A counterforce elimination intelligent control method for a loader

Abstract

This invention discloses an intelligent control method for loader with reaction force elimination, belonging to the technical field of intelligent loader control, including: Step S1: Constructing a loader operation status perception system; Step S2: Constructing a reaction force prediction model based on collected real-time parameters; Step S3: Analyzing the instantaneous reaction force data output by the reaction force prediction model; Step S4: Executing a reaction force elimination control strategy through an intelligent control unit; Step S5: Real-time acquisition of feedback parameters after the loader executes control commands; Step S6: Constructing a fault diagnosis and fault-tolerant control module. This invention, by constructing a multi-dimensional operation status perception system, a precise reaction force prediction model, and a collaborative elimination control strategy, achieves real-time perception, accurate prediction, and efficient cancellation of material reaction forces, improving the loader's operation accuracy, efficiency, and safety, and extending equipment lifespan.

Description

Technical Field

[0001] This invention relates to the technical field of intelligent control of loaders, and in particular to an intelligent control method for loader that eliminates reaction force. Background Technology

[0002] Loaders, as core equipment in the engineering construction field, are widely used in mining, road construction, building construction, port loading and unloading, and other scenarios, mainly undertaking key operations such as material loading, transfer, and unloading. During actual operation, especially when loading hard or heavy materials, the material generates a significant instantaneous reaction force on the bucket. This reaction force is transmitted to the loader body through the bucket, boom, and other structures, easily causing overall machine vibration and attitude deviation. This not only reduces operational accuracy and efficiency but also accelerates the wear of critical components such as the hydraulic system and chassis structure, shortening the equipment's service life. In severe cases, it can even lead to the entire machine overturning, causing a safety accident.

[0003] Existing loader control technologies mostly employ traditional hydraulic proportional control or simple closed-loop control. Their core objective is merely to drive basic actions such as boom lifting and bucket tilting, without designing specific strategies to eliminate material reaction forces. This results in significant technical deficiencies: First, they lack precise operational status perception capabilities, failing to acquire the magnitude and direction of material reaction forces in real time, leading to insufficient targeting of control strategies. Second, the control logic is simplistic, focusing primarily on the independent control of individual actuators, failing to achieve coordinated linkage between the boom, bucket, and overall machine movement, making it difficult to comprehensively counteract the impact of reaction forces. Third, control parameters are mostly fixed settings, unable to be dynamically adjusted according to changes in the operational scenario, resulting in poor control stability and adaptability.

[0004] With the increasing demands for operational efficiency, safety, and equipment durability in engineering construction, traditional control methods can no longer meet practical needs. Therefore, developing an intelligent control method that can accurately sense material reaction forces and eliminate them through collaborative control has become a key direction for the technological upgrading of loaders. Summary of the Invention

[0005] To address the shortcomings of existing loader control technologies, such as their inability to effectively handle material reaction forces and poor operational stability, this invention provides an intelligent control method for loader systems that eliminates reaction forces. By constructing a multi-dimensional operational status perception system, a precise reaction force prediction model, and a collaborative elimination control strategy, real-time perception, accurate prediction, and efficient cancellation of material reaction forces are achieved, thereby improving the loader's operational accuracy, efficiency, and safety, and extending the equipment's service life.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] A method for intelligent control of a loader that eliminates reaction force includes the following steps:

[0008] Step S1: Construct a loader operation status perception system. Collect real-time operating parameters and environmental parameters during the loader operation process through multi-sensor fusion. The real-time operating parameters include boom lifting angle, bucket tilting angle, boom cylinder pressure, bucket cylinder pressure, loader travel speed, and engine speed. The environmental parameters include material type on the working surface, material accumulation height, and working surface slope.

