line and the equatorial plane is the latitude value of this point, described in this patent as value B. B is an angle value, ranging from 0-90°. Simultaneously, the angle between the plane formed by this point and the two poles and the plane containing the 0-degree meridian is the longitude value, described in this method as value L. L is also an angle value. The two-dimensional plane of the meridian is the xOy plane in the diagram, which is the plane through which the meridian of point P passes the Earth's axis. The extracted portion is... Figure 4 As shown.

[0184] First, determine the relationship between the coordinates (x, y) of point P in its meridian two-dimensional plane coordinate system O-xy and its latitude B:

[0185] x = |PK|cos B = (N+H)cos B

[0186] y = |PQ|sin B = [N(1-e 2 )+H]sin B

[0187] In the formula, PK and PQ represent the line segment and their respective lengths. (Here, the formula is modified by adding an absolute value symbol.)

[0188] Secondly, the coordinates (x, y) of the meridian in the two-dimensional plane are obtained in the geodetic rectangular coordinate system (X, Y, Z):

[0189] X = xcosβ, Y = xsinβ, Z = y

[0190] (x,y) represents the coordinates of a certain latitude and longitude point in the meridian two-dimensional plane. The task here is to transform the coordinates of this point in the plane to the coordinates (X,Y,Z) of the geodetic rectangular coordinate system. L is the angle between the x-axis of the meridian two-dimensional plane in the upper left figure and the x-axis of the geodetic rectangular coordinate system.

[0191] Finally, the coordinates (X,Y,Z) of P(L,B,H) in the geodetic coordinate system and P(L,B,H) in the geodetic rectangular coordinate system are obtained.

[0192]

[0193] Where N represents the radius of curvature of the ellipse, e 2 This represents the eccentricity of the ellipsoid.

[0194]

[0195] Where a represents the major semi-axis, which is 6378137m; b represents the minor semi-axis, which is 6356752.3142451793m;

[0196] After obtaining the elliptical radius of curvature N through the above transformation process, we can obtain the specific (X, Y, Z) coordinates of all towers on the required inspection route. Since this patent only involves orientation-related issues, and the inspection height is generally a fixed value, the correlation of coordinate Z is not considered here. Therefore, the data can be reduced from three-dimensional space to a two-dimensional plane.

[0197] Before starting an inspection mission, an initial inspection route needs to be set. Due to environmental obstacles, aircraft range, and efficiency considerations, using tower coordinates as inspection points would result in a large number of towers being traversed during a single inspection mission. Setting inspection points one by one would significantly reduce mission efficiency. Since inspections often use fixed-wing UAVs, excessive coordinated turns would also drastically reduce aircraft range. Moreover, in some practical situations, using tower coordinates as inspection points is not feasible. Considering the above issues, but in order to ensure the completion of the inspection mission and the accuracy of the inspection data, unused tower coordinates can be used as auxiliary data to assist the UAV in orientation, thereby better completing the inspection mission and ensuring the quality and effectiveness of the inspection data.

[0198] Step 2 Target Tower Track Planning Module

[0199] 1. Obtain the coordinates of the target tower using an algorithm.

[0200] The flight path planning section is explained as follows: Figure 5As shown, assume (X1,Y1) is the set previous inspection point; (X2,Y2) is the set next inspection point; (Xi,Yi) and (Xj,Yj) are the tower points passed through in this route; (Xi',Yi') and (Xj',Yj') are the projection points of the tower coordinates on the inspection route. When the aircraft reaches this route, the original inspection route is first replaced, changing (X1,Y1) to (X2,Y2) to (X1,Y1) to (Xi,Yi), (Xi,Yi) to (Xj,Yj), and (Xj,Yj) to (X2,Y2).

[0201] First, for the track segment from (X1,Y1) to (Xi,Yi), it can be determined using the formula...

[0202] (X-X1) / (Xi-X1)=(Y-Y1) / (Yi-Y1)

[0203] 2. Form coordinate points for trajectory planning

[0204] The state transition equation is calculated, and the new inspection route is divided into several target planning points at intervals of 0.1m according to the frequency of the extended Kalman filter algorithm in the UAV flight control. The points (x, y) of n points are used as the new target planning trajectory points. The calculation process starts from the initial value of x, increases x by 0.1 each time, calculates the corresponding y value, and saves these values ​​as the new target planning points.

