Bucket wheel machine high-precision positioning and walking deviation correction control method based on multi-sensor fusion

Abstract

The invention provides a bucket wheel machine high-precision positioning and walking deviation correction control method based on multi-sensor fusion, and relates to the technical field of bulk material storage yard operation equipment control. Obtaining storage yard material point cloud data of a plurality of laser scanning sensors on the bucket wheel machine, bucket wheel machine body attitude data of a plurality of inertial measurement units and traveling wheel rotation turn number data of a plurality of encoders on a traveling mechanism; constructing a storage yard material three-dimensional contour model taking the walking track as a reference, and extracting a material elevation change curve; generating transverse and longitudinal deviation distances through correlation analysis with an ideal walking track elevation curve; dynamically compensating and correcting the deviation distance by using the attitude data and the rotation turn number data to obtain an actual space pose deviation vector; based on the actual space pose deviation vector, a deviation correction control instruction set containing a differential speed control instruction and a rotation angle correction instruction is generated, and high-precision positioning and walking deviation correction control of the bucket wheel machine are achieved.

Description

Technical Field

[0001] This invention relates to the field of control technology for bulk material storage yard operation equipment, and more specifically, to a high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion. Background Technology

[0002] In coal and ore stockpiling operations, bucket wheel excavators, as critical material handling equipment, directly impact operational efficiency and quality through their positioning accuracy and stability. Traditional bucket wheel excavator positioning methods often rely on single sensors or simple mechanical positioning devices. For example, some excavators use limit switches to determine approximate location, but these methods offer low positioning accuracy, only providing a rough estimate and failing to meet the demands of modern, efficient, and precise operations. Other methods employ GPS-based positioning; however, in stockpiling environments, the presence of numerous metal structures and materials obstructing the view makes GPS signals susceptible to interference, leading to significant positioning errors or even signal loss, severely impacting the excavator's normal operation.

[0003] In terms of travel deviation control, most existing methods only consider the speed control of the traveling wheels, attempting to correct the travel deviation of the bucket wheel excavator through simple differential speed adjustment. However, these methods do not fully consider the complex conditions of the material surface in the stockpile and the attitude changes of the bucket wheel excavator itself, such as the undulation of the material, the pitch and roll of the bucket wheel excavator, etc., resulting in unsatisfactory deviation correction effects. The bucket wheel excavator is still prone to deviation during travel, which not only reduces operating efficiency but may also damage the equipment and increase maintenance costs. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion, the method comprising:

[0005] The system acquires real-time point cloud data of the stockpile material from multiple laser scanning sensors installed on the bucket wheel excavator body, real-time attitude data of the bucket wheel excavator body from multiple inertial measurement units installed on the bucket wheel excavator body, and real-time rotation data of the traveling wheels from multiple encoders installed on the bucket wheel excavator traveling mechanism. Based on the overlapping coverage relationship of the scanning areas of different laser scanning sensors in the point cloud data of the stockpile material, a three-dimensional contour model of the stockpile material is constructed with the bucket wheel excavator's travel track as the reference. The three-dimensional contour model of the stockpile material includes material elevation change curves distributed along the extension direction of the travel track. By performing spatial position correlation analysis between the material elevation change curve in the three-dimensional contour model of the stockpile material and the elevation curve of the bucket wheel excavator's preset ideal travel trajectory, the lateral deviation distance and longitudinal deviation distance of the bucket wheel excavator at the current travel position are generated. By using the pitch angle, roll angle and heading angle in the attitude data of the bucket wheel excavator body, and combining the encoder pulse accumulation in the number of rotations of the traveling wheels, the lateral deviation distance and the longitudinal deviation distance are dynamically compensated and corrected to obtain the actual spatial attitude deviation vector of the bucket wheel excavator relative to the traveling track. A set of correction control instructions for the bucket wheel excavator's traveling mechanism is generated based on the actual spatial pose deviation vector. The set of correction control instructions includes differential control instructions for adjusting the speed difference between the traveling wheels on both sides of the bucket wheel excavator's traveling mechanism and slewing angle correction instructions for adjusting the cantilever slewing angle of the bucket wheel excavator.

[0006] Furthermore, embodiments of the present invention also provide a high-precision positioning and travel correction control system for bucket wheel excavators based on multi-sensor fusion, comprising: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion by executing the machine-executable instructions.

[0007] Based on the above, by integrating point cloud data of the stockpile material collected by multiple laser scanning sensors on the bucket wheel excavator body, attitude data of the bucket wheel excavator body collected by multiple inertial measurement units, and rotation data of the traveling wheels collected by multiple encoders on the traveling mechanism, a three-dimensional contour model of the stockpile material is constructed based on the overlapping coverage relationship of the scanning areas of different laser scanning sensors. The material elevation change curve is extracted from this model, which can characterize the morphological features of the stockpile material surface. By performing spatial position correlation analysis between the material elevation change curve and the preset ideal traveling trajectory elevation curve, lateral and longitudinal deviation distances are generated, achieving preliminary quantification of the bucket wheel excavator's position deviation. Dynamic compensation and correction of the deviation distance are performed using the bucket wheel excavator body attitude data and traveling wheel rotation data to obtain the actual spatial pose deviation vector, further improving the positioning accuracy and effectively overcoming the limitations of single-sensor positioning and the influence of environmental factors on positioning accuracy. A correction control command set is generated based on the actual spatial pose deviation vector, including differential speed control commands and slewing angle correction commands. This enables precise correction of the bucket wheel excavator's movement from multiple dimensions, ensuring that the bucket wheel excavator always travels stably along the ideal trajectory, greatly improving the operating efficiency and safety of the bucket wheel excavator. Attached Figure Description

[0008] Figure 1This is a schematic diagram of the execution flow of the high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion provided in an embodiment of the present invention.

[0009] Figure 2 This is a schematic diagram of exemplary hardware and software components of a high-precision positioning and travel correction control system for bucket wheel excavators based on multi-sensor fusion, provided in an embodiment of the present invention. Detailed Implementation

[0010] Figure 1 This is a flowchart illustrating a high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion, provided in one embodiment of the present invention. A detailed description follows.

[0011] Step S110: Obtain the point cloud data of the stockpile material in real time from multiple laser scanning sensors installed on the bucket wheel excavator body, the attitude data of the bucket wheel excavator body in real time from multiple inertial measurement units installed on the bucket wheel excavator body, and the number of rotations of the traveling wheels in real time from multiple encoders installed on the bucket wheel excavator traveling mechanism.

[0012] In this embodiment, a bucket wheel excavator used in a large open-pit mine stockpile is used as a specific scenario for illustration. A first laser scanning sensor and a second laser scanning sensor are installed at the front end of the bucket wheel excavator body. The first laser scanning sensor is located on the left side of the cantilever, and the second laser scanning sensor is located on the right side of the cantilever. The scanning planes of both the first and second laser scanning sensors are perpendicular to the ground.

[0013] The bucket wheel excavator body is also equipped with a first inertial measurement unit, which is fixed at the geometric center of the main structure. This unit is used to measure the pitch, roll, and yaw angles of the excavator body in real time. The excavator's traveling mechanism includes a left-side traveling wheel set and a right-side traveling wheel set. The first encoder of the left-side traveling wheel set is mounted on the left drive wheel axle, and the second encoder of the right-side traveling wheel set is mounted on the right drive wheel axle.

[0014] After the bucket wheel excavator begins operation, the first laser scanning sensor continuously operates at a preset first scanning frequency, collecting real-time point cloud data of the first material in the forward stockpile. This first material point cloud data is a set of first spatial points in the first local coordinate system of the first laser scanning sensor. Each first spatial point in the set consists of a first lateral coordinate value, a first longitudinal coordinate value, and a first vertical coordinate value. The second laser scanning sensor continuously operates at a preset second scanning frequency, collecting real-time point cloud data of the second material in the forward stockpile. This second material point cloud data is a set of second spatial points in the second local coordinate system of the second laser scanning sensor. Each second spatial point in the set consists of a second lateral coordinate value, a second longitudinal coordinate value, and a second vertical coordinate value. The first inertial measurement unit outputs real-time attitude data of the bucket wheel excavator body. This attitude data is a data packet containing three angular components: the pitch angle around the lateral axis of the bucket wheel excavator body, the roll angle around the longitudinal axis of the bucket wheel excavator body, and the yaw angle around the vertical axis of the bucket wheel excavator body. The first encoder records the number of rotations of the left travel wheel in real time. The data output by the first encoder is the cumulative pulse count of the left encoder, which records the total number of pulses that the left travel wheel has rotated from the initial moment to the current moment. The second encoder records the number of rotations of the right travel wheel in real time. The data output by the second encoder is the cumulative pulse count of the right encoder, which records the total number of pulses that the right travel wheel has rotated from the initial moment to the current moment.

[0015] Step S120: Based on the overlapping coverage relationship of the scanning areas corresponding to different laser scanning sensors in the point cloud data of the stockpile material, construct a three-dimensional contour model of the stockpile material with the bucket wheel excavator's traveling track as the reference. The three-dimensional contour model of the stockpile material includes material elevation change curves distributed along the extension direction of the traveling track.

[0016] Step S121: Divide the plurality of laser scanning sensors into a left scanning sensor group and a right scanning sensor group according to their installation positions.

[0017] The first and second laser scanning sensors are grouped according to their physical installation locations. Since the first laser scanning sensor is installed on the left side of the bucket wheel excavator cantilever, it is grouped into the left-side scanning sensor group. Since the second laser scanning sensor is installed on the right side of the bucket wheel excavator cantilever, it is grouped into the right-side scanning sensor group.

[0018] Step S122: Extract the spatial coordinates of each point cloud in the material point cloud data collected by the left scanning sensor group, and perform coordinate transformation based on the lateral coordinate value in the spatial coordinates of each point cloud and the installation position coordinates of the left scanning sensor group to generate the left point cloud coordinate set of the left point cloud data corresponding to the left scanning sensor group in a unified spatial coordinate system.

[0019] Acquire the first material point cloud data collected by the first laser scanning sensor in the left-side scanning sensor group. For each first spatial point in the first material point cloud data, extract the first lateral coordinate value, first longitudinal coordinate value, and first vertical coordinate value of the first spatial point in the first local coordinate system of the first laser scanning sensor itself. Read the pre-calibrated first installation position coordinates of the first laser scanning sensor. The first installation position coordinates are three-dimensional coordinate values ​​in a pre-established unified spatial coordinate system for the entire field. This unified spatial coordinate system has one corner of the stockpile as the origin, the direction of the travel track as the longitudinal axis, the direction perpendicular to the travel track as the lateral axis, and the vertical height direction as the vertical axis. Acquire the pre-calibrated first self-attitude angle values ​​of the first laser scanning sensor, which include the first pitch angle value, the first roll angle value, and the first heading angle value. Construct a first rigid body transformation matrix based on the first installation position coordinates and the first self-attitude angle values. The first rigid body transformation matrix includes a three-dimensional rotation matrix and a three-dimensional translation vector. The first horizontal coordinate, first vertical coordinate, and first perpendicular coordinate of the first spatial point are used to construct the first original coordinate vector. This first original coordinate vector is then multiplied sequentially by the three-dimensional rotation matrix and added to the three-dimensional translation vector to obtain the first transformed coordinate value of the first spatial point in the unified spatial coordinate system. This coordinate transformation process is repeated for each first spatial point in the first material point cloud data. The first transformed coordinate values ​​corresponding to all first spatial points constitute the left-side point cloud coordinate set.

[0020] Step S123: Extract the spatial coordinates of each point cloud in the material point cloud data collected by the right scanning sensor group, and perform coordinate transformation based on the lateral coordinate values ​​in the spatial coordinates of each point cloud and the installation position coordinates of the right scanning sensor group to generate the right point cloud coordinate set of the right point cloud data corresponding to the right scanning sensor group in a unified spatial coordinate system.

[0021] Acquire the second material point cloud data collected by the second laser scanning sensor in the right-side scanning sensor group. For each second spatial point in the second material point cloud data, extract the second lateral coordinate value, second longitudinal coordinate value, and second vertical coordinate value of the second spatial point in the second local coordinate system of the second laser scanning sensor itself. Read the pre-calibrated second installation position coordinates of the second laser scanning sensor, which are three-dimensional coordinate values ​​in a unified spatial coordinate system across the entire field. Acquire the pre-calibrated second self-attitude angle values ​​of the second laser scanning sensor, which include the second pitch angle value, second roll angle value, and second yaw angle value. Construct a second rigid body transformation matrix based on the second installation position coordinates and the second self-attitude angle values. The second rigid body transformation matrix contains a three-dimensional rotation matrix and a three-dimensional translation vector. Construct a second original coordinate vector from the second lateral coordinate values, second longitudinal coordinate values, and second vertical coordinate values ​​of the second spatial point. Multiply the second original coordinate vector with the three-dimensional rotation matrix and then add it with the three-dimensional translation vector to obtain the second transformed coordinate value of the second spatial point in the unified spatial coordinate system across the entire field. Repeat the above coordinate transformation process for each second spatial point in the second material point cloud data. The second transformed coordinate values ​​corresponding to all second spatial points constitute the point cloud coordinate set on the right.

[0022] Step S124: Perform spatial coordinate fusion on the left point cloud coordinate set and the right point cloud coordinate set, and identify the point cloud coordinate points whose spatial coordinates coincide with those in the left and right point cloud coordinate sets as the overlapping area point cloud coordinate set.

[0023] After obtaining the left and right point cloud coordinate sets, the overlapping areas of the two point cloud coordinate sets are identified. Each left point cloud coordinate point in the left point cloud coordinate set is traversed. For the currently traversed left point cloud coordinate point, the spatial Euclidean distance between this left point cloud coordinate point and each right point cloud coordinate point in the right point cloud coordinate set is calculated. The spatial Euclidean distance D = [(XL-XR)^2 + (YL-YR)^2 + (ZL-ZR)^2]^(1 / 2), where XL, YL, and ZL are the horizontal, vertical, and lateral coordinate values ​​of the left point cloud coordinate point, and XR, YR, and ZR are the horizontal, vertical, and lateral coordinate values ​​of the right point cloud coordinate point. When the calculated spatial Euclidean distance D is less than a preset overlap determination distance threshold Dth, it is determined that the left and right point cloud coordinate points represent the same physical point on the stockpile material and belong to the overlapping area. All left-side point cloud coordinates that meet the above conditions, along with their matching right-side point cloud coordinates, constitute the overlapping region point cloud coordinate set.

[0024] Step S125: Based on the spatial coordinates of each point cloud coordinate point in the overlapping area point cloud coordinate set, calculate the coordinate deviation value between the left and right point cloud coordinate sets in the overlapping area, and use the coordinate deviation value to perform global coordinate consistency calibration on the left and right point cloud coordinate sets to obtain the calibrated left and right point cloud coordinate sets.

