Agricultural tractor body leveling control method and system

Abstract

The invention relates to the technical field of mechanical intelligent control, in particular to an agricultural tractor body leveling control method and system, and the method comprises the following steps: constructing a data collection device of an agricultural tractor, and obtaining the inclination angle of the agricultural tractor body according to the data collection device; a multi-degree-of-freedom cooperative decoupling control framework is established, and a preliminary leveling control instruction of the agricultural tractor is obtained in combination with the vehicle body inclination angle; obtaining a terrain discrete elevation point set of the agricultural tractor, and constructing a terrain elevation prediction model according to the terrain discrete elevation point set; based on the preliminary leveling control instruction, utilizing the terrain elevation prediction model to obtain an optimized leveling control instruction of the agricultural tractor; and the vehicle body inclination angle is regulated and controlled according to the optimized leveling control instruction, and vehicle body leveling of the agricultural tractor is achieved. The operation stability, the adaptability and the intelligent level of the agricultural tractor in the unstructured terrain are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical intelligent control, and in particular to a body leveling control method and system for an agricultural tractor. Background Art

[0002] Currently, agricultural tractor body leveling technology primarily relies on a combination of mechanical structures and hydraulic systems. Mainstream leveling solutions include hydraulic differential height adjustment and bending and twisting mechanisms. The former uses hydraulic cylinders to adjust the left and right height differences of the vehicle body for simple leveling. While its structure is simple, its adaptability is limited. The latter utilizes an articulated frame to passively adapt to the terrain on uneven roads. While this improves maneuverability, it sacrifices active control precision. Control strategies often employ a proportional-integral-derivative controller (PID) algorithm based on an inclination sensor. This algorithm uses real-time detection of the vehicle body's inclination angle to actuate a hydraulic valve to adjust the cylinder's extension and retraction. A single hydraulic cylinder controls the lateral tilt angle, and the PID algorithm is combined to optimize response speed. Some technologies incorporate Beidou positioning and an inertial measurement unit (IMU) for path tracking. Position information is acquired through satellite positioning, and attitude data is acquired by the IMU to generate basic leveling commands.

[0003] However, existing technologies mostly focus on a single degree of freedom, fail to fully coordinate multi-degree-of-freedom parameters such as vehicle pitch and tillage depth, and rely on fixed thresholds to trigger leveling; there are the following defects: lack of multi-degree-of-freedom coordination, traditional methods independently control vehicle posture and tillage depth, ignoring the dynamic coupling between the two, and fixed PID parameters cannot dynamically compensate for such coupling effects; insufficient terrain pre-adaptation capability, relying on real-time tilt angle feedback, lack of terrain prediction mechanism, resulting in leveling action lagging behind actual terrain changes; passive safety protection, anti-rollover design mainly uses mechanical roll cages, which only provide protection after rollover, and no active protection.

[0004] In light of this, the present invention proposes a body leveling control method and system for agricultural tractors. By constructing a multi-degree-of-freedom collaborative decoupling control architecture, integrating body angle data and tillage depth parameters, establishing a three-degree-of-freedom coupled dynamics model, and designing a three-channel collaborative control strategy for lateral leveling, longitudinal posture stabilization, and constant tillage depth, this method achieves multi-parameter dynamic decoupling and joint optimization. Millimeter-wave radar terrain scanning and a Gaussian regression prediction model are introduced to generate a short-term posture change prediction sequence, driving control command feedforward compensation to ensure that leveling actions are ahead of terrain changes. A hierarchical active protection mechanism is designed, and a rollover risk assessment system based on dynamic thresholds is constructed. This implements a three-level intervention strategy, from preventative posture stabilization to emergency protection, upgrading passive protection to active risk management. This method significantly improves body stability, operational accuracy, and operational safety in complex terrain, providing key technical support for the intelligentization of modern agricultural equipment. Summary of the Invention

[0005] In view of the defects in the prior art, the present invention provides a method and system for controlling the leveling of an agricultural tractor body.

[0006] To achieve the above-mentioned objectives, in a first aspect, the present invention provides a method for controlling body leveling of an agricultural tractor, the method comprising the following steps: constructing a data acquisition device for the agricultural tractor, and obtaining the body tilt angle of the agricultural tractor based on the data acquisition device; establishing a multi-degree-of-freedom collaborative decoupling control architecture, and obtaining preliminary leveling control instructions for the agricultural tractor in combination with the body tilt angle; obtaining a terrain discrete elevation point set of the agricultural tractor, and constructing a terrain elevation prediction model according to the terrain discrete elevation point set; based on the preliminary leveling control instruction, obtaining an optimized leveling control instruction for the agricultural tractor using the terrain elevation prediction model; and regulating the body tilt angle according to the optimized leveling control instruction to achieve body leveling of the agricultural tractor. The present invention constructs a high-precision data acquisition device through multi-sensor fusion, combines it with a dynamic filtering algorithm to improve the measurement reliability of the vehicle body tilt angle, and provides accurate state perception for leveling control; the multi-degree-of-freedom collaborative decoupling architecture realizes the joint optimization of lateral leveling, longitudinal posture stability and tillage depth parameters, breaking through the limitations of single-degree-of-freedom control and enhancing the leveling accuracy and dynamic response under complex working conditions; the terrain elevation prediction model is based on radar scanning and Gaussian regression technology to establish advanced terrain perception capabilities, drive control command feedforward compensation, and effectively solve the leveling lag problem; through the coordination of prediction model optimization instructions and hierarchical active protection strategies, accurate pre-adjustment and safety intervention of leveling actions are achieved, thereby improving the operating stability, adaptability and intelligence level of agricultural tractors in unstructured terrain.