[0205] like Figure 3 The dashed lines mark these points, which are then saved and used as input data for the UAV flight control position control loop. These points serve as the new target positions for the aircraft, and the flight control system moves the aircraft toward the target points to complete the tower orientation design.

[0206] The method for traversing routes (Xi,Yi) to (Xj,Yj) and (Xj,Yj) to (X2,Y2) is the same, only the required coordinate points differ. (There's no universal formula for how to substitute these coordinates; the result needs to be presented. The formula doesn't seem complicated; it simply requires calculating the routes (Xi,Yi) to (Xj,Yj) and (Xj,Yj) to (X2,Y2) using this method.)

[0207] The specific formula is as follows:

[0208] Formula for route (Xi,Yi) to (Xj,Yj)

[0209] (X-Xi) / (Xj-Xi)=(Y-Yi) / (Yj-Yi)

[0210] Formula for route (Xj,Yj) to (X2,Y2)

[0211] (X-Xj) / (X2-Xj)=(Y-Yj) / (Y2-Yj)

[0212] The specific locations are explained below. Figure 5 As shown: the coordinate points required to form the flight path planning.

[0213] Step 3 Target Tower Orientation Processing Module

[0214] 1. Extended Kalman Filter Algorithm for Data Fusion

[0215] During flight, the UAV uses an extended Kalman filter algorithm to filter and fuse IMU data with GPS data, calculates the actual position coordinates of the aircraft, and transmits them to the position control loop in the flight controller. By comparing the target coordinates with the actual coordinates, the flight status and attitude of the aircraft are controlled to ensure mission completion.

[0216] During flight, the UAV's position and attitude data are primarily calculated and updated in real time using the Extended Kalman Filter (EKF) algorithm. The EKF equation mainly consists of two parts: time prediction update and measurement correction update. The specific data filtering and fusion process employs the EKF formula, and the specific steps are as follows:

[0217] The first step is to estimate the current state based on the system state transition equations and the estimated state from the previous time step:

[0218] The predicted prior state estimate is:

[0219]

[0220] The second step is to predict the covariance of the current estimation error based on the covariance of the estimation error from the previous time step:

[0221] Prediction prior error covariance P k∣k-1 :

[0222]

[0223] The third step is to calculate the Kalman gain using the error covariance and the observation matrix;

[0224] Kalman gain K k

[0225]

[0226] The fourth step involves using the Kalman filter gain to compensate for the observations in the estimate from the first step, resulting in an updated state estimate.

[0227]

[0228] Step 5: Correct the estimation error covariance from step 2: Obtain the updated posterior estimation error covariance matrix.

[0229] P k∣k =(I n -K k H k )P k∣k-1 (5).

[0230] Of the five formulas mentioned above, only formula 1 is expressed in the form of a function f(x). This is because the extended Kalman filter is a state calculation for a nonlinear process, which covers a wide range of topics. The specific calculation process of formula (1) is as follows:

[0231] Formula (1) mainly describes how the IMU sensor provides predicted values ​​of the aircraft's position. The IMU consists of two sensors: an accelerometer and a gyroscope. The gyroscope obtains the real-time three-axis angular velocity ω of the aircraft, the accelerometer obtains the real-time three-axis acceleration values ​​of the aircraft, and the GPS obtains the actual measured position and velocity values ​​of the aircraft. The uses and calculation process of the IMU data are as follows:

[0232] a. The conversion process between gyroscope data and attitude quaternion q (attitude quaternion is used in the following two parts to calculate and predict the aircraft velocity value through integration of accelerometer data, so it is explained here)

[0233] Read the three-axis angular velocity ω from the gyroscope, integrate it to obtain the three-axis angular change Δang_meas, and then subtract the gyroscope offset Δang_bias (the inherent error of the sensor) to obtain the true angular change Δang_truth.

[0234]

[0235] Δ ang_truth =Δ ang_meas -Δ ang_bias

[0236] The change in quaternion Δ quat It can then be represented by the actual change in angle.

[0237]

[0238] Quaternion multiplication can be used to represent the rotation of the state from time K to time K+1, thus obtaining the quaternion state transition equation. The multiplication can be represented in matrix form as follows:

[0239]

[0240] b. Accelerometer data is used to predict aircraft velocity and position (to obtain the predicted position value):

[0241] First, read the velocity change from the inertial measurement unit (accelerometer in the IMU), note that this is in the body coordinate system. Then, subtract the velocity offset (its inherent error) to obtain the velocity change within the body coordinate system.