[0025] After obtaining the coordinate set of the overlapping region point cloud, for each pair of matching points in the overlapping region point cloud coordinate set, calculate the three-dimensional coordinate deviation vector between the spatial coordinates of the left and right point cloud coordinates of the matching point pair. The horizontal coordinate deviation component ΔX = XL - XR, the vertical coordinate deviation component ΔY = YL - YR, and the vertical coordinate deviation component ΔZ = ZL - ZR are all calculated. The calculated three-dimensional coordinate deviation vectors for all matching point pairs in the overlapping region point cloud coordinate set are summed to obtain the total deviation vector. Then, each component of the total deviation vector is divided by the total number of matching point pairs N to obtain the average deviation vector (ΔXavg, ΔYavg, ΔZavg). This average deviation vector is used as the global coordinate deviation correction. The global coordinate deviation correction is applied to every point cloud coordinate point in the right-hand point cloud coordinate set. Specifically, the horizontal coordinate deviation component ΔXavg of the average deviation vector is added to the horizontal coordinate value of each right-hand point cloud coordinate point; the vertical coordinate deviation component ΔYavg of the average deviation vector is added to the vertical coordinate value of the right-hand point cloud coordinate point; and the vertical coordinate deviation component ΔZavg of the average deviation vector is added to the vertical coordinate value of the right-hand point cloud coordinate point. After this correction, the left-hand point cloud coordinate set remains unchanged, while the right-hand point cloud coordinate set is translated to a position that is spatially aligned precisely with the left-hand point cloud coordinate set, resulting in the calibrated left-hand and right-hand point cloud coordinate sets.

[0026] Step S126: Divide the calibrated left point cloud coordinate set and the calibrated right point cloud coordinate set into a spatial grid according to the extension direction of the walking track. Divide the yard area into multiple continuous grid cells along the extension direction of the walking track, with each grid cell corresponding to a segment of the walking track.

[0027] Define a fixed grid step size Lgrid, which represents the length of each grid cell in the direction of the walking track. Starting from the beginning of the walking track, divide the entire walking track into a series of continuous intervals with Lgrid as the interval. Each interval corresponds to a grid cell, and the position of the walking track interval corresponding to the i-th grid cell is from (i-1). Lgrid to i Lgrid. Each grid cell covers the entire possible width of the stockpile in the lateral direction perpendicular to the walking track, ranging from the lateral minimum Xmin to the lateral maximum Xmax.

[0028] Step S127: Within each grid cell, extract the vertical coordinate values ​​of all point cloud coordinate points in the left and right point cloud coordinate subsets corresponding to that grid cell, and perform statistical processing on the vertical coordinate values ​​to obtain the material representative elevation value corresponding to each grid cell.

[0029] For each grid cell, from the calibrated left and right point cloud coordinate sets, select all point cloud points whose spatial coordinates lie within the boundary of that grid cell, forming a point cloud subset corresponding to that grid cell. The criterion for determining whether a point cloud point is within a grid cell is: the point cloud point's vertical coordinate value lies within [(i-1)]. Lgrid, i Within the [Lgrid] interval, the horizontal coordinate value of the point cloud point lies within the interval [Xmin, Xmax]. Extract the vertical coordinate value of each point cloud point in this subset. Perform statistical processing on all vertical coordinate values ​​corresponding to this grid cell, and calculate the average value of all vertical coordinate values ​​as the material representative elevation value of this grid cell. The material representative elevation value Hi = (Z1 + Z2 + ... + ZM) / M, where M is the total number of point cloud points in the subset corresponding to this grid cell, and Z1 to ZM are the vertical coordinate values ​​of each point cloud point.

[0030] Step S128: Associate and map the location of the walking track interval corresponding to each grid cell with the material representative elevation value corresponding to each grid cell to generate a material elevation change curve with the walking track interval location as the abscissa and the material representative elevation value as the ordinate.

[0031] The location of the travel track interval corresponding to each grid cell is associated with the material representative elevation value of that grid cell to form a data pair, wherein the location of the travel track interval is taken as the coordinate value of the center position of the grid cell, Pi=(i-0.5). Lgrid. The data pairs of all grid cells constitute a discrete data sequence. Using the position Pi of the travel track interval as the x-axis and the material's representative elevation value Hi as the y-axis, the above data points are connected sequentially to obtain a continuous material elevation variation curve.

[0032] Step S129: Superimpose the material elevation change curve with the spatial geometric parameters of the bucket wheel excavator's travel track to establish a three-dimensional contour model of the stockpile material based on the bucket wheel excavator's travel track. The three-dimensional contour model of the stockpile material includes the spatial three-dimensional coordinates corresponding to each coordinate point on the material elevation change curve.

[0033] Step S1291: Obtain the spatial geometric parameter set of the bucket wheel excavator's travel track. The spatial geometric parameter set includes the track centerline coordinate sequence, track width, track height, and track extension direction vector in a unified spatial coordinate system.

[0034] Read the pre-stored spatial geometric parameter set of the bucket wheel excavator's travel track. This spatial geometric parameter set includes: the coordinates of a series of discrete points on the centerline of the travel track in a unified spatial coordinate system across the entire field. These discrete points are sampled at equal intervals according to the track mileage, forming a coordinate sequence of the track centerline. The coordinates of the j-th sampled point are represented as (Xtrack_j, Ytrack_j, Ztrack_j), where Xtrack_j is the lateral coordinate value, Ytrack_j is the longitudinal coordinate value, and Ztrack_j is the vertical coordinate value; the physical width value of the track, Wtrack; the height value of the track rail, Htrack; and a unit direction vector describing the track's extension direction, represented as (Vx, Vy, Vz).

[0035] Step S1292: Map the position of the travel track interval corresponding to each curve point on the material elevation change curve to the coordinate sequence of the track centerline. Calculate the track centerline coordinate value corresponding to each curve point based on the linear interpolation relationship between the travel track interval position and the track centerline coordinate sequence.

[0036] For each point on the material elevation change curve, its corresponding track interval position Pi is determined. Two sampling points adjacent to Pi are found in the track centerline coordinate sequence. Let the longitudinal coordinates of these two sampling points be Ytrack_k and Ytrack_k+1, satisfying Ytrack_k ≤ Pi ≤ Ytrack_k+1. The spatial coordinates of these two sampling points are (Xtrack_k, Ytrack_k, Ztrack_k) and (Xtrack_k+1, Ytrack_k+1, Ztrack_k+1), respectively. Based on the linear interpolation relationship, the precise track centerline spatial coordinates at the track interval position Pi are calculated. The lateral coordinate value of the track centerline is Xcenter_i = Xtrack_k + (Xtrack_k+1 - Xtrack_k). (Pi-Ytrack_k) / (Ytrack_k+1-Ytrack_k), where Ycenter_i is the longitudinal coordinate of the track centerline and Zcenter_i is the vertical coordinate of the track centerline and Zcenter_i is the vertical coordinate of the track centerline and Ztrack_k+(Ztrack_k+1-Ztrack_k). (Pi-Ytrack_k) / (Ytrack_k+1-Ytrack_k).

[0037] Step S1293: Using the coordinates of the track centerline corresponding to each curve point as the reference point, and combining the track width and the track height, construct a track cross-section geometric model corresponding to each curve point. The track cross-section geometric model includes the spatial coordinates of each key point on the track cross-section.

[0038] For each curve point on the material elevation change curve, using the calculated track centerline coordinates (Xcenter_i, Ycenter_i, Zcenter_i) as the reference point, and combining the track width Wtrack and track height Htrack, a track cross-sectional geometric model corresponding to that curve point is constructed. The track cross-sectional geometric model includes the spatial coordinates of each key point on the track cross-section. These key points include the left edge point, right edge point, left top surface point, and right top surface point. The spatial coordinates of the left edge point are (Xcenter_i - Wtrack / 2, Ycenter_i, Zcenter_i), the right edge point is (Xcenter_i + Wtrack / 2, Ycenter_i, Zcenter_i), the left top surface point is (Xcenter_i - Wtrack / 2, Ycenter_i, Zcenter_i + Htrack), and the right top surface point is (Xcenter_i + Wtrack / 2, Ycenter_i, Zcenter_i + Htrack).

[0039] Step S1294: Take the material representative elevation value corresponding to each curve point as the elevation offset of the track cross-section geometric model corresponding to that curve point in the vertical direction, and superimpose the elevation offset onto the spatial coordinates of each key point in the track cross-section geometric model to generate the material cross-section contour point cloud coordinate set corresponding to each curve point.

[0040] For each curve point on the material elevation change curve, the representative material elevation value Hi corresponding to that curve point is used as the vertical elevation offset of the track cross-section geometric model corresponding to that curve point. The elevation offset Hi is superimposed onto the vertical coordinates of the spatial coordinates of each key point in the track cross-section geometric model to generate a set of material cross-section contour point cloud coordinates for each curve point. Specifically, for the left edge point of the track, the superimposed spatial coordinates are (Xcenter_i-Wtrack / 2, Ycenter_i, Zcenter_i+Hi); for the right edge point of the track, the superimposed spatial coordinates are (Xcenter_i+Wtrack / 2, Ycenter_i, Zcenter_i+Hi); for the left side of the top surface of the track, the superimposed spatial coordinates are (Xcenter_i-Wtrack / 2, Ycenter_i, Zcenter_i+Htrack+Hi); and for the right side of the top surface of the track, the superimposed spatial coordinates are (Xcenter_i+Wtrack / 2, Ycenter_i, Zcenter_i+Htrack+Hi). All the above-mentioned superimposed key points together constitute the point cloud coordinate set of the material cross-section contour corresponding to the curve point.

[0041] Step S1295: Connect the point cloud coordinate sets of the material cross-section contour corresponding to adjacent curve points continuously, and fit the point cloud coordinates of the corresponding positions in the point cloud coordinate sets of the material cross-section contour of adjacent curve points to the spatial curve according to the extension direction of the walking track, so as to generate a continuous curved surface mesh model of the material surface in three-dimensional space.

[0042] For the i-th and (i+1)-th curve points, extract the corresponding material cross-sectional contour point cloud coordinate sets. Fit the key points in the material cross-sectional contour point cloud coordinate set of the i-th curve point to the corresponding key points in the material cross-sectional contour point cloud coordinate set of the (i+1)-th curve point using spatial curve fitting along the direction of the travel track. Specifically, connect the left edge point of the track of the i-th curve point to the left edge point of the track of the (i+1)-th curve point to form a spatial curve along the direction of the travel track; connect the right edge point of the track of the i-th curve point to the right edge point of the track of the (i+1)-th curve point to form another spatial curve; connect the left edge point of the top surface of the track of the i-th curve point to the left edge point of the top surface of the track of the (i+1)-th curve point to form a third spatial curve; and connect the right edge point of the top surface of the track of the i-th curve point to the right edge point of the top surface of the track of the (i+1)-th curve point to form a fourth spatial curve. These four spatial curves, together with the material cross-sectional contours of adjacent curve points, enclose a continuous surface mesh model of the material surface in three-dimensional space.

[0043] Step S1296: Extract the spatial three-dimensional coordinates of each grid node in the continuous surface mesh model, associate and store the spatial three-dimensional coordinates of each grid node with the corresponding travel track interval position, and form a material space coordinate mapping table indexed by the travel track interval position.

[0044] In the continuous surface mesh model, the spatial three-dimensional coordinates of each mesh node are extracted. Each mesh node corresponds to a specific travel track interval position and a specific position on the cross-section. The spatial three-dimensional coordinates of each mesh node are associated and stored with the corresponding travel track interval position Pi, forming a material spatial coordinate mapping table indexed by the travel track interval position Pi. The data structure of this mapping table is a set of key-value pairs, where the key is the travel track interval position Pi, and the value is the set of spatial three-dimensional coordinates of each key point on the material cross-sectional profile corresponding to that position.

[0045] Step S1297: Based on the material space three-dimensional coordinates corresponding to the positions of each walking track interval in the material space coordinate mapping table, construct the spatial three-dimensional coordinates corresponding to each coordinate point on the material elevation change curve. The spatial three-dimensional coordinates include the position coordinates along the extension direction of the walking track, the lateral position coordinates perpendicular to the extension direction of the walking track, and the elevation coordinates in the vertical direction.

[0046] Based on the material space coordinates corresponding to the positions Pi of each travel track interval in the material space coordinate mapping table, construct the spatial three-dimensional coordinates of each coordinate point on the material elevation change curve. The position coordinate along the travel track extension direction of each coordinate point on the material elevation change curve is the travel track interval position Pi corresponding to that curve point. The lateral position coordinate perpendicular to the travel track extension direction is taken as the lateral coordinate value Xcenter_i of the track centerline, and the vertical elevation coordinate is taken as the material representative elevation value Hi corresponding to that curve point. Therefore, the spatial three-dimensional coordinates corresponding to each coordinate point on the material elevation change curve are represented as (Xcenter_i, Pi, Hi).

[0047] Step S1298: Solidify the correspondence between the position of the travel track interval corresponding to each coordinate point on the material elevation change curve and the three-dimensional spatial coordinates to generate a three-dimensional outline model of the stockpile material based on the bucket wheel excavator travel track.

[0048] The correspondence between the position Pi of the travel track interval corresponding to each coordinate point on the material elevation change curve and the corresponding three-dimensional spatial coordinates (Xcenter_i, Pi, Hi) of that coordinate point is solidified to generate a three-dimensional contour model of the stockpile material based on the bucket wheel excavator's travel track. This three-dimensional contour model of the stockpile material includes the three-dimensional spatial coordinates corresponding to each coordinate point on the material elevation change curve distributed along the extension direction of the travel track, and fully describes the three-dimensional geometry of the stockpile material surface.

[0049] Step S130: Perform spatial position correlation analysis by comparing the material elevation change curve in the three-dimensional contour model of the stockpile material with the preset ideal travel trajectory elevation curve of the bucket wheel excavator, and generate the lateral deviation distance and longitudinal deviation distance of the bucket wheel excavator at the current travel position.

[0050] Step S131: Obtain the number of rotations of the traveling wheel at the current moment from the encoder installed on the traveling mechanism of the bucket wheel excavator. Calculate the theoretical traveling distance of the traveling mechanism along the traveling track based on the product of the encoder pulse accumulation in the traveling wheel rotation data and the circumference of the traveling wheel.

[0051] Obtain the current cumulative pulse value Pleft from the first encoder and the current cumulative pulse value Pright from the second encoder. Read the circumference Cleft of the left travel wheel and the circumference Cright of the right travel wheel. Calculate the theoretical travel distance Dleft = Pleft of the left travel wheel. Cleft calculates the theoretical walking distance of the right-side wheel: Dright = Pright Cright. The average of the theoretical travel distance Dleft of the left traveling wheel and the theoretical travel distance Dright of the right traveling wheel is taken as the theoretical travel distance Dtheoretical of the bucket wheel excavator's traveling mechanism along the traveling track: Dtheoretical = (Dleft + Dright) / 2.

[0052] Step S132: Based on the theoretical walking distance, locate the corresponding first elevation curve point on the material elevation change curve of the three-dimensional contour model of the stockpile material, and extract the spatial three-dimensional coordinates corresponding to the first elevation curve point as the measured material spatial coordinates corresponding to the current walking position.

[0053] Based on the calculated theoretical walking distance Dtheoretical, the corresponding first elevation curve point is located on the material elevation change curve of the three-dimensional contour model of the stockpile material. Specifically, the curve point on the material elevation change curve that is closest to the theoretical walking distance Dtheoretical within the walking track interval position Pi is found, and this curve point is taken as the first elevation curve point. The spatial three-dimensional coordinates corresponding to the first elevation curve point are extracted. These spatial three-dimensional coordinates include the position coordinates along the extension direction of the walking track, the lateral position coordinates perpendicular to the extension direction of the walking track, and the vertical elevation coordinates. These spatial three-dimensional coordinates are taken as the measured material spatial coordinates corresponding to the current walking position, represented as (Xmeasured, Ymeasured, Zmeasured).

[0054] Step S133: Based on the theoretical walking distance, locate the corresponding second elevation curve point on the preset ideal walking trajectory elevation curve of the bucket wheel excavator, and extract the spatial three-dimensional coordinates corresponding to the second elevation curve point as the spatial coordinates of the ideal walking trajectory corresponding to the current walking position.