[0007] Optionally, the method of constructing a data acquisition device for an agricultural tractor and obtaining the body tilt angle of the agricultural tractor based on the data acquisition device includes: combining a wheel speed sensor, a six-axis inertial measurement unit, and a suspension point displacement sensor to construct the data acquisition device; obtaining wheel speed signals, angle data, and hydraulic cylinder extension and contraction amount based on the data acquisition device to construct a state vector for the agricultural tractor; constructing a variance-driven sliding window and dynamically filtering the state vector based on the sliding window to obtain an optimized state vector; and performing tilt angle calculation based on the optimized state vector to obtain the attitude angle of the agricultural tractor, and using the attitude angle as the body tilt angle. The present invention significantly improves the measurement accuracy of the body tilt angle through multi-source sensor fusion and dynamic filtering mechanisms, achieving deep coupling of kinematic parameters and attitude data, and constructing a state vector that comprehensively reflects the dynamic characteristics of the body. The variance-driven sliding window filtering algorithm suppresses sensor noise and transient interference in real time, effectively eliminating the impact of data jumps on tilt angle calculation. The optimized state vector is jointly calculated through multiple parameters to output high-confidence attitude angle data, providing accurate data input for leveling control.

[0008] Optionally, establishing a multi-degree-of-freedom collaborative decoupling control architecture includes: obtaining the three-degree-of-freedom generalized coordinate vector of the agricultural tractor, constructing a three-degree-of-freedom coupling model of the agricultural tractor based on the three-degree-of-freedom generalized coordinate vector; constructing a collaborative decoupling control strategy for the agricultural tractor, the collaborative decoupling control strategy including a lateral leveling channel, a longitudinal attitude stabilization channel, and a constant tillage depth channel; and using the three-degree-of-freedom coupling model and the collaborative decoupling control strategy as the multi-degree-of-freedom collaborative decoupling control architecture. The present invention achieves dynamic decoupling and joint optimization of vehicle body angle data and tillage depth parameters by constructing a three-degree-of-freedom coupling model and a collaborative decoupling control strategy. The lateral leveling channel ensures roll stability, the longitudinal attitude stabilization channel suppresses bump vibration, and the constant tillage depth channel maintains consistency in working depth. The three channels work together to break through the limitations of traditional single-degree-of-freedom control, effectively compensating for multi-parameter coupling effects, improving leveling accuracy and response speed in complex terrain, ensuring high stability and working quality of the agricultural tractor, and enhancing its adaptability to multiple working conditions.

[0009] Optionally, the method of obtaining a preliminary leveling control instruction for the agricultural tractor in combination with the vehicle body tilt angle includes: determining a dynamic leveling control target for the agricultural tractor's vehicle body based on the vehicle body tilt angle; obtaining a decoupling input item for the agricultural tractor in combination with the three-degree-of-freedom coupling model and the vehicle body dynamic leveling control target; and parsing the decoupling input item according to the collaborative decoupling control strategy to obtain the preliminary leveling control instruction. The present invention achieves precise adjustment of vehicle body posture through dynamic leveling target setting and multi-degree-of-freedom decoupling control; generates a dynamic control target based on the real-time tilt angle, calculates independent control input items in combination with the three-degree-of-freedom coupling model, and then converts the multi-parameter coupling effect into an executable instruction through a collaborative decoupling strategy; effectively improves the accuracy and response speed of leveling instructions, ensures the coordinated optimization of angle data and tillage depth parameters in complex terrain, significantly enhances operation stability and leveling accuracy, and provides reliable protection for precision agricultural operations.

[0010] Optionally, obtaining the agricultural tractor's terrain discrete elevation point set and constructing a terrain elevation prediction model based on the terrain discrete elevation point set includes: scanning the terrain using millimeter-wave radar to obtain an original point cloud set, preprocessing the original point cloud set to obtain the terrain discrete elevation point set; compensating the terrain discrete elevation point set based on the agricultural tractor's body motion posture to obtain an optimized terrain discrete elevation point set; and constructing the terrain elevation prediction model based on the optimized terrain discrete elevation point set in combination with Gaussian regression. The present invention utilizes millimeter-wave radar and Gaussian regression algorithms to construct a terrain elevation prediction model, improving its adaptability to complex terrain. Motion compensation of the high-precision point cloud data acquired by radar scanning eliminates interference with terrain perception caused by body vibration and posture changes, ensuring accurate terrain modeling. Combined with the non-parametric nature of Gaussian regression, the model can capture local terrain details. Furthermore, it provides advanced terrain information for leveling, enabling control commands to pre-compensate for ground undulations. This effectively addresses the leveling lag caused by feedback delays in traditional methods and improves the operational stability of agricultural tractors in unstructured terrain.

[0011] Optionally, the method of obtaining the optimized leveling control instruction for the agricultural tractor based on the preliminary leveling control instruction and utilizing the terrain elevation prediction model includes: obtaining a short-term posture change prediction sequence of the agricultural tractor according to the terrain elevation prediction model; pre-evolving the posture of the agricultural tractor according to the short-term posture change prediction sequence to obtain a time-domain posture evolution vector; obtaining a predicted compensation amount based on the time-domain posture evolution vector, and combining the preliminary leveling control instruction and the predicted compensation amount to obtain the optimized leveling control instruction. The present invention implements feedforward optimization of the leveling instruction through the terrain elevation prediction model, significantly improving the predictability and accuracy of the control system; the time-domain evolution vector constructed based on the short-term posture prediction sequence can quantify the vehicle posture deviation caused by the future terrain in advance, generate a predicted compensation amount to dynamically correct the preliminary instruction; effectively compensate for the hysteresis defect of traditional PID control, make the leveling action ahead of the actual terrain change, reduce the amplitude of posture fluctuation, and provide forward control capability for the intelligentization of agricultural tractors.

[0012] Optionally, the body tilt angle is regulated according to the optimized leveling control instruction to achieve body leveling of the agricultural tractor, including: controlling the solenoid valve to input or output hydraulic oil according to the optimized leveling instruction to control the extension and retraction of the hydraulic rod to achieve regulation of the body tilt angle; during the regulation process, the body tilt angle is monitored in real time and dynamically intervened to achieve body leveling of the agricultural tractor. The present invention achieves precise regulation of the body posture through hydraulic actuators and closed-loop monitoring; the solenoid valve controls the flow of hydraulic oil to drive the extension and retraction of the hydraulic rod, converting the optimized instruction into a high-precision mechanical action, improving the leveling response speed and execution accuracy; real-time monitoring serves as a dynamic feedback loop, and the body posture is continuously corrected in combination with the tilt sensor data, effectively suppressing posture deviation caused by external disturbances; ensuring high stability and strong anti-interference capability of the leveling process, while reducing the risk of operation interruption due to posture loss of control.