[0009] Step S2: Construct a reaction force prediction model based on the collected real-time parameters. Calculate the magnitude and direction of the instantaneous reaction force generated by the material on the bucket during the loader's loading, lifting, and unloading processes using the reaction force prediction model. The reaction force prediction model adopts an improved BP neural network model, with its input layer being the real-time operating parameters and environmental parameters collected in step S1, and its output layer being the magnitude and direction vector of the instantaneous reaction force.

[0010] Step S3: Based on the instantaneous reaction force data output by the reaction force prediction model, and combined with the dynamic model of the loader, a reaction force elimination control strategy is constructed. The reaction force elimination control strategy includes boom cylinder pressure compensation control, bucket cylinder pressure compensation control, and overall machine attitude coordinated adjustment control.

[0011] Step S4: The intelligent control unit executes the reaction force elimination control strategy, outputs control commands to the hydraulic execution system and travel system of the loader in real time, dynamically adjusts the oil supply pressure and flow of the boom cylinder and bucket cylinder, and simultaneously adjusts the loader's travel speed and steering angle to offset the reaction force of the material.

[0012] Step S5: Collect feedback parameters after the loader executes control commands in real time, compare the feedback parameters with the preset target parameters, calculate the control error, and dynamically correct the control parameters of the reaction force elimination control strategy through the PID algorithm to ensure the stability of the reaction force elimination effect.

[0013] As a preferred embodiment of the intelligent control method for loader with reaction force elimination described in this invention, the multi-sensor fusion in step S1 specifically includes: using an angle sensor to collect the boom lifting angle and bucket tilting angle, with a sampling frequency of 100Hz; using a pressure sensor to collect the pressure in the rodless and rod chambers of the boom cylinder and the rodless and rod chambers of the bucket cylinder, with a measurement accuracy of ±0.5MPa; using a speed sensor to collect the loader's travel speed, with a measurement range of 0-30km / h; using a speed sensor to collect the engine speed, with a measurement range of 800-2500r / min; using a vision sensor and a lidar sensor to collect the material type, material accumulation height, and slope of the working surface; and using image recognition algorithms and point cloud processing algorithms to fuse and analyze the collected data, with a material type recognition accuracy of not less than 95% and a slope measurement accuracy of ±0.5°.

[0014] As a preferred embodiment of the intelligent control method for loader with reaction force elimination described in this invention, the construction process of the improved BP neural network model in step S2 includes: Step S21: Collecting historical operating data and corresponding measured reaction force data of the loader under different operating scenarios, constructing training datasets and test datasets. The operating scenarios include sand loading, gravel loading, coal loading, and unloading operations. The ratio of the training dataset to the test dataset is 7:3; Step S22: Normalizing the training dataset, mapping the input parameters and output parameters to the [0,1] interval, and using the min-max normalization algorithm; Step S23: Setting the structure of the BP neural network, with 12 nodes in the input layer corresponding to 12 input parameters, and implicit... The hidden layer is set to two layers: the first hidden layer has 20 nodes, the second hidden layer has 15 nodes, and the output layer has 3 nodes, corresponding to the x-axis, y-axis, and z-axis components of the reaction force, respectively. Step S24: The momentum gradient descent algorithm is used to optimize the weights and thresholds of the BP neural network. A momentum factor α is introduced, with a value ranging from 0.1 to 0.3. The learning rate η ranges from 0.01 to 0.05. The network is iteratively trained using the training dataset. Training is stopped when the number of iterations reaches 1000 or the network error is less than 0.001. Step S25: The trained BP neural network is validated using the test dataset. If the model prediction error is greater than 5%, the network structure and training parameters are readjusted until the prediction accuracy requirements are met.