[0055] The pre-set ideal travel trajectory elevation curve of the bucket wheel excavator is read. This ideal travel trajectory elevation curve describes the ideal spatial coordinates of each position of the bucket wheel excavator when it travels along the travel track under ideal conditions. Based on the calculated theoretical travel distance Dtheoretical, the corresponding second elevation curve point is located on the ideal travel trajectory elevation curve. Specifically, the curve point on the ideal travel trajectory elevation curve whose position coordinates are closest to the theoretical travel distance Dtheoretical is found, and this curve point is taken as the second elevation curve point. The spatial three-dimensional coordinates corresponding to the second elevation curve point are extracted. These spatial three-dimensional coordinates include the position coordinates along the extension direction of the travel track, the lateral position coordinates perpendicular to the extension direction of the travel track, and the vertical elevation coordinates. These spatial three-dimensional coordinates are taken as the ideal travel trajectory spatial coordinates corresponding to the current travel position, represented as (Xideal, Yideal, Zideal).

[0056] Step S134: Input the measured material spatial coordinates and the ideal walking trajectory spatial coordinates into the spatial coordinate difference calculation unit, and calculate the difference between the measured material spatial coordinates and the ideal walking trajectory spatial coordinates in the horizontal coordinate axis direction as the horizontal deviation distance, and the difference between the measured material spatial coordinates and the ideal walking trajectory spatial coordinates in the vertical coordinate axis direction as the vertical deviation distance.

[0057] The measured material space coordinates (Xmeasured, Ymeasured, Zmeasured) are compared with the ideal travel trajectory space coordinates (Xideal, Yideal, Zideal). The lateral deviation distance ΔX_deviation = Xmeasured - Xideal is calculated, representing the lateral offset of the bucket wheel excavator from its current position relative to the ideal travel trajectory. The longitudinal deviation distance ΔY_deviation = Ymeasured - Xideal is calculated, representing the longitudinal offset of the bucket wheel excavator from its current position relative to the ideal travel trajectory.

[0058] Step S135: Mark the lateral deviation distance and the longitudinal deviation distance with symbols. Determine the lateral deviation direction mark of the lateral deviation distance and the longitudinal deviation direction mark of the longitudinal deviation distance based on the spatial position relationship between the measured material spatial coordinates and the ideal travel trajectory spatial coordinates. The lateral deviation direction mark is used to indicate the lateral offset direction of the bucket wheel excavator relative to the ideal travel trajectory, and the longitudinal deviation direction mark is used to indicate the longitudinal offset direction of the bucket wheel excavator relative to the ideal travel trajectory.

[0059] The lateral deviation direction is determined by the sign of the lateral deviation distance ΔX_deviation. If ΔX_deviation is greater than 0, the lateral deviation direction is positive, indicating that the bucket wheel excavator has deviated laterally in a positive direction relative to the ideal travel trajectory; if ΔX_deviation is less than 0, the lateral deviation direction is negative, indicating that the bucket wheel excavator has deviated laterally in a negative direction relative to the ideal travel trajectory. The longitudinal deviation direction is determined by the sign of the longitudinal deviation distance ΔY_deviation. If ΔY_deviation is greater than 0, the longitudinal deviation direction is positive, indicating that the bucket wheel excavator has deviated longitudinally in a positive direction relative to the ideal travel trajectory; if ΔY_deviation is less than 0, the longitudinal deviation direction is negative, indicating that the bucket wheel excavator has deviated longitudinally in a negative direction relative to the ideal travel trajectory.

[0060] Step S136: Proportionally correlate the lateral deviation distance with the structural width of the bucket wheel excavator's traveling mechanism to generate a ratio of lateral deviation distance to structural width as a lateral relative deviation coefficient; proportionally correlate the longitudinal deviation distance with the structural length of the bucket wheel excavator's traveling mechanism to generate a ratio of longitudinal deviation distance to structural length as a longitudinal relative deviation coefficient.

[0061] Read the structural width Wstructure and structural length Lstructure of the bucket wheel excavator's traveling mechanism. Calculate the lateral relative deviation coefficient Rlateral = ΔX_deviation / Wstructure, which represents the ratio of the lateral deviation distance to the structural width of the bucket wheel excavator's traveling mechanism. Calculate the longitudinal relative deviation coefficient Rlongitudinal = ΔY_deviation / Lstructure, which represents the ratio of the longitudinal deviation distance to the structural length of the bucket wheel excavator's traveling mechanism.

[0062] Step S137: Obtain the sequence of lateral relative deviation coefficients and the sequence of longitudinal relative deviation coefficients recorded by the bucket wheel excavator traveling mechanism during the historical traveling process. Extract the historical lateral relative deviation coefficients corresponding to multiple consecutive historical moments from the sequence of lateral relative deviation coefficients, and extract the historical longitudinal relative deviation coefficients corresponding to multiple consecutive historical moments from the sequence of longitudinal relative deviation coefficients.

[0063] The lateral and longitudinal relative deviation coefficient sequences recorded by the bucket wheel excavator's traveling mechanism during historical travel are retrieved from the historical database. The lateral relative deviation coefficient sequence records the lateral relative deviation coefficient values ​​Rlateral_t1, Rlateral_t2, ..., Rlateral_tn corresponding to historical times t1, t2, ..., tn. The longitudinal relative deviation coefficient sequence records the longitudinal relative deviation coefficient values ​​Rlongitudinal_t1, Rlongitudinal_t2, ..., Rlongitudinal_tn corresponding to historical times t1, t2, ..., tn. The historical lateral relative deviation coefficients corresponding to the K consecutive historical times preceding the current time are extracted from the lateral relative deviation coefficient sequence, i.e., Rlateral_t-K+1, Rlateral_t-K+2, ..., Rlateral_t. The historical longitudinal relative deviation coefficients corresponding to the K consecutive historical times preceding the current time are extracted from the longitudinal relative deviation coefficient sequence, i.e., Rlongitudinal_t-K+1, Rlongitudinal_t-K+2, ..., Rlongitudinal_t.

[0064] Step S138: Construct a lateral deviation trend characteristic curve based on the historical lateral relative deviation coefficients corresponding to the consecutive multiple historical moments, and calculate the slope of the lateral deviation trend characteristic curve on the time axis as the lateral deviation change rate.

[0065] Using historical timestamps t-K+1, t-K+2, ..., t as the x-axis and the corresponding historical relative lateral deviation coefficients Rlateral_t-K+1, Rlateral_t-K+2, ..., Rlateral_t as the y-axis, a lateral deviation trend characteristic curve is constructed. This curve is then linearly fitted, and the slope of the fitted line is calculated as the lateral deviation change rate Slateral. The lateral deviation change rate Slateral represents the speed and direction of change of the relative lateral deviation coefficient over time.

[0066] Step S139: Construct a longitudinal deviation trend characteristic curve based on the historical longitudinal relative deviation coefficients corresponding to the consecutive multiple historical moments, and calculate the slope of the longitudinal deviation trend characteristic curve on the time axis as the longitudinal deviation change rate.

[0067] Using historical timestamps t-K+1, t-K+2, ..., t as the x-axis and the corresponding historical longitudinal relative deviation coefficients Rlongitudinal_t-K+1, Rlongitudinal_t-K+2, ..., Rlongitudinal_t ​​as the y-axis, a longitudinal deviation trend characteristic curve is constructed. This curve is then linearly fitted, and the slope of the fitted line is calculated as the longitudinal deviation change rate Slongitudinal. The longitudinal deviation change rate Slongitudinal represents the speed and direction of change of the longitudinal relative deviation coefficient over time.

[0068] Step S1310: Perform a correlation analysis between the current lateral relative deviation coefficient and the lateral deviation change rate, and generate a lateral deviation cumulative effect parameter based on the product of the lateral relative deviation coefficient and the lateral deviation change rate.

[0069] Multiplying the current relative lateral deviation coefficient Rlateral_current with the lateral deviation rate of change Slateral yields the lateral deviation cumulative effect parameter Clateral_accum = Rlateral_current. Slateral. The parameter Clateral_accum, representing the cumulative effect of lateral bias, is used to characterize the extent of the cumulative impact of lateral bias under the current trend of change.

[0070] Step S1311: Perform a correlation analysis between the longitudinal relative deviation coefficient at the current moment and the longitudinal deviation change rate, and generate a longitudinal deviation cumulative effect parameter based on the product of the longitudinal relative deviation coefficient and the longitudinal deviation change rate.

[0071] Multiply the current longitudinal relative deviation coefficient Rlongitudinal_current by the longitudinal deviation change rate Slongitudinal to obtain the longitudinal deviation cumulative effect parameter Clongitudinal_accum = Rlongitudinal_current. Slongitudinal. The parameter Clongitudinal_accum, representing the cumulative effect of longitudinal deviation under the current trend of change, is used to characterize the extent of the cumulative impact of longitudinal deviation.

[0072] Step S1312: Obtain the traveling speed of the bucket wheel excavator's traveling mechanism, correlate the lateral deviation cumulative effect parameter with the traveling speed to generate a lateral deviation speed coupling factor, and correlate the longitudinal deviation cumulative effect parameter with the traveling speed to generate a longitudinal deviation speed coupling factor.

[0073] Obtain the current travel speed Vcurrent of the bucket wheel excavator's traveling mechanism. Multiply the lateral deviation cumulative effect parameter Clateral_accum with the current travel speed Vcurrent to obtain the lateral deviation speed coupling factor Flateral_coupling = Clateral_accum. Vcurrent. The longitudinal deviation cumulative effect parameter Clongitudinal_accum is multiplied by the current walking speed Vcurrent to obtain the longitudinal deviation velocity coupling factor Flongitudinal_coupling = Clongitudinal_accum. Vcurrent.

[0074] Step S1313: The lateral relative deviation coefficient, the lateral deviation change rate, the lateral deviation cumulative effect parameter, and the lateral deviation velocity coupling factor are vectorized and combined to generate a comprehensive lateral deviation feature vector.

[0075] The current relative lateral deviation coefficient Rlateral_current, the lateral deviation rate of change Slateral, the lateral deviation cumulative effect parameter Clateral_accum, and the lateral deviation velocity coupling factor Flateral_coupling are vectorized and combined in a predetermined order to generate a comprehensive lateral deviation feature vector Vlateral_feature=[Rlateral_current, Slateral, Clateral_accum, Flateral_coupling]. This comprehensive lateral deviation feature vector Vlateral_feature is a four-dimensional vector that comprehensively describes the current state, trend, cumulative effect, and coupling relationship with velocity of the lateral deviation.

[0076] Step S1314: The longitudinal relative deviation coefficient, the longitudinal deviation change rate, the longitudinal deviation cumulative effect parameter, and the longitudinal deviation velocity coupling factor are vectorized and combined to generate a longitudinal deviation comprehensive feature vector.

[0077] The longitudinal relative deviation coefficient Rlongitudinal_current, the longitudinal deviation change rate Slongitudinal, the longitudinal deviation cumulative effect parameter Clongitudinal_accum, and the longitudinal deviation velocity coupling factor Flongitudinal_coupling at the current moment are vectorized and combined in a predetermined order to generate a comprehensive longitudinal deviation feature vector Vlongitudinal_feature=[Rlongitudinal_current, Slongitudinal, Clongitudinal_accum, Flongitudinal_coupling]. This comprehensive longitudinal deviation feature vector Vlongitudinal_feature is a four-dimensional vector that comprehensively describes the current state, trend, cumulative effect, and coupling relationship with velocity of the longitudinal deviation.

[0078] Step S1315: Input the lateral deviation comprehensive feature vector and the longitudinal deviation comprehensive feature vector into the pose deviation prediction model, and perform cross-correlation analysis on the lateral deviation comprehensive feature vector and the longitudinal deviation comprehensive feature vector through the feature fusion layer of the pose deviation prediction model to generate the coupling influence coefficient of lateral deviation and longitudinal deviation.

[0079] The lateral deviation feature vector (Vlateral_feature) and the longitudinal deviation feature vector (Vlongitudinal_feature) are input into the pose deviation prediction model. The model includes a feature fusion layer that concatenates the two vectors to form an eight-dimensional fused feature vector, Vfusion = [Vlateral_feature, Vlongitudinal_feature]. The model then uses a multi-layer fully connected network to perform nonlinear transformations and feature extraction on the fused feature vector Vfusion, ultimately outputting the coupling influence coefficient Ccoup between the lateral and longitudinal deviations. This coefficient Ccoup is a value between 0 and 1, representing the degree of mutual influence between the lateral and longitudinal deviations.

[0080] Step S1316: Perform secondary correction on the lateral deviation distance and the longitudinal deviation distance according to the coupling influence coefficient to obtain the secondary corrected lateral deviation distance and the secondary corrected longitudinal deviation distance.

[0081] The calculated coupling effect coefficient Ccoup is applied to the lateral deviation distance ΔX_deviation and the longitudinal deviation distance ΔY_deviation for secondary correction. The lateral deviation distance after secondary correction is ΔX_corrected = ΔX_deviation. (1+Ccoup α), where α is a preset coupling influence weighting coefficient. The longitudinal deviation distance after secondary correction is ΔY_corrected = ΔY_deviation. (1+Ccoup β), where β is a preset coupling effect weighting coefficient. Through secondary correction, the coupling effect between lateral and longitudinal deviations is incorporated into the deviation calculation, resulting in a more accurate deviation value.

[0082] Step S140: Using the pitch angle, roll angle and heading angle in the attitude data of the bucket wheel excavator body, combined with the encoder pulse accumulation in the number of rotations of the traveling wheels, the lateral deviation distance and the longitudinal deviation distance are dynamically compensated and corrected to obtain the actual spatial attitude deviation vector of the bucket wheel excavator relative to the traveling track.

[0083] Step S141: Extract the pitch angle of the bucket wheel machine body at the current moment from the attitude data of the bucket wheel machine body, and calculate the horizontal projection offset of the front end point of the bucket wheel machine on the horizontal plane caused by the pitch motion of the bucket wheel machine cantilever due to the product of the pitch angle value and the cantilever length of the bucket wheel machine.

[0084] Extract the pitch angle value θpitch from the attitude data of the bucket wheel excavator body output by the first inertial measurement unit. Read the cantilever length Lboom of the bucket wheel excavator. Calculate the horizontal projection offset of the bucket wheel excavator's front end point on the horizontal plane caused by the cantilever pitch motion: Offsetpitch = Lboom sin(θpitch). This horizontal projection offset, Offsetpitch, represents the horizontal offset distance of the front point relative to the root of the cantilever caused by the cantilever pitch motion.

[0085] Step S142: Extract the roll angle of the bucket wheel excavator body at the current moment from the attitude data of the bucket wheel excavator body, and calculate the vertical displacement difference of the two traveling wheels of the bucket wheel excavator in the vertical direction caused by the roll motion of the bucket wheel excavator body based on the product of the roll angle value and the width of the bucket wheel excavator body.

[0086] Extract the roll angle θroll from the attitude data of the bucket wheel excavator body output by the first inertial measurement unit. Read the width Wbody of the bucket wheel excavator body. Calculate the vertical displacement difference ΔZroll = Wbody between the two traveling wheels of the bucket wheel excavator in the vertical direction caused by the roll motion of the bucket wheel excavator body. sin(θroll). This vertical displacement difference ΔZroll represents the height difference in the vertical direction of the two traveling wheels caused by the lateral roll of the bucket wheel excavator body.

[0087] Step S143: Extract the heading angle of the bucket wheel excavator body at the current moment from the attitude data of the bucket wheel excavator body, and calculate the change in trajectory curvature of the bucket wheel excavator's travel trajectory caused by the heading deflection motion of the bucket wheel excavator based on the ratio of the heading angle value to the wheel gauge of the bucket wheel excavator's traveling mechanism.