[0013] Optionally, the real-time monitoring and dynamic intervention of the vehicle body tilt angle to achieve the body leveling of the agricultural tractor includes: real-time monitoring of the vehicle body tilt angle to obtain the rollover risk coefficient of the agricultural tractor; constructing a graded active protection strategy for the agricultural tractor based on the rollover risk coefficient; and dynamically intervening in the vehicle body tilt angle according to the graded active protection strategy to achieve the body leveling of the agricultural tractor. The present invention significantly improves operational safety by constructing a rollover risk assessment and graded intervention mechanism; real-time monitoring and continuous solution of vehicle body tilt angle data, dynamic calculation of the rollover risk coefficient, and implementation of risk quantification assessment; based on the active protection strategy divided by risk level, triggering multi-level interventions such as preventive stabilization, active anti-rollover, or emergency protection for different risk levels, upgrading traditional passive protection to active risk management; effectively reducing the probability of rollover in complex terrain operations, while dynamically adjusting the leveling intervention intensity to optimize operational efficiency while ensuring safety, achieving a balance between high safety and high operational efficiency.

[0014] Optionally, the hierarchical active protection strategy for the agricultural tractor based on the rollover risk coefficient includes: constructing a dynamic threshold function, and obtaining a dynamic safety threshold based on the dynamic threshold function; obtaining the rollover risk level of the agricultural tractor by combining the rollover risk coefficient and the dynamic safety threshold, wherein the rollover risk level includes a first-level warning, a second-level intervention, and a third-level protection; constructing the hierarchical active protection strategy based on the rollover risk level, wherein the hierarchical active protection strategy includes a first-level preventive posture stabilization strategy, a second-level active anti-rollover strategy, and a third-level emergency protection strategy. The present invention improves the safety and intelligence level of operations through dynamic risk quantification and a hierarchical protection mechanism; the dynamic threshold function generates an adaptive safety boundary in combination with real-time working condition parameters to avoid misjudgment or omission of fixed thresholds in complex terrain; the three-level protection strategy based on the risk level division realizes a gradient response from risk warning to emergency protection: the first-level strategy prevents risk accumulation by fine-tuning the posture, the second-level strategy initiates active posture stabilization to suppress tilt angle development, and the third-level strategy triggers emergency measures such as hydraulic locking; effectively reduces the rollover accident rate, and at the same time maximizes operating efficiency within the safety boundary by dynamically adjusting the intervention intensity.

[0015] In a second aspect, the present invention provides an agricultural tractor body leveling control system, which implements the agricultural tractor body leveling control method provided by the present invention. The system is characterized in that the system includes an input device, an output device, a processor, and a memory, wherein the input device, output device, processor, and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions. The present invention implements an agricultural tractor body leveling control system through hardware collaborative design; the input and output devices establish a data closed loop, realizing real-time interaction between sensor signal acquisition and actuator control; the processor is equipped with a control algorithm to complete multi-source data fusion, dynamic decoupling calculation, and feedforward compensation decision-making; the memory solidifies the intelligent control program to ensure algorithm stability under complex working conditions; the leveling accuracy, response speed, and safety redundancy of agricultural machinery in unstructured terrain are improved, and operational stability is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a method for controlling body leveling of an agricultural tractor according to an embodiment of the present invention; Figure 2 This is a framework diagram of an agricultural tractor body leveling control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the present invention. In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0018] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, those of ordinary skill in the art will appreciate that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0019] See Figure 1 One embodiment of the present invention provides a method for controlling the leveling of an agricultural tractor body, the method comprising the following steps: S1. Construct a data acquisition device for an agricultural tractor, and obtain a body tilt angle of the agricultural tractor according to the data acquisition device.

[0020] In this embodiment, a wheel speed sensor, a six-axis inertial measurement unit, and a suspension point displacement sensor are combined to construct a data acquisition device; wheel speed signals, angle data, and hydraulic cylinder extension and contraction amount are obtained according to the data acquisition device to construct a state vector of the agricultural tractor; a variance-driven sliding window is constructed, and the state vector is dynamically filtered according to the sliding window to obtain an optimized state vector; the inclination angle is calculated based on the optimized state vector to obtain the attitude angle of the agricultural tractor, and the attitude angle is used as the inclination angle of the vehicle body.

[0021] The data acquisition system of agricultural tractors integrates multiple types of sensors to comprehensively perceive vehicle posture information. The wheel speed sensor, a key component, is installed at the end of the drive wheel axle and calculates actual driving speed by monitoring the wheel rotation frequency. The selection of the sensor should take into account both range and resolution. For example, a non-contact Hall effect sensor can be used to avoid mechanical wear and adapt to complex farmland conditions. The deployment of the six-axis inertial measurement unit takes into account installation stiffness and position, and is preferably fixed near the tractor's center of mass. Its three-axis accelerometer and three-axis gyroscope can simultaneously collect roll angle, pitch angle, and angular velocity data. Its update frequency is no less than 100Hz to meet dynamic response requirements. The suspension point displacement sensor uses a wire-type or magnetostrictive displacement sensor, installed at the piston rod end of the front and rear suspension hydraulic cylinders, respectively. It monitors the hydraulic cylinder's extension and contraction in real time, directly reflecting the vehicle height adjustment status. The sensor layout should follow the principle of redundant design. For example, wheel speed sensors can be arranged symmetrically on the left and right drive wheels to improve data reliability through cross-validation, while avoiding electromagnetic interference between sensors. Shielded cables and independent power modules are used to ensure signal integrity.