[0015] As a preferred embodiment of the intelligent control method for eliminating reaction force in a loader as described in this invention, the dynamic model construction process of the loader in step S3 includes: establishing a rectangular coordinate system with the center of mass of the loader as the origin, the x-axis along the forward direction of the loader, the y-axis along the horizontal direction of the loader, and the z-axis perpendicular to the ground and upward; based on Newton's laws of motion and the principle of rigid body dynamics, establishing the translational motion equation and rotational motion equation of the loader, considering the inertial force, gravity, ground support force, driving force of the hydraulic cylinder, and reaction force of the material of the boom, bucket, and material;

[0016] Equations of translational motion:

[0017] ;

[0018] ;

[0019] ;

[0020] Equations of rotational motion:

[0021] ;

[0022] ;

[0023] ;

[0024] Where m is the mass of the loader, ax, ay, and az are the accelerations of the center of mass in the x, y, and z axes, respectively, Fx, Fy, and Fz are the components of the material reaction force in the x, y, and z axes, respectively, Ffx and Ffy are the frictional forces in the x and y axes, respectively, N is the ground support force, g is the acceleration due to gravity, Jx, Jy, and Jz are the moments of inertia of the loader about the x, y, and z axes, respectively, βx, βy, and βz are the angular accelerations about the x, y, and z axes, respectively, and Mx, My, and Mz are the torques generated by the material reaction force about the x, y, and z axes, respectively.

[0025] As a preferred embodiment of the intelligent control method for eliminating reaction force in a loader as described in this invention, the boom cylinder pressure compensation control in step S3 specifically involves: calculating the required compensation pressure for the boom cylinder based on the z-axis component of the reaction force output by the reaction force prediction model, where Pcomp is the compensation pressure, S is the effective area of ​​the rodless chamber of the boom cylinder, and ηh is the hydraulic system efficiency; the intelligent control unit adjusts the oil supply pressure of the boom cylinder through a proportional relief valve according to the calculated compensation pressure, so that the total driving force of the boom cylinder is equal to the sum of the z-axis component of the material reaction force and the z-axis component of the weight of the boom and bucket.

[0026] As a preferred embodiment of the intelligent control method for loader to eliminate reaction force described in this invention, the bucket cylinder pressure compensation control in step S3 specifically involves: calculating the required compensation pressure of the bucket cylinder based on the x-axis and y-axis components of the reaction force output by the reaction force prediction model; obtaining the mapping relationship between the extension / retraction of the bucket cylinder and the reaction force components through inverse kinematics solution; and determining the compensation pressure difference between the rodless chamber and the rod chamber of the bucket cylinder; the intelligent control unit adjusts the inlet and outlet flow rates of the bucket cylinder through a proportional directional valve to achieve dynamic compensation of the bucket cylinder pressure, thereby offsetting the influence of the material reaction force on the bucket's tilting posture.

[0027] As a preferred embodiment of the intelligent control method for eliminating reaction force in a loader according to the present invention, the overall machine attitude coordinated adjustment control in step S3 specifically involves: based on the dynamic model of the loader, and combined with the magnitude and direction of the reaction force, calculating the required adjustment attitude parameters of the loader, including the driving speed correction and steering angle correction; when the component of the reaction force along the x-axis is greater than a preset threshold, the intelligent control unit controls the traveling system to reduce the driving speed, and the reduction is proportional to the x-axis component of the reaction force; when the component of the reaction force along the y-axis is greater than a preset threshold, the intelligent control unit controls the steering system to adjust the steering angle so that the driving direction of the loader is consistent with the direction of the y-axis component of the reaction force, thereby reducing the risk of lateral overturning.

[0028] As a preferred embodiment of the intelligent control method for eliminating reaction force in a loader according to the present invention, wherein: the feedback parameters in step S5 include boom lifting angle deviation, bucket tilting angle deviation, actual pressure of boom cylinder, actual pressure of bucket cylinder, and overall machine center of gravity acceleration; the preset target parameters are target values ​​for boom lifting angle, bucket tilting angle, target pressure of boom cylinder, target pressure of bucket cylinder, and zero target value for center of gravity acceleration; the control parameters of the PID algorithm include proportional coefficient Kp, integral... The coefficients Ki and the derivative coefficient Kd are initially tuned using a particle swarm optimization algorithm for the PID control parameters. After initial tuning, the values ​​of Kp range from 5 to 15, Ki range from 0.1 to 0.5, and Kd range from 1 to 5. During the control process, Kp, Ki, and Kd are dynamically adjusted according to the rate of change of the control error. When the absolute value of the control error is greater than 0.1, Kp is increased and Ki is decreased; when the absolute value of the control error is less than 0.01, Kp is decreased and Ki is increased to ensure the speed and stability of the control process.