[0088] Extract the current heading angle value θyaw from the attitude data of the bucket wheel excavator output by the first inertial measurement unit. Read the wheelbase Wwheelbase of the bucket wheel excavator's traveling mechanism. Calculate the change in trajectory curvature Curvaturechange = tan(θyaw) / Wwheelbase caused by the heading deflection motion of the bucket wheel excavator. This change in trajectory curvature Curvaturechange represents the influence of the heading angle change on the curvature of the traveling trajectory.

[0089] Step S144: Extract the encoder pulse accumulation of the left walking wheel and the encoder pulse accumulation of the right walking wheel from the walking wheel rotation data. Calculate the actual walking distance of the left walking wheel based on the product of the encoder pulse accumulation of the left walking wheel and the circumference of the left walking wheel. Calculate the actual walking distance of the right walking wheel based on the product of the encoder pulse accumulation of the right walking wheel and the circumference of the right walking wheel.

[0090] Obtain the current cumulative pulse value Pleft from the first encoder and the current cumulative pulse value Pright from the second encoder. Read the circumference Cleft and Cright of the left and right traveling wheels. Calculate the actual travel distance Dleft_actual = Pleft of the left traveling wheel. Cleft calculates the actual walking distance of the right-side wheel. Dright_actual = Pright Cright.

[0091] Step S145: Calculate the difference between the actual travel distance of the left travel wheel and the actual travel distance of the right travel wheel as the travel difference between the left and right travel wheels. Based on the ratio of the travel difference between the left and right travel wheels to the distance between the travel wheels of the bucket wheel excavator, calculate the change in the heading deflection angle of the bucket wheel excavator body caused by the asynchronous movement of the left and right travel wheels.

[0092] Calculate the travel difference between the left and right traveling wheels, ΔDtrack = Dleft_actual - Dright_actual. Read the wheel spacing, Wtrack_spacing. Calculate the change in the bucket wheel excavator's heading angle, Δθyaw_drive = ΔDtrack / Wtrack_spacing, caused by the asynchronous movement of the left and right traveling wheels. This change in heading angle, Δθyaw_drive, represents the change in the bucket wheel excavator's orientation due to the asynchronous movement of the traveling wheels.

[0093] Step S146: Input the horizontal projection offset, the vertical displacement difference, the trajectory curvature change, and the heading deflection angle change into the attitude compensation matrix operation unit to generate a lateral compensation factor for correcting the lateral deviation distance and a longitudinal compensation factor for correcting the longitudinal deviation distance.

[0094] The horizontal projection offset (Offsetpitch), vertical displacement difference (ΔZroll), trajectory curvature change (Curvaturechange), and heading deflection angle change (Δθyaw_drive) are input into the attitude compensation matrix operation unit. The attitude compensation matrix operation unit constructs an attitude compensation matrix Mcompensation, a 4x2 matrix containing lateral and longitudinal compensation coefficients. Through matrix multiplication, the input four-dimensional vector [Offsetpitch, ΔZroll, Curvaturechange, Δθyaw_drive] is multiplied by the attitude compensation matrix Mcompensation to obtain a two-dimensional output vector [Flateral_comp, Flongitudinal_comp], where Flateral_comp is the lateral compensation factor and Flongitudinal_comp is the longitudinal compensation factor.

[0095] Step S147: Multiply the lateral compensation factor with the lateral deviation distance to obtain the corrected lateral deviation distance, and multiply the longitudinal compensation factor with the longitudinal deviation distance to obtain the corrected longitudinal deviation distance.

[0096] The lateral deviation distance ΔX_corrected obtained in step S1316 after secondary correction is multiplied by the lateral compensation factor Flateral_comp to obtain the corrected lateral deviation distance ΔX_final = ΔX_corrected. Flateral_comp. The longitudinal deviation distance ΔY_corrected obtained in step S1316 after secondary correction is multiplied by the longitudinal compensation factor Flongitudinal_comp to obtain the corrected longitudinal deviation distance ΔY_final = ΔY_corrected. Flongitudinal_comp.

[0097] Step S148: Obtain the historical posture deviation vector sequence accumulated during the historical travel of the bucket wheel excavator, and extract the historical lateral deviation distance and historical longitudinal deviation distance corresponding to multiple consecutive historical moments adjacent to the current moment from the historical posture deviation vector sequence.

[0098] The historical posture deviation vector sequence accumulated during the historical travel of the bucket wheel excavator is retrieved from the historical database. This sequence records the posture deviation vectors corresponding to historical times t1, t2, ..., tn, and each vector contains a historical lateral deviation distance and a historical longitudinal deviation distance. From this sequence, the historical lateral deviation distances ΔX_history_t-M+1, ΔX_history_t-M+2, ..., ΔX_history_t and the historical longitudinal deviation distances ΔY_history_t-M+1, ΔY_history_t-M+2, ..., ΔY_history_t corresponding to the M consecutive historical times preceding the current time are extracted.

[0099] Step S149: Construct a lateral deviation time series prediction model based on the changing trend of the historical lateral deviation distance corresponding to the consecutive multiple historical moments, and construct a longitudinal deviation time series prediction model based on the changing trend of the historical longitudinal deviation distance corresponding to the consecutive multiple historical moments.

[0100] Based on the historical lateral deviation distances ΔX_history_t-M+1, ΔX_history_t-M+2, ..., ΔX_history_t corresponding to M consecutive historical time points, an autoregressive moving average model is used to construct a lateral deviation time series prediction model. This autoregressive moving average model determines the autoregressive order and moving average order of the model by analyzing the autocorrelation and moving average characteristics of the historical lateral deviation distances, and then fits the parameters of the lateral deviation time series prediction model. Based on the historical longitudinal deviation distances ΔY_history_t-M+1, ΔY_history_t-M+2, ..., ΔY_history_t corresponding to M consecutive historical time points, the same autoregressive moving average model method is used to construct a longitudinal deviation time series prediction model.

[0101] Step S1410: Input the corrected lateral deviation distance into the lateral deviation time series prediction model to predict the lateral deviation at future times, and obtain the predicted lateral deviation distance. Input the corrected longitudinal deviation distance into the longitudinal deviation time series prediction model to predict the longitudinal deviation at future times, and obtain the predicted longitudinal deviation distance.

[0102] The corrected lateral deviation distance ΔX_final is input as the current observation value into the lateral deviation time series prediction model. Based on its internal autoregressive and moving average parameters, the lateral deviation time series prediction model outputs the predicted lateral deviation distance ΔX_pred for the next time step. Similarly, the corrected longitudinal deviation distance ΔY_final is input as the current observation value into the longitudinal deviation time series prediction model. Based on its internal autoregressive and moving average parameters, the longitudinal deviation time series prediction model outputs the predicted longitudinal deviation distance ΔY_pred for the next time step.

[0103] Step S1411: The corrected lateral deviation distance, the corrected longitudinal deviation distance, the predicted lateral deviation distance, the predicted longitudinal deviation distance, and the travel difference of the left and right traveling wheels are vectorized and combined to generate the actual spatial pose deviation vector of the bucket wheel machine relative to the traveling track. The actual spatial pose deviation vector includes lateral deviation components, longitudinal deviation components, predicted lateral deviation components, predicted longitudinal deviation components, and traveling wheel synchronization deviation components.

[0104] The corrected lateral deviation distance ΔX_final is used as the lateral deviation component, the corrected longitudinal deviation distance ΔY_final as the longitudinal deviation component, the predicted lateral deviation distance ΔX_pred as the predicted lateral deviation component, the predicted longitudinal deviation distance ΔY_pred as the predicted longitudinal deviation component, and the travel difference ΔDtrack as the travel wheel synchronization deviation component. These components are vectorized and combined in a predetermined order to generate the actual spatial pose deviation vector Vpose=[ΔX_final, ΔY_final, ΔX_pred, ΔY_pred, ΔDtrack] of the bucket wheel excavator relative to the travel track. This actual spatial pose deviation vector Vpose is a five-dimensional vector that comprehensively describes the current actual spatial pose deviation state of the bucket wheel excavator, its future deviation trend, and the synchronization status of the travel wheels.

[0105] Step S150: Generate a correction control instruction set for the bucket wheel excavator traveling mechanism based on the actual spatial pose deviation vector. The correction control instruction set includes a differential control instruction for adjusting the speed difference between the traveling wheels on both sides of the bucket wheel excavator traveling mechanism and a slewing angle correction instruction for adjusting the slewing angle of the bucket wheel excavator cantilever.

[0106] Step S151: Analyze the lateral deviation component in the actual spatial pose deviation vector, and generate a differential control command for driving the left and right traveling wheels of the bucket wheel excavator to generate a speed difference based on the value of the lateral deviation component and the lateral deviation direction identifier. The differential control command includes the target speed of the left traveling wheel and the target speed of the right traveling wheel.

[0107] The lateral deviation component ΔX_final is extracted from the actual spatial pose deviation vector Vpose, and the corresponding lateral deviation direction indicator is obtained. Based on the absolute value of the lateral deviation component ΔX_final and the lateral deviation direction indicator, the speed adjustment amount in the differential control command is determined. If the lateral deviation direction indicator is positive, the target speed of the left travel wheel is set to Nleft_target = Nbase + Kdiff. |ΔX_final|, Target rotational speed of the right-side wheel Nright_target=Nbase-Kdiff |ΔX_final|, where Nbase is the base speed and Kdiff is the differential control gain coefficient. If the lateral deviation direction indicator is negative, then the target speed of the left travel wheel is set to Nleft_target = Nbase - Kdiff. |ΔX_final|, Target rotational speed of the right-side wheel: Nright_target=Nbase+Kdiff |ΔX_final|.

[0108] Step S152: Analyze the lateral deviation prediction component in the actual spatial pose deviation vector, and generate a pre-steering control command for adjusting the travel direction of the bucket wheel excavator travel mechanism in advance based on the value of the lateral deviation prediction component and the lateral deviation prediction direction identifier. The pre-steering control command includes the pre-adjustment speed of the left travel wheel and the pre-adjustment speed of the right travel wheel.

[0109] The lateral deviation prediction component ΔX_pred is extracted from the actual spatial pose deviation vector Vpose, and the lateral deviation prediction direction indicator is determined based on the sign of this component. The speed pre-adjustment amount in the presteering control command is determined based on the absolute value of the lateral deviation prediction component ΔX_pred and the lateral deviation prediction direction indicator. If the lateral deviation prediction direction indicator is positive, the pre-adjustment speed of the left travel wheel is set to Nleft_pre = Nbase + Kpre. |ΔX_pred|, Pre-adjustment speed of the right-side travel wheel Nright_pre=Nbase-Kpre |ΔX_pred|, where Kpre is the pre-steering control gain coefficient. If the lateral deviation prediction direction indicator is negative, then the pre-adjustment speed of the left travel wheel is set to Nleft_pre = Nbase - Kpre. |ΔX_pred|, Pre-adjustment speed of the right-side travel wheel Nright_pre=Nbase+Kpre |ΔX_pred|.

[0110] Step S153: The target speed of the left travel wheel in the differential control command is superimposed with the pre-adjusted speed of the left travel wheel in the pre-steering control command to generate the final speed control parameter of the left travel wheel. The target speed of the right travel wheel in the differential control command is superimposed with the pre-adjusted speed of the right travel wheel in the pre-steering control command to generate the final speed control parameter of the right travel wheel.

[0111] The target speed of the left travel wheel in the differential control command, Nleft_target, is added to the pre-adjustment speed of the left travel wheel in the pre-steering control command, Nleft_pre, to obtain the final control parameter for the left travel wheel's speed, Nleft_final = Nleft_target + Nleft_pre. Similarly, the target speed of the right travel wheel in the differential control command, Nright_target, is added to the pre-adjustment speed of the right travel wheel in the pre-steering control command, Nright_pre, to obtain the final control parameter for the right travel wheel's speed, Nright_final = Nright_target + Nright_pre.

[0112] Step S154: Analyze the longitudinal deviation component in the actual spatial pose deviation vector, and generate a speed control command for adjusting the traveling speed of the bucket wheel excavator traveling mechanism based on the value of the longitudinal deviation component and the longitudinal deviation direction identifier. The speed control command includes the target speed of the traveling mechanism.

[0113] The longitudinal deviation component ΔY_final is extracted from the actual spatial pose deviation vector Vpose, and the corresponding longitudinal deviation direction indicator is obtained. Based on the absolute value of the longitudinal deviation component ΔY_final and the longitudinal deviation direction indicator, the speed adjustment amount in the speed control command is determined. If the longitudinal deviation direction indicator is positive, it indicates that the bucket wheel excavator is ahead of the ideal trajectory, and the target speed of the traveling mechanism is set to Vtarget = Vbase - Kv. |ΔY_final|, where Vbase is the base speed and Kv is the speed control gain coefficient. If the longitudinal deviation direction indicator is negative, it means that the bucket wheel excavator is lagging behind the ideal trajectory, then the target speed of the traveling mechanism is set to Vtarget = Vbase + Kv. |ΔY_final|.

[0114] Step S155: Analyze the longitudinal deviation prediction component in the actual spatial pose deviation vector, and generate an acceleration pre-adjustment command for adjusting the travel acceleration of the bucket wheel excavator's traveling mechanism in advance based on the value of the longitudinal deviation prediction component and the longitudinal deviation prediction direction identifier. The acceleration pre-adjustment command includes the target acceleration of the traveling mechanism.

[0115] The longitudinal deviation prediction component ΔY_pred is extracted from the actual spatial pose deviation vector Vpose, and the longitudinal deviation prediction direction indicator is determined based on the sign of this component. The acceleration pre-adjustment amount in the acceleration pre-adjustment command is determined based on the absolute value of the longitudinal deviation prediction component ΔY_pred and the longitudinal deviation prediction direction indicator. If the longitudinal deviation prediction direction indicator is positive, it indicates that the bucket wheel excavator is predicted to be ahead, and the target acceleration of the traveling mechanism is set to Atarget = -Ka. |ΔY_pred|, where Ka is the acceleration control gain coefficient. If the longitudinal deviation prediction direction indicator is negative, it indicates that the bucket wheel excavator will lag in the future, then the target acceleration of the traveling mechanism is set to Atarget = +Ka. |ΔY_pred|.

[0116] Step S156: Integrate and correlate the target speed of the walking mechanism in the speed control command with the target acceleration of the walking mechanism in the acceleration pre-adjustment command to generate control parameters for the speed change curve of the walking mechanism.

[0117] The target speed Vtarget of the traveling mechanism in the speed control command is taken as the target final value of the speed change curve, and the target acceleration Atarget of the traveling mechanism in the acceleration pre-adjustment command is taken as the acceleration slope of the speed change curve. Through integration, a speed change curve smoothly transitions from the current speed Vcurrent to the target speed Vtarget is generated. The speed value at each time point on this speed change curve is determined by V(t) = Vcurrent + ∫Atargetdt, where the integration interval extends from the current time to the time when the target speed is reached. This speed change curve serves as the control parameter for the traveling mechanism's speed change curve.

[0118] Step S157: Analyze the synchronous deviation component of the traveling wheels in the actual spatial pose deviation vector, and generate a synchronous balance control command for balancing the speed of the traveling wheels on both sides of the bucket wheel excavator traveling mechanism based on the value of the synchronous deviation component of the traveling wheels. The synchronous balance control command includes compensation for the speed of the left traveling wheel and compensation for the speed of the right traveling wheel.