[0022] The data acquisition device realizes the synchronous acquisition of wheel speed signals, angle data and displacement sensor signals. The pulse signal output by the wheel speed sensor is converted into a standard square wave by a shaping circuit to calculate the actual speed; the six-axis inertial measurement unit transmits the original acceleration and angular velocity data, which needs to be temperature compensated and zero-biased to eliminate environmental interference; the suspension point displacement sensor outputs an analog voltage signal, which is combined with the calibration curve to analyze the actual displacement. In order to solve the problem of timing deviation of multi-source signals, a hardware trigger synchronization mechanism is adopted. The wheel speed and displacement data are aligned through the global clock signal based on the angle data acquisition time. Then the state vector of the agricultural tractor is constructed to satisfy the following relationship: in, is the state vector, is the roll angle, is the pitch angle, is the rolling angular velocity, is the pitch angular velocity, is the height of the suspension point, Indicates transpose.

[0023] A variance-driven sliding window is designed. The state vector is dynamically filtered within a fixed-length time window (e.g., 500ms) to obtain an optimized state vector. The variance of each element within the window is calculated. When the variance exceeds a preset threshold, the current data is deemed to be affected by transient interference, and the abnormal window data is smoothed using an exponentially weighted average method. If the variance is within the normal range, the median value of the window is directly used as the optimized state vector. This dynamic filtering mechanism can adapt to bumpy farmland conditions, suppressing abnormal jumps while maintaining dynamic response characteristics. For example, when a tractor crosses a ditch, it can effectively filter the transient impact signal of the hydraulic cylinder displacement sensor, preventing malfunction of the leveling system.

[0024] The above sliding window satisfies the following relationship: in, is the window sliding length, is the angular variance.

[0025] The above optimized state vector includes: in, To optimize the angle, is the actual angle, is the window sliding length, is the index variable, is the historical filtering value within the window.

[0026] The roll and pitch angles in the optimized state vector directly reflect the vehicle body's tilt state and are compensated for by combining them with suspension displacement information. First, a complementary filtering algorithm fuses angular velocity data with accelerometer data to eliminate gyroscope integral drift errors. Second, a geometric relationship between suspension displacement and vehicle body tilt angle is established. When the tractor rolls, the displacement difference between the left and right suspension hydraulic cylinders and the roll angle form a sinusoidal relationship. Finally, using the optimized state vector as observation data, an extended Kalman filter fusion solution is employed, combining the state equation and the observation equation to obtain the optimal attitude angle estimate. This attitude angle serves as the vehicle body tilt angle, providing precise input for the subsequent leveling control algorithm.

[0027] S2. Establish a multi-degree-of-freedom collaborative decoupling control architecture, and obtain preliminary leveling control instructions for the agricultural tractor in combination with the vehicle body tilt angle.

[0028] In this embodiment, a three-degree-of-freedom generalized coordinate vector of an agricultural tractor is obtained, and a three-degree-of-freedom coupling model of the tractor is constructed based on the three-degree-of-freedom generalized coordinate vector; a collaborative decoupling control strategy for the agricultural tractor is constructed, and the collaborative decoupling control strategy includes a lateral leveling channel, a longitudinal posture stabilization channel, and a constant tillage depth channel; the three-degree-of-freedom coupling model and the collaborative decoupling control strategy are used as a multi-degree-of-freedom collaborative decoupling control architecture; the dynamic leveling control target of the agricultural tractor's body is determined based on the body tilt angle; the decoupling input items of the agricultural tractor are obtained by combining the three-degree-of-freedom coupling model and the body dynamic leveling control target; and the decoupling input items are analyzed according to the collaborative decoupling control strategy to obtain preliminary leveling control instructions.

[0029] The three-degree-of-freedom coupling model for an agricultural tractor uses the vehicle's lateral tilt angle, longitudinal pitch angle, and tillage depth displacement as the three-degree-of-freedom generalized coordinate vectors. Dynamic analysis establishes the dynamic relationship between these degrees of freedom. The lateral tilt angle reflects the degree of left-right tilt of the vehicle and is determined by the displacement difference between the left and right hydraulic cylinders. The longitudinal pitch angle represents the height difference between the front and rear of the vehicle and is related to the front and rear suspension load distribution and the terrain gradient. The tillage depth displacement describes the depth of the implement's penetration into the soil and is influenced by both the vehicle's posture and soil resistance. The model construction considers the following coupling relationships: When the tractor is operating on a side slope, lateral tilt causes a redistribution of wheel loads, which in turn causes a change in the longitudinal pitch angle. When adjusting the tillage depth, changes in the hydraulic cylinder force are transmitted through the vehicle frame to the front and rear suspensions, forming a multi-degree-of-freedom torque coupling. The dynamic equations are established as a three-degree-of-freedom coupling model using the Lagrangian dynamic equations, providing a foundation for subsequent control decoupling.

[0030] The above three-degree-of-freedom generalized coordinate vectors satisfy the following relationship: in, is the three-degree-of-freedom generalized coordinate vector, is the roll angle, is the pitch angle, is the tillage depth displacement, Indicates transpose.

[0031] The above Lagrangian dynamic equations satisfy the following relationship: in, is the mass matrix, is the three-degree-of-freedom generalized coordinate vector, is the second-order derivative of the three-degree-of-freedom generalized coordinate vector, is the Coriolis-centripetal force matrix, is the first-order derivative of the three-degree-of-freedom generalized coordinate vector, is the stiffness matrix, is the external force vector.

[0032] The collaborative decoupling control strategy employs a layered architecture, decomposing complex control tasks into three independent control channels and achieving parameter coupling through a coordination layer. The lateral leveling channel uses the vehicle's roll angle as its control target, generating pressure compensation commands for the left and right hydraulic cylinders through a controller. Its design must consider nonlinear characteristics and employ variable gain parameters to adapt to varying load conditions. The longitudinal attitude stabilization channel uses the pitch angle as its control target, combining vehicle speed and terrain gradient information to suppress vehicle nodding under braking or acceleration conditions through feedforward compensation. The constant tillage depth channel establishes a mapping relationship between tillage depth displacement and hydraulic system flow, dynamically adjusting tillage depth control parameters to accommodate changes in soil hardness. The three channels prioritize commands through a coordination layer. For example, under extreme side slope conditions, the lateral leveling channel can temporarily take over hydraulic system flow distribution, sacrificing some tillage depth accuracy to ensure vehicle stability.