[0029] As a preferred embodiment of the intelligent control method for loader with reaction force elimination described in this invention, the method further includes step S6: constructing a fault diagnosis and fault-tolerant control module to monitor the operating status of multiple sensors, hydraulic execution system, and walking system in real time; when a sensor fault is detected, redundant data in the data fusion algorithm is used to replace the faulty sensor data; when a hydraulic execution system fault is detected, the intelligent control unit switches to a backup control strategy to reduce the control accuracy requirements and ensure that the loader can safely complete the current operation or stop; the response time of fault diagnosis does not exceed 0.5s.

[0030] As a preferred embodiment of the intelligent control method for loader with reaction force elimination described in this invention, step S1 further includes filtering the collected real-time parameters and using a Kalman filter algorithm to eliminate sensor noise. Here, X(k) is the state vector at time k, A is the state transition matrix, B is the control input matrix, U(k) is the control input vector, W(k) is the process noise, Z(k) is the observation vector, H is the observation matrix, and V(k) is the observation noise. The parameters are smoothed using the Kalman filter algorithm, reducing the parameter fluctuation amplitude to within 5%.

[0031] The beneficial effects of this invention are:

[0032] Precise perception and prediction of reaction force: Through multi-sensor fusion technology, the system achieves comprehensive perception of the working status and environment. Combined with an improved BP neural network model, it accurately predicts the magnitude and direction of the material reaction force with a prediction error of less than 5%. This provides reliable data support for the elimination of reaction force and solves the core defect of traditional technology that cannot accurately perceive reaction force.

[0033] Multi-dimensional collaborative control eliminates reaction forces: A multi-dimensional control strategy is constructed to compensate for boom cylinder pressure, bucket cylinder pressure, and coordinate the overall machine posture. This enables the linkage control of the hydraulic actuator system and the walking system, which can offset the material reaction forces in different directions in all directions, effectively reduce the vibration amplitude of the whole machine, and improve the operation accuracy.

[0034] Dynamic adaptive control enhances stability: The PID control parameters are tuned using a particle swarm optimization algorithm and dynamically adjusted according to the control error, enabling the control strategy to adapt to changes in different operating scenarios. This results in fast control response, high stability, and control error that can be kept within 0.01.

[0035] Fault tolerance ensures operational safety: The addition of a fault diagnosis and fault-tolerant control module can quickly detect and handle sensor and execution system faults. Through redundant data replacement and backup control strategy switching, it ensures operational continuity and safety, reduces downtime due to failures, and improves system reliability.

[0036] Extend equipment lifespan: By effectively eliminating material reaction forces, the wear intensity of key components such as the hydraulic system and chassis structure is reduced. Actual tests have verified that the lifespan of key equipment components is extended by more than 20%, reducing user operating costs. Attached Figure Description

[0037] Figure 1 This is a basic flowchart illustrating a reaction force elimination intelligent control method for loaders provided in one embodiment of the present invention. Detailed Implementation

[0038] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0039] Reference Figure 1 As an embodiment of the present invention, the specific implementation steps of a loader intelligent control method for eliminating reaction force are as follows:

[0040] Step S1: Construct a loader operation status perception system and deploy various types of sensors: Use a WDD35D4 angle sensor to collect the boom lifting angle and bucket tilting angle, with a sampling frequency of 100Hz; Use a PT124G-111 pressure sensor to collect the rodless and rod chamber pressures of the boom cylinder and bucket cylinder, with a measurement accuracy of ±0.5MPa; Use a Hall effect speed sensor (model: HS-100) to collect the travel speed, with a measurement range of 0-30km / h; Use a magnetoelectric speed sensor (model: CS-200) to collect the engine speed, with a measurement range of 800-2500r / min; Use a USB 3.0 industrial camera (vision sensor) and a Velodyne VLP-16 lidar sensor to collaboratively collect environmental parameters of the working surface, use a convolutional neural network image recognition algorithm to analyze the material type (crushed stone), with an accuracy of 96.5%, and use a point cloud clustering algorithm to measure the material accumulation height (1.2m) and the slope of the working surface (3.2°), with a slope measurement accuracy of ±0.3°. Meanwhile, the Kalman filter algorithm is used to filter the collected parameters. After filtering, the boom lifting angle fluctuation range is reduced from 8% to 3%, and the cylinder pressure fluctuation range is reduced from 10% to 4%, ensuring the reliability of the sensing data.

[0041] Step S2: Construct an improved BP neural network reaction force prediction model: Collect historical operating data (1200 sets) of a 5-ton loader in scenarios of gravel loading, sand loading, coal loading and unloading, and corresponding measured reaction force data. Divide the data into a training dataset (840 sets) and a test dataset (360 sets) at a 7:3 ratio. Map the training dataset to the [0,1] interval using the min-max normalization algorithm. Set the neural network structure to 12 input layers, 20 hidden layers 1-15 hidden layers 2-3 output layers. Train the network using the momentum gradient descent algorithm, setting the momentum factor α=0.2 and the learning rate η=0.03. After 920 iterations, the network error reaches 0.0009, and training is stopped. Validate the model using the test dataset. The average prediction error is 3.8%, which meets the accuracy requirements. Input the real-time parameters collected in step S1 into the model to predict the instantaneous reaction force of the gravel on the bucket: x-axis component F. x =4800N, y-axis component Fᵧ=1900N, z-axis component Fz=11500N.

[0042] Step S3: Constructing the loader's dynamic model and reaction force elimination control strategy: Establish a rectangular coordinate system with the loader's center of mass as the origin, with the x-axis along the forward direction, the y-axis along the horizontal direction, and the z-axis perpendicular to the ground and upward; establish translational and rotational motion equations based on Newton's laws of motion and rigid body dynamics principles. It is known that the 5-ton loader has a total mass m = 5000 kg, an effective area S = 0.02 m² for the rodless chamber of the boom cylinder, a hydraulic system efficiency ηh = 0.9, and a gravitational acceleration g = 9.8 m / s².

[0043] Based on the reaction force data, the control parameters are calculated as follows: the boom cylinder compensation pressure Pcomp is calculated based on the z-axis component of the reaction force; the bucket cylinder compensation pressure difference is obtained through inverse kinematics and is 780 kPa; the overall machine attitude adjustment parameters are: the travel speed correction is -3 km / h (i.e., the travel speed is reduced by 3 km / h), and the steering angle correction is 7°.

[0044] Step S4: The intelligent control unit (using an STM32H743 microprocessor) executes the control strategy and outputs control commands: the boom cylinder oil supply pressure is increased to the calculated compensation pressure through the proportional relief valve; the bucket cylinder oil inlet and return flow is adjusted through the proportional directional valve to achieve 780kPa pressure difference compensation; the travel system controller reduces the travel speed from 6km / h to 3km / h and adjusts the steering angle by 7° to achieve reaction force cancellation.

[0045] Step S5: Collect feedback parameters: the deviation of the boom lifting angle is 0.7°, the deviation of the bucket flipping angle is 0.4°, the deviation between the actual pressure and the target pressure of the boom cylinder is 75 kPa, and the control error is 0.007; the parameters are dynamically adjusted using the PID algorithm, Kp is adjusted from 12 to 9, Ki is adjusted from 0.3 to 0.4, and after adjustment, the angle deviation is reduced to 0.2°, the pressure deviation is reduced to 28 kPa, and the control is stable.