[0242]

[0243] Δ vel_truth =Δ vel_meas -Δ vel_bias Delocity bias states

[0244] Next, the velocity transfer equation in the ground coordinate system is derived. Since the accelerometer eliminates the influence of gravitational acceleration, this needs to be added when calculating the velocity at time K+1. It's also important to note the coordinate system; a transfer matrix derived from quaternions is used here. The specific transfer equation is shown below.

[0245]

[0246] The attitude quaternion (calculated from the gyroscope data mentioned earlier) is used here.

[0247]

[0248] c. Calculation of predicted values ​​for aircraft position state variables:

[0249] The displacement formula in junior high school physics is x = x0 + vt, so the displacement formula is relatively simple. The details are as follows:

[0250]

[0251] At this point, we have obtained the prior position estimate required for the extended Kalman filter, which completes equation (1).

[0252] The remaining formulas (2) to (5) can be calculated directly according to their contents.

[0253] The main function of formula (2) is to calculate the relevant covariance value obtained by formula 2 based on the error of the IMU sensor itself. This value can be calculated by formula and is subsequently used to expand the Kalman gain value.

[0254] Formula 3 calculates a weight, which we use to calculate the actual position of the aircraft. Currently, we have two data points: one is the position value calculated by integrating the aircraft's own IMU sensor, and the other is the value measured by the actual GPS. The weight of each is mainly determined by its own error covariance, which is determined by Formula 2.

[0255] After calculating the weight according to Formula 3, proceed to Formula 4. Formula 4 represents a fusion process of GPS measured position data and IMU integrated position data. The weight of the two is determined by the gain value calculated by Formula 3.

[0256] 2. Obtain the real-time location data of the target drone.

[0257] For formula (4), z′ k This is the current (i.e., the k-th) cycle state measurement value. Since GPS data directly provides the aircraft's three-axis position coordinates in the geodetic coordinate system, we directly use it as the aircraft's position observation value without any further transformation. and The actual position data of the target UAV is estimated by formula (4) based on the state variable m. The process is iterated, and in each iteration, the state quantity m of the target UAV at the current time K is calculated: The current values ​​in the table;

[0258] Equation (5) is an iterative update of the covariance matrix, which will provide relevant estimated covariance matrix values ​​for the next extended Kalman filter cycle.

[0259] The error covariance matrix P is obtained through formula (5). k∣k :P k∣k =(I n -K k H k )P k∣k-1 The error covariance matrix is ​​updated in each iteration by calculating the value of the state variable m at the next moment. The real-time position data of the target UAV is obtained by iterating through each iteration of formulas (4) and (5).

[0260] The meanings and related explanations of each term in the formula are as follows:

[0261] Among them, state variables Among them, attitude quaternions:

[0262]

[0263] V NED This represents the machine's velocity in the NED coordinate system.

[0264]

[0265] P NED This indicates the machine's position in the NED coordinate system:

[0266]

[0267] Δθ b This represents the incremental angular deviation (bias) in the XYZ body coordinate system:

[0268]

[0269] ΔV b This represents the incremental velocity bias in the XYZ body coordinate system.

[0270]

[0271] Additionally f:R n ×R m ×R n →R n and h:R n ×R m →R m It is a nonlinear function; m and n are parameters illustrating the nonlinear function, indicating the dimension of its internal terms; ω k-1 G is an ι×1 dimensional system noise vector, where ι is the dimension of the state parameters, assumed to be extracted from a zero-mean multivariate normal distribution N; k For the input control matrix; F k H is the state transition matrix; k Let Q be the observation matrix. Both Q and R follow a normal distribution, and Q... k :ω k ~N(0,Q) k ), R k :v k ~N(0,R k ).

[0272] k is the loop number. Let u be the posterior estimate of the (k-1)th iteration. k-1 This is the control matrix for the k-th cycle. F is a nonlinear function estimated a priori. k-1 Let P be the state transition matrix for the (k-1)th cycle. k-1∣k-1 G is the posterior estimate of the covariance for the (k-1)th iteration. k-1 Let Q be the control input matrix for the (k-1)th cycle. k-1 Let be the covariance matrix of the (k-1)th cycle, assuming it follows a multivariate zero-mean normal distribution. P is the transpose of the control input matrix for the (k-1)th iteration; k∣k-1 For the prior error covariance, H is the measurement matrix for the k-th cycle. k Let R′ be the observation matrix. k The observation noise with zero mean Gaussian covariance in the k-th cycle. For prior state estimation, and ∈ state variables m, z′ k The state measurement value for the current (i.e., the kth) cycle, z′ k|k-1 I is the state measurement value for the (k-1)th cycle. n It is an n-dimensional identity matrix.