[0033] The dynamic leveling control target of an agricultural tractor's vehicle body is set based on a combination of the real-time perception of the vehicle body tilt angle and the operating conditions. The roll and pitch angle data acquired by the data acquisition device is used to construct a three-dimensional spatial description of the vehicle body's posture, and the current tilt angle is compared with the preset safe operating threshold. The control targets need to be set in layers: the basic target is to restore the vehicle body tilt angle to within the safety threshold to ensure that the working plane of the agricultural machinery remains parallel to the ground; the advanced target needs to consider the consistency requirements of the tillage depth, and fit the optimal tillage depth-attitude mapping curve through historical operating data, so that the leveling process takes into account both stability and work quality. In addition, the control target incorporates a dynamic compensation mechanism to adjust the target value according to the tractor's driving speed and the rate of terrain change. For example, the leveling threshold is moderately relaxed during high-speed operation to avoid frequent movements, while the threshold is tightened during low-speed and fine operations to improve control accuracy.

[0034] The three-degree-of-freedom coupling model uses the vehicle's center of mass as the origin and establishes a dynamic coupling relationship between lateral leveling, longitudinal posture stabilization, and tilling depth control. Based on the current vehicle tilt angle, the model uses geometric relationships to calculate the required displacement compensation for each hydraulic outrigger. For example, when roll angle deviation is detected, the model decomposes the overall leveling requirement into independent extension and retraction commands for the left and right hydraulic cylinders. Simultaneously, considering the dynamic characteristics of the hydraulic system, the displacement requirements are converted into solenoid valve opening commands using the flow-pressure characteristic curve. A hierarchical control structure is employed to generate decoupling inputs: the bottom layer is the closed-loop position control of the hydraulic actuator, the middle layer is the three-degree-of-freedom dynamic decoupling compensation, and the top layer is the allocation of leveling targets based on the vehicle tilt angle. This effectively isolates the dynamic coupling effects between the various degrees of freedom, for example, avoiding additional interference with tilling depth control when adjusting the roll angle. Decoupling inputs include the generalized error vector, external disturbance feedforward terms, state constraint boundaries, and coupling matrix parameters.

[0035] The collaborative decoupling control strategy converts decoupled inputs into executable control commands through priority scheduling and energy optimization. The strategy first prioritizes lateral leveling, longitudinal stabilization, and tillage depth control channels. For example, in extreme tilt conditions, the lateral leveling channel is given the highest priority, suspending non-urgent tillage depth adjustments. Second, the strategy incorporates an energy management module to limit the total control variable based on hydraulic system pressure and oil temperature to prevent system overheating and energy consumption surges caused by excessive leveling. The parsing process utilizes fuzzy logic to map continuously varying decoupled inputs into discrete solenoid valve drive signals. For example, converting a hydraulic cylinder extension and retraction request into a 0%-100% duty cycle command. The resulting preliminary leveling control command includes the target displacement for each hydraulic outrigger, the solenoid valve switching sequence, and the time constant for the transition process. This ensures the coordinated multi-degree-of-freedom control while reducing control algorithm complexity through decoupling. For example, maintaining a constant tillage depth while maintaining vehicle levelness significantly improves operational accuracy and reliability in complex terrain.

[0036] S3. Obtain a terrain discrete elevation point set of the agricultural tractor, and construct a terrain elevation prediction model based on the terrain discrete elevation point set.

[0037] In this embodiment, the terrain is scanned based on the millimeter-wave radar to obtain an original point cloud set, which is then preprocessed to obtain a terrain discrete elevation point set; the terrain discrete elevation point set is compensated according to the body motion posture of the agricultural tractor to obtain an optimized terrain discrete elevation point set; based on the optimized terrain discrete elevation point set, a terrain elevation prediction model is constructed in combination with Gaussian regression.

[0038] Millimeter-wave radar is the core sensor, achieving non-contact scanning by emitting high-frequency electromagnetic waves and receiving reflected signals from the ground. Mounted on the front bumper or roof of the agricultural tractor, the radar covers a forward detection range of 10-15 meters at a 45° angle, ensuring sufficient terrain preview distance at operating speeds. During scanning, the radar generates a sector-shaped detection area using rotational scanning or phased array electronic scanning, generating a raw point cloud with each scanning cycle. This point cloud contains 3D coordinates and reflection intensity information. To increase point cloud density, an interleaved scanning pattern is employed: When the tractor is traveling in a straight line, the radar performs a full-coverage scan at a fixed angular resolution. During cornering, the radar adjusts the scanning sector priority based on trajectory prediction, prioritizing the inboard terrain data. Once the raw point cloud is generated, it undergoes spatiotemporal synchronization to unify the point cloud data acquired from different scanning cycles into the vehicle coordinate system, eliminating point cloud distortion caused by the tractor's motion.

[0039] Preprocessing the raw point cloud consists of three core steps: noise filtering, data registration, and vehicle motion compensation. First, an outlier removal algorithm based on statistical distribution is used to calculate the average distance within the neighborhood of each data point, eliminating outliers and effectively eliminating non-ground reflection interference such as vegetation and gravel. Second, a voxel grid filter is used to divide the point cloud into a three-dimensional grid, retaining the centroid point within each grid cell to achieve uniform data density while preserving terrain features. Vehicle motion compensation requires the integration of real-time tractor posture information. When the vehicle pitches or rolls, the angle between the radar scanning plane and the ground changes, leading to systematic deviations. The vehicle body's Euler angle data is acquired using a six-axis inertial measurement unit (IMU). A rotation matrix is established between the point cloud coordinate system and the geographic coordinate system, and coordinate transformation correction is performed on each data point. For example, when a 2° pitch angle is detected, the point cloud is projected onto a horizontal reference plane using a three-dimensional rotation transformation to eliminate terrain distortion caused by attitude changes, ultimately generating a high-precision, optimized terrain discrete elevation point set.

[0040] The above-mentioned vehicle body motion posture compensation includes vehicle body posture compensation and motion compensation.

[0041] The vehicle body posture compensation satisfies the following relationship: in, is the point cloud coordinate after compensation, is the pitch rotation matrix, is the roll rotation matrix, is the original point cloud coordinate, To compensate for the installation height of the millimeter wave radar, is the roll angle, is the pitch angle.