[0046] Step S6: The fault diagnosis and fault tolerance control module monitors the status of each system in real time, and no faults are detected throughout the process, and the system runs normally.

[0047] Implementation effect: After adopting the method of the present invention, the vibration amplitude of the whole machine during the loading of crushed stones by the loader is reduced from 18 mm to 5.5 mm, the operation accuracy is improved by 38%, the wear of the boom cylinder is reduced by 24%, the effective elimination of the reaction force is achieved, and the operation stability and equipment durability are improved.

[0048] In the present invention, through the multi-sensor fusion technology, the comprehensive perception of the operation status and the environment is realized, and the magnitude and direction of the material reaction force are accurately predicted by combining the improved BP neural network model, and the prediction error is less than 5%, providing reliable data support for the elimination of the reaction force, solving the core defect that the traditional technology cannot accurately perceive the reaction force, constructing a multi-dimensional control strategy of boom cylinder pressure compensation, bucket cylinder pressure compensation and overall machine attitude cooperative adjustment, realizing the linkage control of the hydraulic execution system and the walking system, and can offset the material reaction forces in different directions in all directions, effectively reducing the vibration amplitude of the whole machine, improving the operation accuracy, using the particle swarm optimization algorithm to tune the PID control parameters, and dynamically adjusting the parameters according to the control error, so that the control strategy can adapt to the changes of different operation scenarios, with fast control response speed and high stability, and the control error can be stabilized within 0.01. A fault diagnosis and fault tolerance control module is added, which can quickly detect and handle sensor and actuator system faults, and ensure the operation continuity and safety through redundant data replacement, standby control strategy switching, etc., reduce the fault shutdown rate, improve the system reliability, effectively eliminate the material reaction force, reduce the wear intensity of key components such as the hydraulic system and the chassis structure, and through actual measurement and verification, the service life of the key components of the equipment is extended by more than 20%, reducing the user's use cost.

[0049] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A reaction force cancellation intelligent control method for a loader, characterized by, Includes the following steps: Step S1: Construct a loader operation status perception system. Collect real-time operating parameters and environmental parameters during the loader operation process through multi-sensor fusion. The real-time operating parameters include boom lifting angle, bucket tilting angle, boom cylinder pressure, bucket cylinder pressure, loader travel speed, and engine speed. The environmental parameters include material type on the working surface, material accumulation height, and working surface slope. Step S2: Construct a reaction force prediction model based on the collected real-time parameters. Calculate the magnitude and direction of the instantaneous reaction force generated by the material on the bucket during the loader's loading, lifting, and unloading processes using the reaction force prediction model. The reaction force prediction model adopts an improved BP neural network model, with its input layer being the real-time operating parameters and environmental parameters collected in step S1, and its output layer being the magnitude and direction vector of the instantaneous reaction force. Step S3: Based on the instantaneous reaction force data output by the reaction force prediction model, and combined with the dynamic model of the loader, a reaction force elimination control strategy is constructed. The reaction force elimination control strategy includes boom cylinder pressure compensation control, bucket cylinder pressure compensation control, and overall machine attitude coordinated adjustment control. Step S4: The intelligent control unit executes the reaction force elimination control strategy, outputs control commands to the hydraulic execution system and travel system of the loader in real time, dynamically adjusts the oil supply pressure and flow of the boom cylinder and bucket cylinder, and simultaneously adjusts the loader's travel speed and steering angle to offset the reaction force of the material. Step S5: Collect feedback parameters after the loader executes control commands in real time, compare the feedback parameters with the preset target parameters, calculate the control error, and dynamically correct the control parameters of the reaction force elimination control strategy through the PID algorithm to ensure the stability of the reaction force elimination effect.