[0119] The synchronization deviation component ΔDtrack of the traveling wheels is extracted from the actual spatial pose deviation vector Vpose. The speed compensation amount in the synchronization balance control command is determined based on the value of ΔDtrack. The speed compensation of the left traveling wheel is calculated as Nleft_comp = -Ksync. ΔDtrack, right-side wheel speed compensation Nright_comp=+Ksync ΔDtrack, where Ksync is the synchronous balance control gain coefficient. When ΔDtrack is positive, it means that the actual travel distance of the left travel wheel is greater than that of the right. In this case, negative compensation is needed to reduce the speed of the left travel wheel, and positive compensation is needed to increase the speed of the right travel wheel to reduce the travel difference.

[0120] Step S158: The final speed control parameter of the left traveling wheel is fused with the speed compensation of the left traveling wheel in the synchronous balance control command to generate the execution speed of the left traveling wheel; the final speed control parameter of the right traveling wheel is fused with the speed compensation of the right traveling wheel in the synchronous balance control command to generate the execution speed of the right traveling wheel.

[0121] The final speed control parameter Nleft_final of the left traveling wheel is added to the speed compensation Nleft_comp of the left traveling wheel in the synchronization balance control command, resulting in the left traveling wheel execution speed Nleft_exec = Nleft_final + Nleft_comp. Similarly, the final speed control parameter Nright_final of the right traveling wheel is added to the speed compensation Nright_comp of the right traveling wheel in the synchronization balance control command, resulting in the right traveling wheel execution speed Nright_exec = Nright_final + Nright_comp.

[0122] Step S159: Generate a differential drive execution signal for the bucket wheel excavator's traveling mechanism based on the operating speed of the left traveling wheel and the operating speed of the right traveling wheel, and generate a speed drive execution signal for the bucket wheel excavator's traveling mechanism based on the control parameters of the traveling mechanism's speed change curve.

[0123] The left-side traveling wheel rotation speed Nleft_exec is converted into a first frequency control signal for the first drive motor, which controls the rotation speed of the left-side traveling wheel drive motor. The right-side traveling wheel rotation speed Nright_exec is converted into a second frequency control signal for the second drive motor, which controls the rotation speed of the right-side traveling wheel drive motor. The first and second frequency control signals together constitute the differential drive execution signal. The speed variation curve control parameters of the traveling mechanism are converted into a speed control signal for the traveling mechanism drive motor. This speed control signal contains the desired speed value at each time point and is used to control the overall traveling speed of the traveling mechanism.

[0124] Step S1510: Analyze the lateral deviation component and the longitudinal deviation component in the actual spatial pose deviation vector. Based on the spatial vector synthesis relationship between the lateral deviation component and the longitudinal deviation component, generate a slewing angle correction command for adjusting the cantilever slewing angle of the bucket wheel excavator. The slewing angle correction command includes the cantilever slewing angle adjustment value and the cantilever slewing direction identifier.

[0125] Step S15101: Extract the value of the lateral deviation component and the lateral deviation direction identifier from the actual spatial pose deviation vector, and extract the value of the longitudinal deviation component and the longitudinal deviation direction identifier from the actual spatial pose deviation vector.

[0126] Extract the values ​​of the lateral deviation component ΔX_final and the corresponding lateral deviation direction identifier from the actual spatial pose deviation vector Vpose, and extract the values ​​of the longitudinal deviation component ΔY_final and the corresponding longitudinal deviation direction identifier.

[0127] Step S15102: Construct a spatial vector model of the current position and orientation deviation of the bucket wheel excavator based on the values ​​of the lateral deviation component and the longitudinal deviation component. The spatial vector model takes the center point of the bucket wheel excavator's traveling mechanism as the vector starting point and the vector direction of the vector synthesis of the lateral deviation component and the longitudinal deviation component as the vector direction.

[0128] A spatial vector model is constructed using the center point of the bucket wheel excavator's traveling mechanism as the vector starting point, the lateral deviation component ΔX_final as the lateral vector component, and the longitudinal deviation component ΔY_final as the longitudinal vector component. The vector direction of this spatial vector model is determined by the vector synthesis of the lateral and longitudinal deviation components. The angle θvector between the vector synthesis direction and the lateral coordinate axis is calculated using the arctangent function atan2(ΔY_final, ΔX_final).

[0129] Step S15103: Calculate the angle between the vector direction of the spatial vector model and the current rotation direction of the bucket wheel excavator boom as the boom rotation angle deviation value.

[0130] Obtain the angle θcurrent_arm between the current rotation direction of the bucket wheel excavator boom and the longitudinal axis of the travel track. Calculate the angle Δθarm = θvector - θcurrent_arm between the vector direction of the spatial vector model and the current rotation direction of the boom. This angle Δθarm is used as the deviation value of the boom rotation angle.

[0131] Step S15104: Determine the cantilever rotation direction indicator based on the symbol combination relationship between the lateral deviation direction indicator and the longitudinal deviation direction indicator. The cantilever rotation direction indicator is used to indicate whether the cantilever rotation angle adjustment direction is clockwise or counterclockwise.

[0132] The cantilever slewing direction is determined by the combination of the symbols of the lateral and longitudinal deviation direction indicators. When both the lateral and longitudinal deviation direction indicators are positive, the cantilever slewing direction is clockwise. When both are negative, the cantilever slewing direction is counter-clockwise. Similarly, when both are negative and positive, the cantilever slewing direction is clockwise.

[0133] Step S15105: The cantilever rotation angle deviation value is proportionally correlated with the rotation angle range of the bucket wheel excavator cantilever to generate a cantilever rotation angle adjustment value, which is used to indicate the angle value that the cantilever needs to rotate.

[0134] Read the maximum slewing angle range θarm_max of the bucket wheel excavator boom. Proportionally correlate the boom slewing angle deviation value Δθarm with the maximum slewing angle range θarm_max, and calculate the boom slewing angle adjustment value θarm_adjust=Δθarm. (θarm_max / θarm_max_norm), where θarm_max_norm is a normalized reference value. This cantilever rotation angle adjustment value θarm_adjust represents the required rotation angle of the cantilever.

[0135] Step S15106: Obtain material point cloud data on the cantilever rotation path in real time from multiple laser scanning sensors installed on the cantilever of the bucket wheel excavator, and construct a material elevation change curve on the cantilever rotation path based on the material point cloud data on the cantilever rotation path.

[0136] The material point cloud data along the cantilever's rotation path is acquired in real time by the third and fourth laser scanning sensors installed on the cantilever of the bucket wheel excavator. Using the same method as steps S121 to S128, coordinate transformation, fusion calibration, mesh generation, and statistical processing are performed on the point cloud data acquired by the third and fourth laser scanning sensors to construct a material elevation variation curve along the cantilever's rotation path. This material elevation variation curve describes the material height distribution that the cantilever may encounter during rotation.

[0137] Step S15107: Input the cantilever rotation angle adjustment value into the cantilever rotation path planning unit, and perform obstacle avoidance correction on the cantilever rotation angle adjustment value according to the material elevation change curve on the cantilever rotation path to generate the obstacle avoidance corrected cantilever rotation angle adjustment value.

[0138] The cantilever rotation angle adjustment value θarm_adjust is input into the cantilever rotation path planning unit. Based on the material elevation change curve along the cantilever rotation path, the unit analyzes whether the cantilever tip will collide with the material during rotation at the θarm_adjust angle. If the analysis indicates that there are locations on the rotation path where the material height exceeds the height of the cantilever tip, the cantilever rotation angle adjustment value θarm_adjust is corrected by decreasing the rotation angle or adjusting the rotation direction until a collision-free rotation path is planned, generating the obstacle avoidance corrected cantilever rotation angle adjustment value θarm_adjust_obstacle.

[0139] Step S15108: Combine and encode the cantilever rotation angle adjustment value after obstacle avoidance correction with the cantilever rotation direction identifier to generate the basic control parameters of the cantilever rotation control command.

[0140] The cantilever rotation angle adjustment value θarm_adjust_obstacle after obstacle avoidance correction is combined with the cantilever rotation direction identifier and encoded to form the basic control parameters of the cantilever rotation control command. These basic control parameters include the required rotation angle and rotation direction information for the cantilever.

[0141] Step S15109: Obtain the current travel speed of the bucket wheel excavator's traveling mechanism, perform speed coupling analysis on the current travel speed and the cantilever slewing angle adjustment value after obstacle avoidance correction, and generate slewing angle adjustment parameters based on the matching relationship between the current travel speed and the cantilever slewing angle velocity.

[0142] Obtain the current travel speed Vcurrent of the bucket wheel excavator's traveling mechanism. Perform a velocity coupling analysis between the current travel speed Vcurrent and the obstacle avoidance corrected cantilever rotation angle adjustment value θarm_adjust_obstacle to calculate the angular velocity adjustment parameter ωarm_adjust of the cantilever rotation at the current travel speed. This angular velocity adjustment parameter ωarm_adjust is calculated using the formula ωarm_adjust=Kspeed_couple. Vcurrent θarm_adjust_obstacle is determined, where Kspeed_couple is the speed coupling coefficient. The angular velocity adjustment parameter ωarm_adjust is used to adjust the angular velocity of the cantilever rotation, so that the cantilever rotation speed matches the travel speed.

[0143] Step S15110: The slewing angular velocity adjustment parameter is superimposed on the basic control parameters of the cantilever slewing control command to generate a complete slewing angle correction command that includes the cantilever slewing angle adjustment value, the cantilever slewing direction indicator, and the slewing angular velocity adjustment parameter.

[0144] The slewing angle adjustment parameter ωarm_adjust is superimposed on the basic control parameters of the cantilever slewing control command to generate a complete slewing angle correction command. This complete slewing angle correction command consists of three parts: the cantilever slewing angle adjustment value θarm_adjust_obstacle after obstacle avoidance correction, the cantilever slewing direction indicator, and the slewing angle adjustment parameter ωarm_adjust.

[0145] Step S210: Obtain the feedback travel wheel rotation count data after the bucket wheel excavator travel mechanism executes the correction control command set, and calculate the actual travel distance of the bucket wheel excavator travel mechanism after executing the correction control command set based on the encoder pulse accumulation in the feedback travel wheel rotation count data.

[0146] After the bucket wheel excavator's traveling mechanism executes the correction control command set generated in step S150, it retrieves the accumulated pulse value Pleft_fb of the left feedback encoder after execution from the first encoder and the accumulated pulse value Pright_fb of the right feedback encoder after execution from the second encoder. The circumferences Cleft and Cright of the left and right traveling wheels are read. The actual traveling distance Dleft_fb = Pleft_fb after execution is calculated. Cleft calculates the actual walking distance after the right-side walking wheel is activated: Dright_fb = Pright_fb Cright. The average of the actual travel distances of the left and right traveling wheels, Dleft_fb and Dright_fb, is taken as the actual travel distance after the bucket wheel excavator's traveling mechanism executes the correction control command set: Actual_fb = (Dleft_fb + Dright_fb) / 2.

[0147] Step S220: Obtain the feedback attitude data of the bucket wheel excavator body collected in real time by multiple inertial measurement units installed on the bucket wheel excavator body after executing the correction control command set. The feedback attitude data of the bucket wheel excavator body includes the feedback pitch angle, feedback roll angle and feedback heading angle.

[0148] After the bucket wheel excavator's traveling mechanism completes the set of correction control commands, it again obtains the feedback attitude data of the bucket wheel excavator body from the first inertial measurement unit. This feedback attitude data of the bucket wheel excavator body includes the feedback pitch angle value θpitch_fb, the feedback roll angle value θroll_fb, and the feedback yaw angle value θyaw_fb.

[0149] Step S230: Construct a set of feedback spatial pose parameters after the bucket wheel machine executes the correction control command set based on the feedback travel wheel rotation count data and the feedback bucket wheel machine body attitude data.

[0150] The actual travel distance Dactual_fb after execution, the feedback pitch angle value θpitch_fb, the feedback roll angle value θroll_fb, and the feedback heading angle value θyaw_fb are combined to construct the feedback spatial pose parameter set Spose_fb=[Dactual_fb, θpitch_fb, θroll_fb, θyaw_fb] after the bucket wheel excavator executes the correction control command set.

[0151] Step S240: Perform a deviation comparison analysis between the feedback spatial pose parameter set and the actual spatial pose deviation vector, and calculate the pose residual vector between the feedback spatial pose parameter set and the actual spatial pose deviation vector.

[0152] The actual walking distance Didual_fb in the feedback spatial pose parameter set Spose_fb is compared with the theoretical walking distance Dtheoretical calculated in step S131, and the walking distance residual ΔDresidual = Didual_fb - Dtheoretical is calculated. The feedback pitch angle value θpitch_fb is compared with the pitch angle value θpitch extracted in step S141, and the pitch angle residual Δθpitch_residual = θpitch_fb - θpitch is calculated. The feedback roll angle value θroll_fb is compared with the roll angle value θroll extracted in step S142, and the roll angle residual Δθroll_residual = θroll_fb - θroll is calculated. The feedback yaw angle value θyaw_fb is compared with the yaw angle value θyaw extracted in step S143, and the yaw angle residual Δθyaw_residual = θyaw_fb - θyaw is calculated. The above four residual components are vectorized and combined to obtain the pose residual vector Vresidual=[ΔDresidual, Δθpitch_residual, Δθroll_residual, Δθyaw_residual].

[0153] Step S250: Generate execution effect evaluation parameters for the correction control command set based on the pose residual vector. The execution effect evaluation parameters include lateral deviation residual coefficient, longitudinal deviation residual coefficient, and heading deviation residual coefficient.

[0154] Based on the travel distance residual ΔDresidual in the pose residual vector Vresidual, and combined with the structural length Lstructure of the bucket wheel excavator's traveling mechanism, the longitudinal deviation residual coefficient Rlongitudinal_residual = ΔDresidual / Lstructure is calculated. Based on the heading angle residual Δθyaw_residual in the pose residual vector Vresidual, and combined with the wheelbase Wwheelbase of the bucket wheel excavator's traveling mechanism, the heading deviation residual coefficient Ryaw_residual = Δθyaw_residual is calculated. Wwheelbase. Based on the geometric relationship between the lateral deviation component ΔX_final and the travel distance residual ΔDresidual and heading angle residual Δθyaw_residual, the lateral deviation residual coefficient Rlateral_residual is calculated as Rlateral_residual = ΔX_final - ΔDresidual. sin(Δθyaw_residual). Combine the lateral deviation residual coefficient Rlateral_residual, longitudinal deviation residual coefficient Rlongitudinal_residual, and heading deviation residual coefficient Ryaw_residual to generate the performance evaluation parameter Eevaluation=[Rlateral_residual, Rlongitudinal_residual, Ryaw_residual].

[0155] Step S260: Input the execution effect evaluation parameters into the correction control parameter self-adjustment unit, and adaptively adjust the differential control gain of the differential control command and the slewing angle correction command in the correction control command set through the correction control parameter self-adjustment unit to generate updated differential control gain and updated slewing angle control gain.

[0156] The performance evaluation parameter, Eevaluation, is input to the self-adjusting unit of the correction control parameters. This unit contains a proportional-integral-derivative (PID) controller, which takes the lateral deviation residual coefficient, Rlateral_residual, as input and outputs the adjustment amount ΔKdiff for the differential control gain. The updated differential control gain is Kdiff_new = Kdiff + ΔKdiff. Simultaneously, the self-adjusting unit takes the heading deviation residual coefficient, Ryaw_residual, as input and outputs the adjustment amount ΔKarm for the turn angle control gain. The updated turn angle control gain is Karm_new = Karm + ΔKarm.