[0042] Motion compensation satisfies the following relationship: in, is the longitudinal coordinate of the point cloud after compensation, is the vertical coordinate of the original point cloud, Real-time speed of farm tractors, is the total system delay time.

[0043] The terrain elevation prediction model employs a Gaussian process regression framework, using an optimized set of discrete elevation points as training samples. First, spatial interpolation is performed on the point set to generate a regularly gridded terrain base. The model inputs are two-dimensional spatial coordinates, and the output is the elevation prediction for the corresponding location. The core of Gaussian regression lies in the selection of the covariance function. A linear combination of a squared exponential kernel function and a periodic kernel function is used to capture the continuous terrain variation while preserving local undulations. During training, kernel function hyperparameters, including the characteristic length scale, signal variance, and noise variance, are optimized using maximum likelihood estimation. To improve real-time performance, 5% of induction points are selected from the complete point set to construct a low-dimensional feature space. The resulting terrain elevation prediction model outputs elevation predictions and their confidence intervals for any location, providing advanced terrain information to the leveling system. For example, if a 0.15m terrain rise is predicted 3 meters ahead of the tractor, the model triggers pre-extension of the front axle hydraulic lever, adjusting the vehicle's posture to the optimal state in advance.

[0044] The above terrain elevation prediction model satisfies the following relationship: in, is the predicted value of terrain elevation, is the kernel function vector, is the kernel matrix, is the noise variance, is the identity matrix, is the elevation coordinate of the point cloud after compensation.

[0045] S4. Based on the preliminary leveling control instruction, use the terrain elevation prediction model to obtain an optimized leveling control instruction for the agricultural tractor.

[0046] In this embodiment, a short-term posture change prediction sequence of the agricultural tractor is obtained based on the terrain elevation prediction model; the posture of the agricultural tractor is pre-evolved based on the short-term posture change prediction sequence to obtain a time-domain posture evolution vector; a predicted compensation amount is obtained based on the time-domain posture evolution vector, and an optimized leveling control instruction is obtained by combining the preliminary leveling control instruction and the predicted compensation amount.

[0047] The terrain elevation prediction model generates a sequence of terrain elevation changes within the next 2-5 seconds by fusing millimeter-wave radar scanning data with vehicle body kinematic parameters in real time. It then outputs a predicted point cloud with a time step of 0.1 seconds, with each predicted point containing spatial coordinates and elevation values. Based on the tractor's current speed and steering angle information, the predicted terrain is mapped to the vehicle body coordinate system using a kinematic forward solution algorithm to generate a short-term posture change prediction sequence. For example, when a 0.25m terrain bump is predicted 3 meters ahead, the impact of this terrain feature on the vehicle body's roll angle, pitch angle, and center of mass height is calculated. The prediction sequence is dynamically updated using a sliding window mechanism, with the window length covering the tractor's current position to the predicted range ahead, ensuring that control commands are always generated based on the latest terrain information.

[0048] The above short-term posture change prediction sequence satisfies the following relationship: in, is the predicted change in roll angle, is the elevation-inclination conversion coefficient, is the predicted value of terrain elevation, is the predicted change in pitch angle.

[0049] Based on a short-term posture change prediction sequence, with the current vehicle body state as the initial condition, the posture state at each predicted time step is recursively calculated to obtain the time-domain posture evolution vector. The pre-evolution process considers the dynamic characteristics of the hydraulic actuator, such as the hydraulic cylinder response delay time and maximum extension and retraction speed, and converts the predicted terrain elevation changes into a target displacement sequence for the hydraulic outriggers. The time-domain posture evolution vector consists of three core components: the roll angle change rate, the pitch angle change rate, and the vertical displacement of the center of mass. Each component is estimated through a Kalman filter, fusing the prediction model output with real-time sensor feedback to effectively suppress model error accumulation. For example, on continuously bumpy roads, the time-domain vector can clearly reflect the periodic fluctuation characteristics of the vehicle body posture, providing a quantitative basis for optimizing control commands.

[0050] Based on the time-domain pose evolution vector, a fuzzy control algorithm is used to generate predicted compensations. This algorithm uses the roll angle change rate, pitch angle change rate, and center of mass displacement as input variables and maps them into a compensation space for the hydraulic outrigger extension and retraction. Compensation calculations consider two dimensions: amplitude compensation to offset attitude deviations caused by terrain undulations, and rate compensation to suppress sudden changes in vehicle posture. For example, when the terrain prediction indicates an impending roll angle acceleration of 0.3° / s, the compensation algorithm outputs a pre-extension of the left hydraulic cylinder, causing the vehicle to tilt in the opposite direction to offset the subsequent attitude change. The final optimized leveling control command is generated through a weighted fusion mechanism, combining the steady-state control variable of the preliminary command with the dynamic adjustment of the predicted compensation variable. The weight coefficient is dynamically adjusted based on the tractor's travel speed, increasing the compensation weight during low-speed precision operations and ensuring the baseline stability of the preliminary command during high-speed transfer conditions.

[0051] S5. Adjust the tilt angle of the vehicle body according to the optimized leveling control instruction to achieve leveling of the agricultural tractor body.

[0052] S5 specifically includes the following steps: S51 . Controlling the solenoid valve to input or output hydraulic oil according to the optimized leveling instruction to control the extension and retraction of the hydraulic rod to adjust the tilt angle of the vehicle body.

[0053] In this embodiment, the hydraulic actuator's control is centered around a solenoid valve assembly, utilizing high-precision proportional solenoid valves to achieve on / off switching and flow regulation of the hydraulic circuit. The optimized leveling control command is first parsed into the target pressure and flow requirements for each hydraulic leg. Based on the current feedback from the hydraulic system's pressure sensors, the central controller generates the solenoid valve drive signals using pulse-width modulation technology. The hydraulic rod's extension and retraction speed is controlled in a closed-loop manner by adjusting the solenoid valve opening area. A pressure-flow composite control strategy is employed: during the initial leveling phase, the system rapidly responds to command changes with maximum flow. When approaching the target posture, the system switches to pressure control mode, fine-tuning the valve opening to eliminate static errors. To address flow distribution issues during the coordinated motion of multiple hydraulic cylinders, the system employs a priority scheduling algorithm, dynamically allocating hydraulic pump output flow based on the posture contribution of each leg. Furthermore, the hydraulic actuator incorporates both mechanical limit stops and electronic buffering. When the hydraulic rod's travel is nearing its limit, the solenoid valve drive current is automatically reduced to prevent damage from rigid impact.