2. A reaction force cancellation intelligent control method for a loader as set forth in claim 1, the improvement wherein: The multi-sensor fusion in step S1 specifically includes: using an angle sensor to collect the boom lifting angle and bucket tilting angle, with a sampling frequency of 100Hz; using a pressure sensor to collect the pressure in the rodless and rod chambers of the boom cylinder and the rodless and rod chambers of the bucket cylinder, with a measurement accuracy of ±0.5MPa; using a speed sensor to collect the loader's travel speed, with a measurement range of 0-30km / h; using a speed sensor to collect the engine speed, with a measurement range of 800-2500r / min; and using a vision sensor and a lidar sensor to collect the material type, material accumulation height, and slope of the working surface. The collected data is then fused and analyzed using image recognition algorithms and point cloud processing algorithms, with a material type recognition accuracy of no less than 95% and a slope measurement accuracy of ±0.5°.

3. The intelligent control method for loader with reaction force elimination as described in claim 2, characterized in that: The construction process of the improved BP neural network model described in step S2 includes: Step S21: Collecting historical operating data and corresponding measured reaction force data of the loader under different operating scenarios, constructing training datasets and test datasets. The operating scenarios include sand loading, gravel loading, coal loading and unloading operations, and the ratio of the training dataset to the test dataset is 7:3; Step S22: Normalizing the training dataset, mapping the input parameters and output parameters to the [0,1] interval, and using the min-max normalization algorithm; Step S23: Setting the structure of the BP neural network, with 12 nodes in the input layer corresponding to 12 input parameters, and 2 hidden layers, with 2 nodes in the first hidden layer. The first layer has 0 nodes, the second hidden layer has 15 nodes, and the output layer has 3 nodes, corresponding to the x-axis, y-axis, and z-axis components of the reaction force, respectively; Step S24: The momentum gradient descent algorithm is used to optimize the weights and thresholds of the BP neural network, and a momentum factor α is introduced, with a value range of 0.1-0.

3. The learning rate η ranges from 0.01-0.

05. The network is iteratively trained using the training dataset. Training is stopped when the number of iterations reaches 1000 or the network error is less than 0.001; Step S25: The trained BP neural network is validated using the test dataset. If the model prediction error is greater than 5%, the network structure and training parameters are readjusted until the prediction accuracy requirements are met.

4. The intelligent control method for loader with reaction force elimination as described in claim 3, characterized in that: The process of constructing the dynamic model of the loader in step S3 includes: establishing a rectangular coordinate system with the center of mass of the loader as the origin, with the x-axis along the forward direction of the loader, the y-axis along the horizontal direction of the loader, and the z-axis perpendicular to the ground and upward; based on Newton's laws of motion and the principles of rigid body dynamics, establishing the translational motion equation and rotational motion equation of the loader, considering the inertial force, gravity, ground support force, driving force of the hydraulic cylinder, and reaction force of the material of the boom, bucket, and material; Equations of translational motion: ； ； ； Equations of rotational motion: ； ； ； Where m is the mass of the loader, ax, ay, and az are the accelerations of the center of mass in the x, y, and z axes, respectively, Fx, Fy, and Fz are the components of the material reaction force in the x, y, and z axes, respectively, Ffx and Ffy are the frictional forces in the x and y axes, respectively, N is the ground support force, g is the acceleration due to gravity, Jx, Jy, and Jz are the moments of inertia of the loader about the x, y, and z axes, respectively, βx, βy, and βz are the angular accelerations about the x, y, and z axes, respectively, and Mx, My, and Mz are the torques generated by the material reaction force about the x, y, and z axes, respectively.

5. The intelligent control method for loader with reaction force elimination as described in claim 4, characterized in that: The boom cylinder pressure compensation control described in step S3 is as follows: Based on the reaction force z-axis component output by the reaction force prediction model, the required compensation pressure for the boom cylinder is calculated, where Pcomp is the compensation pressure, S is the effective area of ​​the rodless chamber of the boom cylinder, and ηh is the hydraulic system efficiency; The intelligent control unit adjusts the oil supply pressure of the boom cylinder through the proportional relief valve according to the calculated compensation pressure, so that the total driving force of the boom cylinder is equal to the sum of the z-axis component of the material reaction force and the z-axis component of the weight of the boom and bucket.