[0157] Step S270: Multiply the updated differential control gain with the lateral deviation component in the actual spatial pose deviation vector to generate an adaptively adjusted differential control command.

[0158] The updated differential control gain Kdiff_new is multiplied by the lateral deviation component ΔX_final in the actual spatial pose deviation vector Vpose to obtain the speed adjustment amount ΔNdiff_adaptive=Kdiff_new in the adaptively adjusted differential control command. |ΔX_final|. Based on the lateral deviation direction indicator, the speed distribution of the left and right travel wheels is determined, and an adaptively adjusted differential control command is generated. This command includes the adaptive target speed of the left travel wheel Nleft_target_adaptive and the adaptive target speed of the right travel wheel Nright_target_adaptive.

[0159] Step S280: Perform vector synthesis operation on the updated rotation angle control gain and the lateral deviation component and longitudinal deviation component in the actual spatial pose deviation vector to generate an adaptively adjusted rotation angle correction command.

[0160] The updated rotation angle control gain Karm_new is vector-combined with the lateral deviation component ΔX_final and the longitudinal deviation component ΔY_final in the actual spatial pose deviation vector Vpose to calculate the adaptively adjusted cantilever rotation angle adjustment value θarm_adjust_adaptive=Karm_new. sqrt(ΔX_final^2+ΔY_final^2). The cantilever rotation direction indicator is determined based on the symbol combination of the lateral and longitudinal deviation direction indicators, and an adaptively adjusted rotation angle correction command is generated.

[0161] Step S290: Store the adaptively adjusted differential control command and the adaptively adjusted slewing angle correction command as a reference command set for the next round of correction control of the bucket wheel excavator traveling mechanism.

[0162] The adaptively adjusted differential control command generated in step S270 and the adaptively adjusted slewing angle correction command generated in step S280 are associated and stored to form a reference command set for the next round of correction control. This reference command set is used as the benchmark for initial control commands in subsequent correction control cycles, enabling online learning and continuous optimization of control parameters.

[0163] For example, step S150 may also include: step S310: inputting the actual spatial pose deviation vector into the generator network of the pre-trained generative adversarial network, the generator network outputting a preliminary correction control command sequence, the preliminary correction control command sequence including a set of differential control value suggestions and rotation angle correction value suggestions at discrete time steps.

[0164] A generative adversarial network (GAN) is constructed, consisting of a generator network and a discriminator network. The generator network employs a multilayer perceptron architecture, comprising an input layer, three hidden layers, and an output layer. The number of nodes in the input layer equals the dimension of the actual spatial pose deviation vector, and the number of nodes in the output layer equals the product of the control time domain length and the dimension of the control command. The actual spatial pose deviation vector Vpose is input into the generator network, which performs forward propagation calculations using its internal weight matrix and bias vector, outputting a control sequence vector Csequence=[c1, c2, ..., cT], where T is the number of discrete time steps in the control time domain, and each ci contains two components: a differential control value proposal and a rotation angle correction value proposal. This control sequence vector Csequence is then reshaped into a preliminary correction control command sequence over T time steps.

[0165] Step S320: Obtain the actual correction effect dataset of the bucket wheel excavator traveling mechanism executing correction control commands under historical similar posture deviation states. The actual correction effect data includes the actual posture residual vector after correction and the energy consumption index of the correction process.

[0166] Historical correction records corresponding to pose deviation states similar to the current actual spatial pose deviation vector Vpose are retrieved from the historical operation database. Similarity is determined by calculating the Euclidean distance between the historical pose deviation vector and the current Vpose; states are considered similar when the Euclidean distance is less than a preset similarity threshold. For each historical correction record of a similar state, the actual pose residual vector Vresidual_history after executing the correction control command and the energy consumption index Eenergy_history during the correction process are extracted. The energy consumption index is calculated by multiplying the power of the integral travel motor and the rotary motor by the time. All extracted data constitutes a dataset of real correction results.

[0167] Step S330: Input the preliminary correction control command sequence and the actual spatial pose deviation vector into the discriminator network of the pre-trained generative adversarial network. The discriminator network outputs the probability value that the preliminary correction control command sequence is judged as a real and efficient control command.

[0168] The discriminator network employs a multilayer perceptron structure. The number of nodes in the input layer equals the sum of the dimensions of the actual spatial pose deviation vector and the control sequence vector. The output layer has only one node and uses a sigmoid activation function to output a probability value between 0 and 1. The actual spatial pose deviation vector Vpose is concatenated with the preliminary correction control command sequence Csequence output by the generator network to form the discriminator network's input vector Dinput = [Vpose, Csequence]. The discriminator network performs forward propagation calculations on this input vector and outputs a probability value Pauthentic, which represents the probability that the discriminator network classifies the input preliminary correction control command sequence as a genuine and efficient control command.

[0169] Step S340: Based on the probability value, the parameters of the generator network are updated in reverse by the adversarial training loss function of the generative adversarial network, so that the initial bias control command sequence generated by the generator network approaches the command sequence that the discriminator network cannot distinguish from the real efficient command, and an optimized initial bias control command sequence is generated.

[0170] Construct the adversarial training loss function for the generative adversarial network: LGAN = -log(Pauthentic), where Pauthentic is the probability value output by the discriminator network. Simultaneously, construct the real performance loss function for the generator network: Lesser = ||Vresidual_history||^2 + λ. Eenergy_history, where λ is the energy consumption weight coefficient. The total loss function of the generator network is Ltotal = LGAN + μ. The effect is defined as Ltotal, where μ is the weight coefficient of the effect loss. The gradient of Ltotal with respect to the weight matrices and bias vectors of each layer of the generator network is calculated using backpropagation. The parameters of the generator network are then updated using the Adam optimizer. After multiple iterations of training, the initial bias control command sequence output by the generator network gradually approaches a command sequence that is difficult for the discriminator network to distinguish from real, efficient commands. This trained command sequence output by the generator network is then used as the optimized initial bias control command sequence Csequence_optimized.

[0171] Step S350: Obtain a high-precision point cloud map generated by three-dimensional lidar scanning on the forward travel path of the bucket wheel excavator, perform voxelization processing on the high-precision point cloud map, and generate a three-dimensional accessibility probability grid to represent the accessibility of the terrain.

[0172] A high-precision point cloud map generated by a third-party 3D LiDAR scanner is acquired along the forward travel path of the bucket wheel excavator. This high-precision point cloud map contains dense 3D point cloud data within the forward path area. The high-precision point cloud map is then voxelized, dividing the 3D space into cubic voxel units with a side length of Lvoxel. For each voxel unit, the number of point clouds within that unit is counted, and the height variation variance of that unit is calculated. Based on the number of point clouds within the voxel unit and the height variation variance, the passability probability value Ptraversable for that voxel unit is calculated. The passability probability value Ptraversable is inversely proportional to both the number of point clouds and the height variation variance. The passability probability values ​​of all voxel units constitute a 3D passability probability grid.

[0173] Step S360: Input the three-dimensional accessibility probability grid into a pre-trained convolutional neural network and a recurrent neural network cascaded into a terrain semantic understanding network. The terrain semantic understanding network outputs terrain semantic feature encoding vectors of the path ahead of the bucket wheel machine in the future control time domain.

[0174] A terrain semantic understanding network is constructed, consisting of a concatenated convolutional neural network (CNN) and a recurrent neural network (RNN). The CNN employs a 3D convolutional neural network structure, containing three 3D convolutional layers, three 3D pooling layers, and one flattening layer, used to extract spatial features from a 3D accessibility probability grid. The RNN uses a long short-term memory (LSTM) network structure, containing two LSM layers, used to process the temporal dependencies of the spatial feature sequences output by the CNN. The 3D accessibility probability grid is input into the CNN, where each layer sequentially performs 3D convolution operations, batch normalization, activation function operations, and pooling operations, ultimately outputting a spatial feature vector. This spatial feature vector is then input into the RNN. Each LSM layer of the RNN processes the feature sequences at each time step through its internal input gate, forget gate, and output gate mechanisms, outputting a terrain semantic feature encoding vector, Vterrain, representing the path of the bucket wheel machine ahead in the next few control time domains.

[0175] Step S370: Perform feature fusion between the terrain semantic feature encoding vector and the actual spatial pose deviation vector to generate an enhanced pose context feature vector that incorporates forward-looking terrain semantic information.

[0176] The terrain semantic feature encoding vector Vterrain and the actual spatial pose deviation vector Vpose are fused by concatenating along the feature dimensions to form an enhanced pose context feature vector Vcontext=[Vpose, Vterrain]. The dimension of this enhanced pose context feature vector Vcontext is equal to the sum of the dimensions of the actual spatial pose deviation vector and the terrain semantic feature encoding vector.

[0177] Step S380: Input the enhanced pose context feature vector into the optimized generator network, and the optimized generator network outputs a sequence of correction control instructions fused with terrain semantic understanding.

[0178] The enhanced pose context feature vector Vcontext is input into the optimized generator network in step S340. The optimized generator network performs forward propagation calculation through its internal weight matrix and bias vector, and outputs a sequence of correction control instructions Csequence_terrain=[c1_terrain, c2_terrain, ..., cT_terrain] that incorporates terrain semantic understanding, where each ci_terrain contains differential control value suggestions and turning angle correction value suggestions after incorporating terrain semantic information.

[0179] Step S390: Extract the differential control command and swerve angle correction command corresponding to the current control time domain from the corrective control command sequence fused with terrain semantic understanding, and generate intelligent corrective control commands driven collaboratively by generative adversarial network and terrain semantic understanding network.

[0180] The control instruction c1_terrain corresponding to the first time step is extracted from the control instruction sequence Csequence_terrain fused with terrain semantic understanding. This control instruction contains the differential control value suggestion and the slewing angle correction value suggestion corresponding to the current control time domain. The differential control value suggestion is converted into the execution speed of the left and right travel wheels to generate a differential control instruction. The slewing angle correction value suggestion is converted into the cantilever slewing angle adjustment value and the cantilever slewing direction identifier to generate a slewing angle correction instruction. The above differential control instruction and slewing angle correction instruction together constitute the intelligent correction control instruction driven collaboratively by the generative adversarial network and the terrain semantic understanding network.

[0181] For example, the method further includes: step S410: obtaining the three-dimensional contour model of the stockpile material, the preset ideal travel trajectory elevation curve of the bucket wheel excavator, the actual spatial pose deviation vector, and the historical correction control command execution sequence, and constructing a correction control Markov decision process tuple in units of time steps.

[0182] The material elevation change curve segment at the current moment is extracted from the 3D contour model of the stockpile material; the ideal trajectory segment at the current moment is extracted from the ideal trajectory elevation curve; the deviation state at the current moment is extracted from the actual spatial pose deviation vector Vpose; and the historical correction control command execution sequence of the past N time steps is extracted from the historical database. This information is combined into a state-space representation Sstate. The control action Aaction at the current moment is defined as a two-dimensional vector containing differential speed control and slewing angle control. The deviation state change at the next moment after executing the control action is used as the state transition, and the weighted sum of the deviation reduction and energy consumption is used as the immediate reward Rreward. Thus, a correction control Markov decision process tuple (Sstate, Aaction, Rreward, Sstate_next) is constructed on a time-step basis.

[0183] Step S420: Based on the corrective control Markov decision process tuple, construct a policy network and a value network using the proximal policy optimization algorithm. The policy network is used to output the probability distribution of control actions at a given state, and the value network is used to evaluate the long-term value of the current state.

[0184] Construct a policy network using a multilayer perceptron structure. The number of nodes in the input layer equals the dimension of the state space Sstate, and the number of nodes in the output layer equals twice the dimension of the control action space Aaction. The output layer simultaneously outputs the mean and logarithmic standard deviation of the control actions, used to construct a Gaussian probability distribution. Construct a value network using a multilayer perceptron structure. The number of nodes in the input layer equals the dimension of the state space Sstate, and the number of nodes in the output layer is 1, used to output the long-term value estimate Vvalue of the current state.

[0185] Step S430: In the digital twin simulation environment of the bucket wheel excavator, drive the digital twin of the bucket wheel excavator to execute the control action sampled from the control action probability distribution output by the policy network in the current state, and calculate the next simulation state and immediate reward based on the simulation dynamics model.

[0186] A digital twin simulation environment for a bucket wheel excavator is constructed, comprising a kinematic model of the excavator, a geometric model of the stockpile materials, and a sensor simulation model. Within this environment, the current state Sstate is input into a policy network. The policy network outputs the mean and standard deviation of the control actions, which are then sampled from a Gaussian distribution to obtain the control action Aaction_sample. This Aaction_sample is then input into the bucket wheel excavator's digital twin. The simulation environment calculates the excavator's position and attitude changes based on the kinematic model, calculates the interaction with the materials based on the stockpile material geometry model, and generates the next moment's observation data based on the sensor simulation model, thus obtaining the next simulation state Sstate_next_sim. The immediate reward Rreward_sim is calculated based on the reduction in deviation and the energy consumption.

[0187] Step S440: Calculate the advantage function value of each state-action pair in the simulation trajectory using the generalized advantage estimation operator of the near-end policy optimization algorithm, which is used to evaluate the superiority or inferiority of the control action relative to the average level.

[0188] A simulation trajectory containing T steps is acquired in a simulation environment. This trajectory includes a state sequence, an action sequence, a reward sequence, and a next state sequence. The generalized advantage estimation operator is used to calculate the advantage function value Aadvantage for each state-action pair in this trajectory. The formula for calculating the generalized advantage estimation operator is Aadvantage_t = δ_t + γλδ_{t+1} + (γλ)^2δ_{t+2} + ..., where δ_t = Rreward_t + γVvalue(Sstate_{t+1}) - Vvalue(Sstate_t), γ is the discount factor, and λ is the generalized advantage estimation parameter. The advantage function value Aadvantage_t represents the superiority or inferiority of performing action Aaction_t in state Sstate_t relative to the average level.

[0189] Step S450: Based on the advantage function value, update the parameters of the policy network by maximizing the objective function with pruning mechanism, and simultaneously update the parameters of the value network by minimizing the temporal difference error of the value network.

[0190] The objective function for constructing the policy network is Lpolicy=E[min(ratio_t)]. Aadvantage_t, clip(ratio_t, 1-ε, 1+ε) Let [Aadvantage_t], where ratio_t is the probability ratio of the new policy to the old policy in state Sstate_t, and ε is the pruning threshold. The parameters of the policy network are updated using gradient ascent by maximizing this objective function. The loss function of the value network, Lvalue = E[(Vvalue(Sstate_t) - (Rreward_t + γVvalue(Sstate_{t+1})))^2], is constructed. The parameters of the value network are updated using gradient descent by minimizing this loss function.

[0191] Step S460: Load the trained strategy network into the actual control system of the bucket wheel excavator, making it the core of the online bucket wheel excavator control strategy.

[0192] The network parameters of the strategy network trained in step S450 are exported and loaded into the storage unit of the bucket wheel excavator's actual control system. This strategy network becomes the core of the online bucket wheel excavator control strategy, used to generate control actions based on real-time sensed status in the actual operating environment.

[0193] Step S470: During the actual operation of the bucket wheel excavator, the real-time sensed actual spatial pose deviation vector and other sensor status information are input into the online strategy network, and the strategy network directly outputs the probability distribution of the optimal control action at the current moment.