[0273] This completes the calculation of this round of extended Kalman filtering. The current values ​​of the target UAV's flight state variables *m* at time *K* are calculated, along with the posterior estimation error covariance matrix. The calculated current position of the aircraft, along with the target UAV's trajectory planning coordinates obtained in step 2, are fed into the target UAV's flight control loop for comparison, thus achieving aircraft control. The error covariance matrix will be applied in the next extended Kalman filtering iteration, repeating the process to continuously calculate the UAV's actual position.

[0274] This invention provides a method for directional flight path planning for UAV inspection of power transmission line towers in mountainous areas. It solves the problem that the flight path planning formed by the initial inspection coordinate points cannot accurately give the precise location of the power transmission line towers, resulting in a large error between the tower coordinates given by the UAV based on the initial inspection coordinates and the actual position coordinates of the towers. This affects the efficiency of UAV inspection, reduces the accuracy of the inspected location area, and deteriorates the validity of the data. This invention acquires the latitude and longitude positions of the target tower and performs coordinate system transformation; it places the coordinates of the target tower and the initial inspection coordinates in the same coordinate system, and calculates and processes them using an extended Kalman filter algorithm to obtain n coordinate points for the target tower's trajectory planning. The invention then uses the extended Kalman filter to obtain the real-time position data of the UAV, and transmits the UAV's real-time position data and the n coordinate points of the target tower's trajectory planning to the target UAV's position control loop for comparison and input parameters. This significantly reduces the workload of ground personnel, ensures that all required inspection towers are covered and that the inspection data is valid, enables wireless monitoring of the operational status of mountain circuits within the inspection range, helps inspection personnel accurately and efficiently record the inspection results of mountain lines without climbing, provides effective and accurate target trajectory coordinates, and ensures the effective execution of the system.

Claims

1. A method for planning the directional flight path of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas, characterized in that, The method steps are as follows; Step 1: The data acquisition and processing module converts the pre-collected latitude and longitude coordinates of the target tower into the position coordinates of the target tower in the geodetic rectangular coordinate system, obtaining K position coordinates of the target tower, which are in the same coordinate system as the initial inspection coordinates. Step 2: The trajectory planning module calculates the target tower function of the K-1 segment of the route formed by connecting the position coordinates of K target towers in sequence. The coordinates of n target towers are obtained by frequency calculation and processing using the extended Kalman filter algorithm. The coordinates of the n target towers are the n coordinates of the target tower trajectory planning. The coordinates of the n target towers are the n coordinate points for the target tower trajectory planning. Specifically, the tower points passed between the previous inspection point P1(X1,Y1) and the next inspection point P2(X2,Y2) are Pi(Xi,Yi) and Pj(Xj,Yj); and the projection points on the tower coordinate point inspection line are (Xi',Yi') and (Xj',Yj'). When the target drone travels to the section from (X1,Y1) to (X2,Y2), the original inspection route is replaced, changing (X1,Y1) to (X2,Y2) to (X1,Y1) to (Xi,Yi), (Xi,Yi) to (Xj,Yj), and (Xj,Yj) to (X2,Y2). The formula for solving the equation of the straight line of the track from (X1,Y1) to (Xi,Yi) is: (X-X1) / (Xi-X1)=(Y-Y1) / (Yi-Y1) The formula for solving the equation of the straight line of the track from (Xi,Yi) to (Xj,Yj) is: (X-Xi) / (Xj-Xi)=(Y-Yi) / (Yj-Yi) The formula for solving the equation of the straight line of the trajectory from (Xj,Yj) to (X2,Y2) is: (X-Xj) / (X2-Xj)=(Y-Yj) / (Y2-Yj) Where i and j are both natural numbers, i < n, j < n, n > 1; After obtaining the radius of curvature N of the ellipse, the (X,Y,Z) of all target towers on the required inspection route are obtained; According to the formula for solving the straight line equation of the track, each inspection line segment is calculated in the same way. According to the frequency of the extended Kalman filter algorithm, each inspection line segment is set as the nth coordinate point of the target inspection tower track at a preset distance of d meters. Starting from the initial value of the x value of the nth target coordinate point, the preset value of d meters is added to x, and the y value of the nth target coordinate point is calculated. The nth (x, y) value is saved as the nth coordinate point of the target inspection tower track planning. Connecting the n target coordinate points forms the target inspection tower trajectory plan, where r is: r = {{x1,y1}, {x2,y2}, ..., {x...} n ,y n }}; Step 3: The tower orientation processing module obtains the real-time position data of the target UAV by processing the position data of the UAV during the flight process through an algorithm, and transmits the real-time position data of the target UAV and the target tower trajectory planning n coordinate points to the target UAV flight control position control loop; Specifically, the target inspection tower trajectory is planned with n coordinate points, and the tower orientation processing module obtains the real-time position data of the target UAV through the extended Kalman filter algorithm based on the position data of the UAV during the flight process. This data is then transmitted to the position control loop in the target UAV flight control system, providing relevant control input parameters for the UAV to achieve flight status control.