[0054] S52: During the control process, the body tilt angle is monitored in real time and dynamically intervened to achieve body leveling of the agricultural tractor.

[0055] In this embodiment, the body tilt angle is monitored in real time to obtain the rollover risk coefficient of the agricultural tractor; a hierarchical active protection strategy for the agricultural tractor is constructed based on the rollover risk coefficient; and the body tilt angle is dynamically intervened according to the hierarchical active protection strategy to achieve body leveling of the agricultural tractor.

[0056] During agricultural tractor operation, a dual-redundant monitoring system constructed from a six-axis inertial measurement unit and suspension point displacement sensors enables real-time sensing of the vehicle's tilt angle. Roll, pitch, and angular velocity data are monitored, while displacement sensors simultaneously capture changes in the extension and retraction of the hydraulic outriggers. The monitoring system uses a complementary filtering algorithm to fuse the data from these two sensors. Based on this fused data, the rollover risk factor is calculated, which satisfies the following relationship: in, is the capsizing risk factor, For the quality of agricultural tractors, is the acceleration due to gravity, is the centroid height, is the roll angle, is the speed of the agricultural tractor, is the turning radius of the agricultural tractor, is the soil coefficient, is the friction coefficient, is the normal load of the agricultural tractor.

[0057] In this embodiment, a dynamic threshold function is constructed, and a dynamic safety threshold is obtained based on the dynamic threshold function; the overturning hazard level of the agricultural tractor is obtained by combining the overturning hazard coefficient and the dynamic safety threshold, and the overturning hazard level includes a first-level warning, a second-level intervention, and a third-level protection; a hierarchical active protection strategy is constructed based on the overturning hazard level, and the hierarchical active protection strategy includes a first-level preventive posture stabilization strategy, a second-level active anti-overturning strategy, and a third-level emergency protection strategy.

[0058] First, a dynamic threshold function is constructed. This function integrates real-time operating parameters such as vehicle speed, load, and ground adhesion coefficient. Fuzzy logic reasoning is used to generate a dynamic safety threshold, replacing the traditional fixed threshold scheme and achieving adaptive adjustment of the safety margin. The system continuously calculates the current rollover risk factor and compares it with the dynamic safety threshold. When the rollover risk factor exceeds 80% of the dynamic safety threshold, a level 1 warning is triggered. At this time, a preventive attitude stabilization strategy is initiated, suppressing the trend of attitude deterioration by slightly adjusting the hydraulic leg extension and extension. If the rollover risk factor exceeds the dynamic safety threshold, the system enters level 2 intervention and executes an active anti-rollover strategy. The system takes over hydraulic control and implements forced attitude correction actions, while also limiting engine power output to prevent sudden power changes from exacerbating imbalances. If the rollover risk factor continues to rise to above 120% of the dynamic safety threshold, level 3 protection is immediately activated, initiating an emergency protection strategy, including hydraulic locking, engine shutdown protection, and seatbelt pretensioning. The hierarchical strategy achieves a smooth transition through dynamic weight allocation. For example, in the secondary intervention stage, the system performs active posture stabilization with a weight of 70% and prepares protective measures with a weight of 30%, ensuring the continuity and safety of control instructions, forming a complete protection chain from risk warning to emergency protection.

[0059] The above dynamic threshold function satisfies the following relationship: in, is the dynamic safety threshold, is the soil coefficient, is the ground adhesion coefficient, is the speed of the agricultural tractor, The real-time slope angle of the terrain.

[0060] In this embodiment, when dynamically intervening in the vehicle body tilt angle according to the graded active protection strategy, the system continuously monitors the vehicle body roll angle, pitch angle and hydraulic support leg displacement data, and calculates the rollover risk coefficient in real time. When the rollover risk coefficient exceeds the preset threshold, a three-level protection response mechanism is triggered: in the first-level warning stage, sound and light signals are used to remind the operator to pay attention to the vehicle body posture; in the second-level intervention stage, the opening of the hydraulic solenoid valve is automatically adjusted, and the extension and retraction of the high-risk axial hydraulic rod is controlled first. For example, when the rollover risk coefficient exceeds the limit, the oil supply to the low-side hydraulic cylinder is increased, and the movement amplitude of the opposite hydraulic rod is limited at the same time; in the third-level protection stage, if the risk coefficient continues to rise, the hydraulic locking function is activated and the engine output torque is limited to prevent the vehicle from substantially overturning. The entire intervention process continuously corrects the control amount through closed-loop control to achieve progressive leveling of the vehicle body posture while ensuring safety.

[0061] See Figure 2In an optional embodiment, the present invention provides a body leveling control system for an agricultural tractor. The system includes an input device, an output device, a processor, and a memory, wherein the hardware components are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to invoke the program instructions and execute the specific steps of the embodiments of the body leveling control method for an agricultural tractor provided by the present invention. The body leveling control system for an agricultural tractor provided by the present invention has a complete structure and is objectively stable, thereby enhancing the overall applicability and practical application capabilities of the present invention.