6. The intelligent control method for loader with reaction force elimination as described in claim 5, characterized in that: The bucket cylinder pressure compensation control described in step S3 is as follows: based on the x-axis and y-axis components of the reaction force output by the reaction force prediction model, the required compensation pressure of the bucket cylinder is calculated. The mapping relationship between the extension and retraction of the bucket cylinder and the reaction force components is obtained through inverse kinematics, thereby determining the compensation pressure difference between the rodless chamber and the rod chamber of the bucket cylinder. The intelligent control unit adjusts the inlet and outlet flow rates of the bucket cylinder through the proportional directional valve to achieve dynamic compensation of the bucket cylinder pressure and counteract the influence of the material reaction force on the bucket's tilting posture.

7. The intelligent control method for loader with reaction force elimination as described in claim 6, characterized in that: The overall machine attitude coordinated adjustment control described in step S3 is as follows: Based on the dynamic model of the loader, and combined with the magnitude and direction of the reaction force, the required attitude adjustment parameters of the loader are calculated, including the driving speed correction and steering angle correction; when the component of the reaction force along the x-axis is greater than a preset threshold, the intelligent control unit controls the walking system to reduce the driving speed, and the reduction is proportional to the x-axis component of the reaction force; when the component of the reaction force along the y-axis is greater than a preset threshold, the intelligent control unit controls the steering system to adjust the steering angle so that the driving direction of the loader is consistent with the y-axis component of the reaction force, thereby reducing the risk of lateral overturning.

8. The intelligent control method for loader with reaction force elimination as described in claim 7, characterized in that: The feedback parameters in step S5 include boom lifting angle deviation, bucket tilting angle deviation, actual boom cylinder pressure, actual bucket cylinder pressure, and overall machine center of gravity acceleration. The preset target parameters are target values ​​for boom lifting angle, bucket tilting angle, boom cylinder pressure, bucket cylinder pressure, and zero target value for center of gravity acceleration. The control parameters of the PID algorithm include proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd. The particle swarm optimization algorithm is used to initially tune the PID control parameters. After initial tuning, the value range of Kp is 5-15, the value range of Ki is 0.1-0.5, and the value range of Kd is 1-5. During the control process, Kp, Ki, and Kd are dynamically adjusted according to the rate of change of the control error. When the absolute value of the control error is greater than 0.1, Kp is increased and Ki is decreased; when the absolute value of the control error is less than 0.01, Kp is decreased and Ki is increased to ensure the speed and stability of the control process.

9. The intelligent control method for loader with reaction force elimination as described in claim 8, characterized in that: It also includes step S6: constructing a fault diagnosis and fault-tolerant control module to monitor the operating status of multiple sensors, hydraulic actuators and walking systems in real time. When a sensor fault is detected, redundant data in the data fusion algorithm is used to replace the faulty sensor data. When a hydraulic actuator fault is detected, the intelligent control unit switches to the backup control strategy to reduce the control accuracy requirements and ensure that the loader can safely complete the current operation or stop. The response time of fault diagnosis does not exceed 0.5s.

10. The intelligent control method for loader with reaction force elimination as described in claim 9, characterized in that: Step S1 further includes filtering the acquired real-time parameters and using a Kalman filter algorithm to eliminate sensor noise. Here, X(k) is the state vector at time k, A is the state transition matrix, B is the control input matrix, U(k) is the control input vector, W(k) is the process noise, Z(k) is the observation vector, H is the observation matrix, and V(k) is the observation noise. The parameters are smoothed by the Kalman filter algorithm, reducing the parameter fluctuation amplitude to within 5%.

Images