[0194] During the actual operation of the bucket wheel excavator, data from the first laser scanning sensor, the second laser scanning sensor, the first inertial measurement unit, the first encoder, and the second encoder are collected in real time. The actual spatial pose deviation vector Vpose_current at the current moment is calculated according to steps S110 to S140. Vpose_current is combined with other sensor state information to form the current state Sstate_current. Sstate_current is input into the online policy network. The policy network calculates the mean μaction and log-standard deviation logσaction of the control action through forward propagation, constructing a Gaussian distribution as the probability distribution of the optimal control action at the current moment.

[0195] Step S480: Sample from the optimal control action probability distribution, or select the control action with the highest probability, to generate differential control commands for the bucket wheel excavator traveling mechanism and slewing angle correction commands for the cantilever, thus forming end-to-end intelligent correction control commands directly generated by the near-end strategy optimization network.

[0196] From the optimal control action probability distribution output by the policy network, the control action with the highest probability is selected as the final control action, i.e., the mean of the Gaussian distribution, μaction, is taken as the control action value. This control action value includes a differential control component and a slewing angle correction component. The differential control component is converted into the left and right travel wheel speeds to generate a differential control command. The slewing angle correction component is converted into a cantilever slewing angle adjustment value and a cantilever slewing direction indicator to generate a slewing angle correction command. The above differential control command and slewing angle correction command together constitute the end-to-end intelligent correction control command directly generated by the proximal policy optimization network.

[0197] Step S490: Input the end-to-end intelligent correction control command directly generated by the proximal policy optimization network and the intelligent correction control command generated by the generative adversarial network into a control command fusion module based on a meta-learning framework. The control command fusion module includes a meta-learner. The meta-learner dynamically adjusts the trust weights of intelligent correction control commands from different sources according to the real-time characteristics of the current operating scenario, and generates the final fused intelligent correction control command.

[0198] A control command fusion module based on a meta-learning framework is constructed, which includes a meta-learner. The meta-learner adopts a multilayer perceptron structure, with the number of input layer nodes equal to the instantaneous feature dimension of the current running scene, and the number of output layer nodes is 2, corresponding to the trust weight w_ppo of the end-to-end intelligent correction control command generated by the proximal policy optimization network and the trust weight w_gan of the intelligent correction control command generated by the generative adversarial network, respectively, and satisfying w_ppo + w_gan = 1. The instantaneous features of the current running scene include the terrain semantic feature encoding vector Vterrain, the norm of the actual spatial pose deviation vector Vpose, and the current walking speed Vcurrent. The above instantaneous features are input into the meta-learner, and the meta-learner outputs the trust weights w_ppo and w_gan. The end-to-end intelligent correction control command Cppo generated by the proximal policy optimization network and the intelligent correction control command Cgan generated by the generative adversarial network are weighted and fused to generate the final fused intelligent correction control command Cfinal = w_ppo. Cppo+w_gan Cgan. The final intelligent correction control command, Cfinal, is the control command executed by the bucket wheel excavator's traveling mechanism, achieving adaptive fusion of intelligent control commands from different sources.

[0199] In one exemplary embodiment, a high-precision positioning and travel correction control system for bucket wheel excavators based on multi-sensor fusion is provided, which includes terminals, servers, etc., and its internal structure diagram can be as follows. Figure 2As shown, it includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements a high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the housing of the high-precision positioning and walking correction control system for bucket wheel excavators based on multi-sensor fusion, or an external keyboard, touchpad, or mouse, etc.

[0200] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion, characterized in that, The method includes: The system acquires real-time point cloud data of the stockpile material from multiple laser scanning sensors installed on the bucket wheel excavator body, real-time attitude data of the bucket wheel excavator body from multiple inertial measurement units installed on the bucket wheel excavator body, and real-time rotation data of the traveling wheels from multiple encoders installed on the bucket wheel excavator traveling mechanism. Based on the overlapping coverage relationship of the scanning areas of different laser scanning sensors in the point cloud data of the stockpile material, a three-dimensional contour model of the stockpile material is constructed with the bucket wheel excavator's travel track as the reference. The three-dimensional contour model of the stockpile material includes material elevation change curves distributed along the extension direction of the travel track. By performing spatial position correlation analysis between the material elevation change curve in the three-dimensional contour model of the stockpile material and the elevation curve of the bucket wheel excavator's preset ideal travel trajectory, the lateral deviation distance and longitudinal deviation distance of the bucket wheel excavator at the current travel position are generated. By using the pitch angle, roll angle and heading angle in the attitude data of the bucket wheel excavator body, and combining the encoder pulse accumulation in the number of rotations of the traveling wheels, the lateral deviation distance and the longitudinal deviation distance are dynamically compensated and corrected to obtain the actual spatial attitude deviation vector of the bucket wheel excavator relative to the traveling track. A set of correction control instructions for the bucket wheel excavator's traveling mechanism is generated based on the actual spatial pose deviation vector. The set of correction control instructions includes differential control instructions for adjusting the speed difference between the traveling wheels on both sides of the bucket wheel excavator's traveling mechanism and slewing angle correction instructions for adjusting the cantilever slewing angle of the bucket wheel excavator.

2. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 1, characterized in that, The step of constructing a three-dimensional contour model of the stockpile material based on the overlapping coverage relationship of the scanning areas corresponding to different laser scanning sensors in the stockpile material point cloud data includes: The plurality of laser scanning sensors are divided into a left scanning sensor group and a right scanning sensor group according to their installation positions; Extract the spatial coordinates of each point cloud from the material point cloud data collected by the left scanning sensor group. Perform coordinate transformation based on the lateral coordinate values ​​of each point cloud and the installation position coordinates of the left scanning sensor group to generate the left point cloud coordinate set of the left point cloud data corresponding to the left scanning sensor group in a unified spatial coordinate system. Extract the spatial coordinates of each point cloud in the material point cloud data collected by the right scanning sensor group, and perform coordinate transformation based on the lateral coordinate value in the spatial coordinates of each point cloud and the installation position coordinates of the right scanning sensor group to generate the right point cloud coordinate set of the right point cloud data corresponding to the right scanning sensor group in a unified spatial coordinate system. Spatial coordinate fusion is performed on the left point cloud coordinate set and the right point cloud coordinate set to identify the point cloud coordinate points whose spatial coordinates coincide with those of the left and right point cloud coordinate sets as the overlapping region point cloud coordinate set. Based on the spatial coordinates of each point cloud coordinate point in the overlapping area point cloud coordinate set, calculate the coordinate deviation value between the left and right point cloud coordinate sets in the overlapping area, and use the coordinate deviation value to perform global coordinate consistency calibration on the left and right point cloud coordinate sets to obtain the calibrated left and right point cloud coordinate sets. The calibrated left and right point cloud coordinate sets are divided into spatial grids according to the extension direction of the walking track. The yard area is divided into multiple continuous grid units along the extension direction of the walking track, and each grid unit corresponds to a segment of the walking track. Within each grid cell, extract the vertical coordinate values ​​of all point cloud coordinate points in the left and right point cloud coordinate subsets corresponding to that grid cell, and perform statistical processing on the vertical coordinate values ​​to obtain the material representative elevation value corresponding to each grid cell. The location of the walking track interval corresponding to each grid cell is associated with the material representative elevation value corresponding to each grid cell, and a material elevation change curve is generated with the location of the walking track interval as the horizontal axis and the material representative elevation value as the vertical axis. The material elevation change curve is superimposed with the spatial geometric parameters of the bucket wheel excavator's travel track to establish a three-dimensional contour model of the stockpile material based on the bucket wheel excavator's travel track. The three-dimensional contour model of the stockpile material includes the spatial three-dimensional coordinates corresponding to each coordinate point on the material elevation change curve.

3. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 1, characterized in that, The spatial position correlation analysis is performed by comparing the material elevation change curve in the three-dimensional contour model of the stockpile material with the preset ideal travel trajectory elevation curve of the bucket wheel excavator, generating the lateral and longitudinal deviation distances of the bucket wheel excavator at its current travel position, including: The number of rotations of the traveling wheel at the current moment is obtained from the encoder installed on the traveling mechanism of the bucket wheel excavator. Based on the product of the cumulative encoder pulse in the number of rotations of the traveling wheel and the circumference of the traveling wheel, the theoretical traveling distance of the traveling mechanism of the bucket wheel excavator along the traveling track is calculated. Based on the theoretical walking distance, locate the corresponding first elevation curve point on the material elevation change curve of the three-dimensional contour model of the stockpile material, and extract the spatial three-dimensional coordinates corresponding to the first elevation curve point as the measured material spatial coordinates corresponding to the current walking position. Based on the theoretical walking distance, locate the corresponding second elevation curve point on the preset ideal walking trajectory elevation curve of the bucket wheel excavator, and extract the spatial three-dimensional coordinates corresponding to the second elevation curve point as the spatial coordinates of the ideal walking trajectory corresponding to the current walking position; The measured material spatial coordinates and the ideal walking trajectory spatial coordinates are input into the spatial coordinate difference calculation unit. The difference between the measured material spatial coordinates and the ideal walking trajectory spatial coordinates in the horizontal coordinate axis direction is calculated as the horizontal deviation distance, and the difference between the vertical coordinates in the vertical coordinate axis direction is calculated as the vertical deviation distance. The lateral deviation distance and the longitudinal deviation distance are marked with symbols. The lateral deviation direction indicator of the lateral deviation distance and the longitudinal deviation direction indicator of the longitudinal deviation distance are determined according to the spatial position relationship between the measured material spatial coordinates and the ideal travel trajectory spatial coordinates. The lateral deviation direction indicator is used to indicate the lateral offset direction of the bucket wheel excavator relative to the ideal travel trajectory, and the longitudinal deviation direction indicator is used to indicate the longitudinal offset direction of the bucket wheel excavator relative to the ideal travel trajectory. The lateral deviation distance is proportionally correlated with the structural width of the bucket wheel excavator's traveling mechanism to generate a ratio of lateral deviation distance to structural width as a lateral relative deviation coefficient. The longitudinal deviation distance is proportionally correlated with the structural length of the bucket wheel excavator's traveling mechanism to generate a ratio of longitudinal deviation distance to structural length as a longitudinal relative deviation coefficient.

4. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 1, characterized in that, The method utilizes the pitch angle, roll angle, and yaw angle from the bucket wheel excavator's body attitude data, combined with the encoder pulse accumulation from the travel wheel rotation count data, to dynamically compensate and correct the lateral deviation distance and the longitudinal deviation distance, thereby obtaining the actual spatial attitude deviation vector of the bucket wheel excavator relative to the travel track, including: Extract the pitch angle of the bucket wheel machine body at the current moment from the attitude data of the bucket wheel machine body, and calculate the horizontal projection offset of the front end point of the bucket wheel machine on the horizontal plane caused by the pitch motion of the bucket wheel machine cantilever based on the product of the pitch angle and the cantilever length of the bucket wheel machine. Extract the roll angle of the bucket wheel excavator body at the current moment from the attitude data of the bucket wheel excavator body, and calculate the vertical displacement difference in the vertical direction of the two traveling wheels of the bucket wheel excavator caused by the roll motion of the bucket wheel excavator body based on the product of the roll angle value and the width of the bucket wheel excavator body. Extract the heading angle of the bucket wheel excavator body at the current moment from the attitude data of the bucket wheel excavator body, and calculate the change in trajectory curvature of the bucket wheel excavator's travel trajectory caused by the heading deflection motion of the bucket wheel excavator based on the ratio of the heading angle value to the wheel gauge of the bucket wheel excavator's traveling mechanism. Extract the encoder pulse accumulation of the left walking wheel and the encoder pulse accumulation of the right walking wheel from the data of the number of rotations of the walking wheel. Calculate the actual walking distance of the left walking wheel based on the product of the encoder pulse accumulation of the left walking wheel and the circumference of the left walking wheel. Calculate the actual walking distance of the right walking wheel based on the product of the encoder pulse accumulation of the right walking wheel and the circumference of the right walking wheel. The difference between the actual travel distance of the left and right traveling wheels is calculated as the travel difference between the left and right traveling wheels. Based on the ratio of the travel difference between the left and right traveling wheels to the distance between the traveling wheels of the bucket wheel excavator, the change in the heading deflection angle of the bucket wheel excavator body caused by the asynchronous movement of the left and right traveling wheels is calculated. The horizontal projection offset, the vertical displacement difference, the trajectory curvature change, and the heading deflection angle change are input into the attitude compensation matrix operation unit to generate a lateral compensation factor for correcting the lateral deviation distance and a longitudinal compensation factor for correcting the longitudinal deviation distance. The lateral compensation factor is multiplied by the lateral deviation distance to obtain the corrected lateral deviation distance, and the longitudinal compensation factor is multiplied by the longitudinal deviation distance to obtain the corrected longitudinal deviation distance. Obtain the historical posture deviation vector sequence accumulated during the historical travel of the bucket wheel excavator, and extract the historical lateral deviation distance and historical longitudinal deviation distance corresponding to multiple consecutive historical moments adjacent to the current moment from the historical posture deviation vector sequence. A horizontal deviation time series prediction model is constructed based on the changing trend of the historical horizontal deviation distance corresponding to multiple consecutive historical moments, and a vertical deviation time series prediction model is constructed based on the changing trend of the historical vertical deviation distance corresponding to multiple consecutive historical moments. The corrected lateral deviation distance is input into the lateral deviation time series prediction model to predict the lateral deviation at future times, and the predicted lateral deviation distance is obtained. The corrected longitudinal deviation distance is input into the longitudinal deviation time series prediction model to predict the longitudinal deviation at future times, and the predicted longitudinal deviation distance is obtained. The corrected lateral deviation distance, the corrected longitudinal deviation distance, the predicted lateral deviation distance, the predicted longitudinal deviation distance, and the travel difference between the left and right traveling wheels are vectorized and combined to generate the actual spatial posture deviation vector of the bucket wheel excavator relative to the traveling track. The actual spatial posture deviation vector includes lateral deviation components, longitudinal deviation components, predicted lateral deviation components, predicted longitudinal deviation components, and traveling wheel synchronization deviation components.

5. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 1, characterized in that, The set of correction control instructions for the bucket wheel excavator traveling mechanism generated based on the actual spatial pose deviation vector includes: The lateral deviation component in the actual spatial pose deviation vector is analyzed. Based on the value of the lateral deviation component and the lateral deviation direction identifier, a differential control command is generated to drive the left and right traveling wheels of the bucket wheel excavator traveling mechanism to generate a speed difference. The differential control command includes the target speed of the left traveling wheel and the target speed of the right traveling wheel. The lateral deviation prediction component in the actual spatial pose deviation vector is analyzed. Based on the value of the lateral deviation prediction component and the lateral deviation prediction direction identifier, a pre-steering control command for adjusting the travel direction of the bucket wheel excavator travel mechanism in advance is generated. The pre-steering control command includes the pre-adjustment speed of the left travel wheel and the pre-adjustment speed of the right travel wheel. The target speed of the left travel wheel in the differential control command is superimposed with the pre-adjusted speed of the left travel wheel in the pre-steering control command to generate the final speed control parameter of the left travel wheel. The target speed of the right travel wheel in the differential control command is superimposed with the pre-adjusted speed of the right travel wheel in the pre-steering control command to generate the final speed control parameter of the right travel wheel. The longitudinal deviation component in the actual spatial pose deviation vector is analyzed, and a speed control command for adjusting the traveling speed of the bucket wheel excavator's traveling mechanism is generated based on the value of the longitudinal deviation component and the longitudinal deviation direction identifier. The speed control command includes the target speed of the traveling mechanism. The longitudinal deviation prediction component in the actual spatial pose deviation vector is analyzed. Based on the value of the longitudinal deviation prediction component and the longitudinal deviation prediction direction identifier, an acceleration pre-adjustment command for adjusting the travel acceleration of the bucket wheel excavator travel mechanism in advance is generated. The acceleration pre-adjustment command includes the target acceleration of the travel mechanism. The target speed of the walking mechanism in the speed control command and the target acceleration of the walking mechanism in the acceleration pre-adjustment command are integrally correlated to generate control parameters for the walking mechanism speed change curve. The synchronous deviation component of the traveling wheel in the actual spatial pose deviation vector is analyzed, and a synchronous balance control command for balancing the speed of the traveling wheels on both sides of the bucket wheel excavator is generated based on the value of the synchronous deviation component of the traveling wheel. The synchronous balance control command includes compensation for the speed of the left traveling wheel and compensation for the speed of the right traveling wheel. The final speed control parameter of the left walking wheel is fused with the speed compensation of the left walking wheel in the synchronous balance control command to generate the execution speed of the left walking wheel; the final speed control parameter of the right walking wheel is fused with the speed compensation of the right walking wheel in the synchronous balance control command to generate the execution speed of the right walking wheel. The differential drive execution signal of the bucket wheel excavator's traveling mechanism is generated based on the operating speed of the left traveling wheel and the operating speed of the right traveling wheel, and the speed drive execution signal of the bucket wheel excavator's traveling mechanism is generated based on the control parameters of the traveling mechanism's speed change curve. The lateral deviation component and the longitudinal deviation component in the actual spatial pose deviation vector are analyzed. Based on the spatial vector synthesis relationship between the lateral deviation component and the longitudinal deviation component, a slewing angle correction command for adjusting the cantilever slewing angle of the bucket wheel excavator is generated. The slewing angle correction command includes the cantilever slewing angle adjustment value and the cantilever slewing direction identifier.

6. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 2, characterized in that, The step of superimposing the material elevation change curve with the spatial geometric parameters of the bucket wheel excavator's travel track to establish a three-dimensional contour model of the stockpile material based on the bucket wheel excavator's travel track includes: Obtain the spatial geometric parameter set of the bucket wheel excavator's travel track. The spatial geometric parameter set includes the track centerline coordinate sequence, track width, track height, and track extension direction vector in a unified spatial coordinate system. Map the position of the travel track interval corresponding to each curve point on the material elevation change curve to the coordinate sequence of the track centerline. Calculate the track centerline coordinate value corresponding to each curve point based on the linear interpolation relationship between the travel track interval position and the track centerline coordinate sequence. Using the coordinates of the track centerline corresponding to each curve point as a reference point, and combining the track width and the track height, a geometric model of the track cross section corresponding to each curve point is constructed. The geometric model of the track cross section includes the spatial coordinates of each key point on the track cross section. The material representative elevation value corresponding to each curve point is used as the vertical elevation offset of the track cross-section geometric model corresponding to that curve point. The elevation offset is superimposed on the spatial coordinates of each key point in the track cross-section geometric model to generate the material cross-section contour point cloud coordinate set corresponding to each curve point. The coordinate sets of the material cross-section contour point cloud corresponding to adjacent curve points are continuously connected. The coordinates of the point cloud corresponding to the material cross-section contour point cloud of adjacent curve points are spatially fitted according to the extension direction of the walking track to generate a continuous surface mesh model of the material surface in three-dimensional space. The spatial three-dimensional coordinates of each grid node are extracted from the continuous curved surface grid model. The spatial three-dimensional coordinates of each grid node are associated with the corresponding walking track interval position and stored to form a material space coordinate mapping table indexed by the walking track interval position. Based on the material space three-dimensional coordinates corresponding to the positions of each travel track interval in the material space coordinate mapping table, construct the spatial three-dimensional coordinates corresponding to each coordinate point on the material elevation change curve. The spatial three-dimensional coordinates include the position coordinates along the extension direction of the travel track, the lateral position coordinates perpendicular to the extension direction of the travel track, and the elevation coordinates in the vertical direction. The correspondence between the location of the travel track interval corresponding to each coordinate point on the material elevation change curve and the three-dimensional spatial coordinates is solidified to generate a three-dimensional outline model of the stockpile material based on the bucket wheel excavator's travel track.

7. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 3, characterized in that, After proportionally correlating the lateral deviation distance with the structural width of the bucket wheel excavator's traveling mechanism to generate a ratio of lateral deviation distance to structural width as a lateral relative deviation coefficient, and proportionally correlating the longitudinal deviation distance with the structural length of the bucket wheel excavator's traveling mechanism to generate a ratio of longitudinal deviation distance to structural length as a longitudinal relative deviation coefficient, the method further includes: Obtain the sequence of lateral relative deviation coefficients and the sequence of longitudinal relative deviation coefficients recorded by the bucket wheel excavator's traveling mechanism during its historical travel process. Extract the historical lateral relative deviation coefficients corresponding to multiple consecutive historical moments from the sequence of lateral relative deviation coefficients, and extract the historical longitudinal relative deviation coefficients corresponding to multiple consecutive historical moments from the sequence of longitudinal relative deviation coefficients. A horizontal deviation trend characteristic curve is constructed based on the historical horizontal relative deviation coefficients corresponding to the consecutive historical moments, and the slope of the horizontal deviation trend characteristic curve on the time axis is calculated as the rate of change of the horizontal deviation. Based on the historical longitudinal relative deviation coefficients corresponding to the consecutive historical moments, a longitudinal deviation trend characteristic curve is constructed, and the slope of the longitudinal deviation trend characteristic curve on the time axis is calculated as the longitudinal deviation change rate. The correlation analysis between the current lateral relative deviation coefficient and the lateral deviation change rate is performed, and the lateral deviation cumulative effect parameter is generated based on the product of the lateral relative deviation coefficient and the lateral deviation change rate. The longitudinal relative deviation coefficient at the current moment is correlated with the longitudinal deviation change rate, and the longitudinal deviation cumulative effect parameter is generated based on the product of the longitudinal relative deviation coefficient and the longitudinal deviation change rate. The travel speed of the bucket wheel excavator's traveling mechanism is obtained, and the lateral deviation cumulative effect parameter is correlated with the travel speed to generate a lateral deviation speed coupling factor. The longitudinal deviation cumulative effect parameter is correlated with the travel speed to generate a longitudinal deviation speed coupling factor. The lateral relative deviation coefficient, the lateral deviation change rate, the lateral deviation cumulative effect parameter, and the lateral deviation velocity coupling factor are vectorized and combined to generate a comprehensive lateral deviation feature vector. The longitudinal relative deviation coefficient, the longitudinal deviation change rate, the longitudinal deviation cumulative effect parameter, and the longitudinal deviation velocity coupling factor are vectorized and combined to generate a comprehensive longitudinal deviation feature vector. The lateral deviation comprehensive feature vector and the longitudinal deviation comprehensive feature vector are input into the pose deviation prediction model. The feature fusion layer of the pose deviation prediction model performs cross-correlation analysis on the lateral deviation comprehensive feature vector and the longitudinal deviation comprehensive feature vector to generate the coupling influence coefficient between the lateral deviation and the longitudinal deviation. The lateral deviation distance and the longitudinal deviation distance are corrected twice based on the coupling influence coefficient to obtain the lateral deviation distance and the longitudinal deviation distance after the second correction.

8. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 4, characterized in that, After vectorizing and combining the corrected lateral deviation distance, the corrected longitudinal deviation distance, the predicted lateral deviation distance, the predicted longitudinal deviation distance, and the travel difference between the left and right traveling wheels to generate the actual spatial pose deviation vector of the bucket wheel excavator relative to the traveling track, the method further includes: The point cloud data of the material in front of the bucket wheel excavator is collected in real time by multiple laser scanning sensors installed on the cantilever, and the elevation change curve of the material in front of the cantilever is constructed based on the point cloud data of the material in front of the cantilever. The material elevation change curve in front of the cantilever is matched with the material elevation change curve in the three-dimensional contour model of the stockpile material, and the curve offset between the material elevation change curve in front of the cantilever and the material elevation change curve in spatial position is calculated. Based on the curve offset, generate the elevation deviation of the material in front of the cantilever relative to the three-dimensional contour model of the stockpile material. A spatial geometric correlation analysis is performed between the elevation deviation in front of the cantilever and the lateral and longitudinal deviation components in the actual spatial pose deviation vector. Based on the geometric projection relationship between the elevation deviation in front of the cantilever and the lateral deviation component, a lateral deviation influence factor in front of the cantilever is generated. Based on the geometric projection relationship between the elevation deviation in front of the cantilever and the longitudinal deviation component, a longitudinal deviation influence factor in front of the cantilever is generated. The lateral deviation influence factor in front of the cantilever is superimposed on the lateral deviation component of the actual spatial pose deviation vector to generate an updated lateral deviation component. The longitudinal deviation influence factor in front of the cantilever is superimposed on the longitudinal deviation component of the actual spatial pose deviation vector to generate an updated longitudinal deviation component. A spatial geometric correlation analysis is performed between the elevation deviation in front of the cantilever and the predicted lateral deviation distance and the predicted longitudinal deviation distance in the actual spatial pose deviation vector. Based on the geometric projection relationship between the elevation deviation in front of the cantilever and the predicted lateral deviation distance, a correction factor for the lateral prediction deviation in front of the cantilever is generated. Based on the geometric projection relationship between the elevation deviation in front of the cantilever and the predicted longitudinal deviation distance, a correction factor for the longitudinal prediction deviation in front of the cantilever is generated. The cantilever front lateral prediction deviation correction factor is superimposed on the lateral deviation prediction component of the actual spatial pose deviation vector to generate an updated lateral deviation prediction component. The cantilever front longitudinal prediction deviation correction factor is superimposed on the longitudinal deviation prediction component of the actual spatial pose deviation vector to generate an updated longitudinal deviation prediction component. The tilt angle and slewing angle of the boom are collected in real time by the tilt sensor installed on the boom of the bucket wheel excavator. The spatial coordinates of the front end point of the boom in space are calculated based on the tilt angle and slewing angle. The spatial coordinates of the cantilever front end point are analyzed in relation to the updated lateral deviation component and the updated longitudinal deviation component. The lateral position deviation factor of the cantilever front end is generated based on the spatial distance relationship between the spatial coordinates of the cantilever front end point and the updated lateral deviation component. The longitudinal position deviation factor of the cantilever front end is generated based on the spatial distance relationship between the spatial coordinates of the cantilever front end point and the updated longitudinal deviation component. The lateral position deviation factor of the cantilever front end is superimposed on the updated lateral deviation component to generate the final lateral deviation component, and the longitudinal position deviation factor of the cantilever front end is superimposed on the updated longitudinal deviation component to generate the final longitudinal deviation component. The final lateral deviation component, the final longitudinal deviation component, the updated lateral deviation prediction component, the updated longitudinal deviation prediction component, and the travel difference between the left and right wheels are vectorized and recombined to generate an updated actual spatial pose deviation vector.

9. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 5, characterized in that, The step involves analyzing the lateral and longitudinal deviation components in the actual spatial pose deviation vector, and generating a slewing angle correction command for adjusting the cantilever slewing angle of the bucket wheel excavator based on the spatial vector composition relationship between the lateral and longitudinal deviation components. This includes: The values ​​of the lateral deviation components and the lateral deviation direction identifiers are extracted from the actual spatial pose deviation vector; the values ​​of the longitudinal deviation components and the longitudinal deviation direction identifiers are extracted from the actual spatial pose deviation vector. A spatial vector model of the current posture deviation of the bucket wheel excavator is constructed based on the values ​​of the lateral deviation component and the longitudinal deviation component. The spatial vector model takes the center point of the bucket wheel excavator's traveling mechanism as the vector starting point and the vector direction of the vector synthesis of the lateral deviation component and the longitudinal deviation component as the vector direction. The angle between the vector direction of the spatial vector model and the current rotation direction of the bucket wheel excavator boom is calculated as the boom rotation angle deviation value; The cantilever rotation direction indicator is determined based on the symbol combination relationship between the lateral deviation direction indicator and the longitudinal deviation direction indicator. The cantilever rotation direction indicator is used to indicate whether the cantilever rotation angle adjustment direction is clockwise or counterclockwise. The cantilever rotation angle deviation value is proportionally correlated with the rotation angle range of the bucket wheel excavator cantilever to generate a cantilever rotation angle adjustment value, which is used to indicate the angle value that the cantilever needs to rotate. The material point cloud data on the cantilever rotation path is collected in real time by multiple laser scanning sensors installed on the cantilever of the bucket wheel excavator. The material elevation change curve on the cantilever rotation path is constructed based on the material point cloud data on the cantilever rotation path. The cantilever rotation angle adjustment value is input into the cantilever rotation path planning unit. The cantilever rotation angle adjustment value is corrected for obstacle avoidance based on the material elevation change curve on the cantilever rotation path, and the cantilever rotation angle adjustment value after obstacle avoidance correction is generated. The cantilever rotation angle adjustment value after obstacle avoidance correction is combined and encoded with the cantilever rotation direction identifier to generate the basic control parameters of the cantilever rotation control command. The current travel speed of the bucket wheel excavator's traveling mechanism is obtained, and the current travel speed is coupled with the cantilever rotation angle adjustment value after obstacle avoidance correction is analyzed. Based on the matching relationship between the current travel speed and the cantilever rotation angle, rotation angle adjustment parameters are generated. The slewing angular velocity adjustment parameters are superimposed on the basic control parameters of the cantilever slewing control command to generate a complete slewing angle correction command that includes the cantilever slewing angle adjustment value, the cantilever slewing direction indicator, and the slewing angular velocity adjustment parameters.

10. The high-precision positioning and travel correction control method for bucket wheel excavators based on multi-sensor fusion according to claim 1, characterized in that, After generating the correction control command set for the bucket wheel excavator's traveling mechanism based on the actual spatial pose deviation vector, the method further includes: Obtain the feedback travel wheel rotation count data after the bucket wheel excavator travel mechanism executes the correction control command set, and calculate the actual travel distance of the bucket wheel excavator travel mechanism after executing the correction control command set based on the encoder pulse accumulation in the feedback travel wheel rotation count data; The feedback attitude data of the bucket wheel excavator body is collected in real time by multiple inertial measurement units installed on the bucket wheel excavator body after executing the correction control command set. The feedback attitude data of the bucket wheel excavator body includes feedback pitch angle, feedback roll angle and feedback heading angle. Based on the feedback walking wheel rotation count data and the feedback bucket wheel excavator body attitude data, a feedback spatial pose parameter set is constructed after the bucket wheel excavator executes the correction control command set; The feedback spatial pose parameter set and the actual spatial pose deviation vector are compared and analyzed to calculate the pose residual vector between the feedback spatial pose parameter set and the actual spatial pose deviation vector. The execution effect evaluation parameters of the correction control command set are generated based on the pose residual vector. The execution effect evaluation parameters include the lateral deviation residual coefficient, the longitudinal deviation residual coefficient, and the heading deviation residual coefficient. The execution effect evaluation parameters are input into the correction control parameter self-adjustment unit. The correction control parameter self-adjustment unit adaptively adjusts the differential control gain of the differential control command and the slewing angle correction command in the correction control command set to generate updated differential control gain and updated slewing angle control gain. The updated differential control gain is multiplied by the lateral deviation component in the actual spatial pose deviation vector to generate an adaptively adjusted differential control command. The updated rotation angle control gain is vector-combined with the lateral and longitudinal deviation components in the actual spatial pose deviation vector to generate an adaptively adjusted rotation angle correction command. The adaptively adjusted differential control command and the adaptively adjusted slewing angle correction command are stored as a reference command set for the next round of correction control of the bucket wheel excavator's traveling mechanism.

Images