2. The method for directional flight path planning of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas according to claim 1, characterized in that, The data acquisition and processing module in step 1 converts the pre-acquired latitude and longitude coordinates of the target tower into the position coordinates of the target tower in a geodetic rectangular coordinate system. The specific steps are as follows: Step 201: Transform the latitude and longitude coordinates P(L,B,H) of the target tower in the geodetic coordinate system to the geodetic space rectangular coordinate system to obtain the coordinates P(X,Y,Z) of the target tower; obtain the coordinates (x,y) of the latitude B value of point P in the coordinate system O-xy of the meridian two-dimensional plane: x = |PK|cosB = (N+H)cosB y=|PQ|sinB=[N(1-e 2 )+H]sinB Wherein, PK and PQ represent the line segment lengths from point P to point K and from point P to point Q, respectively; Step 202 involves obtaining the two-dimensional plane coordinates (x, y) of the meridian in the geodetic rectangular coordinate system (X, Y, Z): X=xcosβ, Y=xsinβ, Z=y; Where β is the angle between the x-axis of the meridian two-dimensional plane and the x-axis of the geodetic rectangular coordinate system; x, y are the coordinates of the latitude and longitude of the desired point P in the meridian two-dimensional plane; Step 203 yields the coordinates (X, Y, Z) of the geodetic coordinate system P(L, B, H) in the geodetic rectangular coordinate system. Where N represents the radius of curvature of the ellipse; e 2 Represents the eccentricity of the ellipsoid: a is the major semi-axis, and b is the minor semi-axis.

3. The method for directional flight path planning of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas according to claim 1, characterized in that, The real-time position data of the target UAV is obtained by filtering and fusing IMU data and GPS data acquired by the target UAV during flight using an extended Kalman filter algorithm to obtain the actual position coordinates of the target aircraft. The extended Kalman filter algorithm combines IMU data with GPS data through filtering. Position data collected during the flight of the target aircraft is used to calculate and update the UAV's position and attitude data in real time through the extended Kalman filter algorithm. The extended Kalman filter algorithm equation includes time prediction update and measurement correction update. The steps of the extended Kalman filter algorithm are as follows: Step 301: Estimate the current state based on the state transition equation and the posterior estimate from the previous time step, and obtain the predicted prior state estimate. Where k is the loop number. Let u be the posterior estimate of the (k-1)th iteration. k-1 This is the control matrix for the k-th cycle. ω is a nonlinear function estimated a priori; k-1 It is the noise error matrix corresponding to each of the ι×1-dimensional parameters, where ι is the dimension of the state parameters; Step 302: Predict the covariance of the current estimation error based on the covariance of the estimation error at the previous time step. Prediction prior error covariance P k∣k-1 : Among them, F k-1 Let P be the state transition matrix for the (k-1)th cycle. k-1∣k-1 G is the posterior estimate of the covariance for the (k-1)th iteration. k-1 Let Q be the control input matrix for the (k-1)th cycle. k-1 Let be the covariance matrix of the (k-1)th cycle. This is the transpose of the control input matrix for the (k-1)th iteration; Step 303 calculates the Kalman gain K from the error covariance and the observation matrix. k : in, H is the measurement matrix for the k-th cycle. k Let R′ be the observation matrix. k The observation noise is the zero-mean Gaussian covariance of the k-th cycle; Step 304 uses the Kalman filter gain to compensate for the observations in the estimate from step 301, thus obtaining the updated posterior state estimate. in, For prior state estimation, and ∈ state variables m, z′ k Let z′ be the current state measurement value in the k-th cycle. k|k-1 This is the state measurement value for the (k-1)th cycle; state variables q is the attitude quaternion: V NED This represents the machine's velocity in the NED coordinate system. P NED This indicates the machine's position in the NED coordinate system: Δθ b This represents the incremental angular deviation (bias) in the XYZ body coordinate system: ΔV b This represents the incremental velocity deviation in the XYZ body coordinate system: Step 305 corrects the estimation error covariance from step 302: This yields the updated posterior estimation error covariance matrix. P k∣k =(I n -K k H k )P k∣k-1 (5) Among them, I n For an n-dimensional identity matrix, H k For the observation matrix, K k For Kalman gain, P k∣k-1 Let be the prior error covariance.