[0062] In summary, the method of the present invention provides a body leveling control method and system for agricultural tractors, which realize multi-parameter dynamic decoupling and joint optimization by constructing a multi-degree-of-freedom collaborative decoupling control architecture, a three-degree-of-freedom coupled dynamics model, and a three-channel collaborative control strategy; constructing a terrain elevation prediction model to generate a short-term posture change prediction sequence, driving the control command feedforward compensation, so that the leveling action is ahead of the terrain change; designing a hierarchical active protection mechanism to realize a three-level intervention strategy from preventive posture stabilization to emergency protection, upgrading passive protection to active risk management, and improving the body stability, operation accuracy and operation safety under complex terrain; the method of the present invention is easy to understand, simple to calculate, with a small workload, and convenient for engineering application, providing a theoretical basis and technical support for the further development of mechanical intelligent control technology.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A method for controlling the leveling of an agricultural tractor body, characterized in that: The steps include: Constructing a data acquisition device for an agricultural tractor, and obtaining a body tilt angle of the agricultural tractor according to the data acquisition device; Establishing a multi-degree-of-freedom collaborative decoupling control architecture and obtaining preliminary leveling control instructions for the agricultural tractor in combination with the vehicle body tilt angle; Acquiring a terrain discrete elevation point set of the agricultural tractor, and constructing a terrain elevation prediction model based on the terrain discrete elevation point set; Based on the preliminary leveling control instruction, obtaining an optimized leveling control instruction for the agricultural tractor using the terrain elevation prediction model; The body tilt angle is regulated according to the optimized leveling control instruction to achieve body leveling of the agricultural tractor.

2. The agricultural tractor body leveling control method according to claim 1, characterized in that: The method of constructing a data acquisition device for an agricultural tractor and obtaining the body tilt angle of the agricultural tractor according to the data acquisition device includes: Combining a wheel speed sensor, a six-axis inertial measurement unit and a suspension point displacement sensor to construct the data acquisition device; Acquiring wheel speed signals, angle data, and hydraulic cylinder extension and contraction amounts according to the data acquisition device to construct a state vector of the agricultural tractor; Constructing a variance-driven sliding window, and dynamically filtering the state vector according to the sliding window to obtain an optimized state vector; The tilt angle is calculated based on the optimized state vector to obtain the posture angle of the agricultural tractor, and the posture angle is used as the tilt angle of the vehicle body.

3. The agricultural tractor body leveling control method according to claim 1, characterized in that: The establishment of a multi-degree-of-freedom collaborative decoupling control architecture includes: Acquiring a three-degree-of-freedom generalized coordinate vector of the agricultural tractor, and constructing a three-degree-of-freedom coupling model of the agricultural tractor based on the three-degree-of-freedom generalized coordinate vector; Constructing a coordinated decoupling control strategy for the agricultural tractor, wherein the coordinated decoupling control strategy includes a lateral leveling channel, a longitudinal posture stabilization channel, and a plowing depth constant channel; The three-degree-of-freedom coupling model and the collaborative decoupling control strategy are used as the multi-degree-of-freedom collaborative decoupling control architecture.

4. The agricultural tractor body leveling control method according to claim 3, characterized in that: The obtaining of a preliminary leveling control instruction for the agricultural tractor in combination with the vehicle body tilt angle includes: determining a vehicle body dynamic leveling control target of the agricultural tractor based on the vehicle body tilt angle; Combining the three-degree-of-freedom coupling model and the vehicle body dynamic leveling control target to obtain a decoupling input item of the agricultural tractor; The decoupling input item is parsed according to the collaborative decoupling control strategy to obtain the preliminary leveling control instruction.

5. The agricultural tractor body leveling control method according to claim 1, characterized in that: The step of obtaining a terrain discrete elevation point set of the agricultural tractor and constructing a terrain elevation prediction model according to the terrain discrete elevation point set includes: Scanning the terrain based on the millimeter-wave radar to obtain an original point cloud set, and preprocessing the original point cloud set to obtain the terrain discrete elevation point set; Compensating the terrain discrete elevation point set according to the body motion posture of the agricultural tractor to obtain an optimized terrain discrete elevation point set; The terrain elevation prediction model is constructed based on the optimized terrain discrete elevation point set and combined with Gaussian regression.

6. The agricultural tractor body leveling control method according to claim 1, characterized in that: The method of obtaining the optimized leveling control instruction for the agricultural tractor based on the preliminary leveling control instruction and using the terrain elevation prediction model comprises: Obtaining a short-term posture change prediction sequence of the agricultural tractor according to the terrain elevation prediction model; Pre-evolving the posture of the agricultural tractor according to the short-term posture change prediction sequence to obtain a time-domain posture evolution vector; A predicted compensation amount is obtained based on the time-domain posture evolution vector, and the optimized leveling control instruction is obtained by combining the preliminary leveling control instruction and the predicted compensation amount.

7. The agricultural tractor body leveling control method according to claim 1, characterized in that: The step of regulating the vehicle body tilt angle according to the optimized leveling control instruction to achieve vehicle body leveling of the agricultural tractor includes: Controlling the solenoid valve to input or output hydraulic oil according to the optimized leveling instruction to control the extension and retraction of the hydraulic rod to adjust the tilt angle of the vehicle body; During the control process, the body tilt angle is monitored in real time and dynamically intervened to achieve body leveling of the agricultural tractor.

8. The agricultural tractor body leveling control method according to claim 7, characterized in that: The real-time monitoring and dynamic intervention of the vehicle body tilt angle to achieve vehicle body leveling of the agricultural tractor includes: Real-time monitoring of the vehicle body tilt angle to obtain the rollover risk coefficient of the agricultural tractor; Constructing a hierarchical active protection strategy for the agricultural tractor based on the rollover risk coefficient; The body leveling of the agricultural tractor is achieved by dynamically intervening in the body tilt angle according to the hierarchical active protection strategy.

9. The agricultural tractor body leveling control method according to claim 8, characterized in that: The step of constructing a hierarchical active protection strategy for the agricultural tractor based on the rollover risk coefficient includes: Constructing a dynamic threshold function, and obtaining a dynamic safety threshold based on the dynamic threshold function; Combining the rollover risk coefficient and the dynamic safety threshold to obtain a rollover risk level of the agricultural tractor, wherein the rollover risk level includes a first-level warning, a second-level intervention, and a third-level protection; The hierarchical active protection strategy is constructed based on the overturning hazard level, and the hierarchical active protection strategy includes a first-level preventive posture stabilization strategy, a second-level active anti-overturning strategy, and a third-level emergency protection strategy.

10. An agricultural tractor body leveling control system, characterized in that: The system includes an input device, an output device, a processor and a memory, wherein the input device, the output device, the processor and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the agricultural tractor body leveling control method according to any one of claims 1 to 9.

Images