4. The method for directional flight path planning of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas, as described in claim 3, is characterized in that... Step 301 estimates the current state based on the state transition equation and the estimated state of the previous time step, and obtains the predicted prior state estimate: the IMU data and the attitude quaternion q are converted. The IMU data is read by two sensors, an accelerometer and a gyroscope. The gyroscope reads the real-time three-axis angular velocity ω value of the target UAV, the accelerometer obtains the real-time three-axis acceleration value of the target UAV in the body coordinate system, and the GPS data obtains the actual measured position and velocity values ​​of the aircraft. The gyroscope reads the three-axis angular velocity ω, and the three-axis angle change Δang_meas is obtained by integrating the three-axis angular velocity. Then, the gyroscope offset Δang_bias is subtracted to obtain the target UAV angle change Δang_truth. D ang_truth =D ang_meas -D ang_bias The change Δ of the quaternion q of the attitude quaternion quat The change in the angle of the target UAV is then used to represent it. Using quaternion multiplication to represent the rotation of the state from time K to time K+1, we obtain the quaternion state transition equation, and the multiplication is represented in matrix form as follows: Where K≥1.

5. The method for directional flight path planning of unmanned aerial vehicle (UAV) inspection of power transmission line towers in mountainous areas according to claim 4, characterized in that, The following steps are taken to obtain the three-axis acceleration values ​​of the target UAV in real time in the body coordinate system based on the accelerometer: Step 401: Read the velocity change from the accelerometer in the IMU data. This velocity change is in the body coordinate system. Then, subtract the velocity offset to obtain the velocity change Δ of the target UAV in the body coordinate system. vel_truth : D vel_truth =D vel_meas -D vel_bias Step 402: Based on the velocity change value calculated in the body coordinate system in step 401, obtain the aircraft velocity value in the ground coordinate system. Since the accelerometer itself eliminates the influence of gravitational acceleration, the influence of gravitational acceleration is added when calculating the velocity at time K+1, resulting in a quaternion-based transfer matrix. The quaternion-based transfer matrix is ​​as follows: in, It is the attitude rotation matrix, the attitude rotation matrix VN, VE, and VD are the calculated aircraft velocity values ​​in three directions in the ground coordinate system. Step 403: Based on the UAV's velocity value in the ground coordinate system calculated in step 402, calculate the state variables of the UAV's position. The values ​​of these state variables will also be used as the predicted values ​​for the extended Kalman filter. The prior position estimates are the predicted values ​​for the target UAV's flight position state, specifically as follows: Where P represents position, NED is the abbreviation for the direction name of the three axes of the geodetic coordinate system (northeast, east, and west), and PN, PE, and PD represent the position coordinates of the three directions.

6. The method for directional flight path planning of unmanned aerial vehicles (UAVs) for inspecting power transmission line towers in mountainous areas, as described in claim 3, is characterized in that... The actual position data of the target UAV is estimated by formula (4). The process is iterated, and in each iteration, the state quantity m of the target UAV at the current time K is calculated: The current values ​​in the table; The error covariance matrix P is obtained through formula (5). k∣k :P k∣k =(I n -K k H k )P k∣k-1 The error covariance matrix is ​​updated in each iteration by calculating the value of the state variable m at the next moment. The real-time position data of the target UAV is obtained by iterating through each iteration of formulas (4) and (5).

7. A method for directional flight path planning for unmanned aerial vehicle (UAV) inspection of power transmission line towers in mountainous areas, as described in claim 1 or 6, characterized in that... Based on the target inspection tower trajectory planning, n coordinate points (x) n ,y n The data is transmitted along with the real-time location data of the target UAV to the target UAV's position control loop, providing input parameters for the target UAV's position control loop to perform position comparison.

Images