Heavy load omnidirectional agv hydraulic floating suspension load balancing and attitude cooperative constraint control method

Abstract

A method for load balancing and attitude coordination constraint control of a heavy-duty omnidirectional AGV with hydraulic floating suspension is disclosed, relating to the field of general control or adjustment systems. The method includes: acquiring vehicle attitude signals, suspension cylinder pressure signals, and displacement signals; determining the actual vertical load and vehicle tilt angle data of each differential steering wheel; determining the load weight, attitude weight, and tilt angle constraint boundary corresponding to the operating conditions; based on a predictive controller, performing a rolling solution within the tilt angle constraint boundary with the objective of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command for each suspension cylinder; performing a positive correlation conversion of the actual vertical load with drive torque to determine the coordination torque command for each differential steering wheel; and adjusting the hydraulic pressure of each suspension cylinder and the output torque of each differential steering wheel based on the pressure adjustment command and the coordination torque command to control the operation of the heavy-duty omnidirectional AGV. Implementing this application can avoid safety risks caused by local overload.

Description

Technical Field

[0001] This application relates to the general field of control or regulation systems, and more particularly to a method for load balancing and attitude coordination constraint control of a heavy-duty omnidirectional AGV hydraulic floating suspension. Background Technology

[0002] Heavy-duty omnidirectional automated guided vehicles (AGVs) are core equipment in modern intelligent manufacturing and smart logistics systems, undertaking the flexible transfer of large-sized and heavy-weight materials. Their applications span aerospace, heavy machinery, mold manufacturing, and other fields. To adapt to the uneven ground conditions commonly found in industrial sites and to ensure the stability of heavy goods during transportation, these AGVs are typically equipped with multiple independent suspension systems.

[0003] In related technologies, a common suspension solution involves equipping each wheel with a hydraulic cylinder and connecting all cylinders to a common accumulator via pipelines, with one hydraulic chamber (e.g., the rodless chamber) in each cylinder. When the AGV is moving, if a wheel is passively lifted due to a road surface bump, the piston of its corresponding cylinder is compressed, and the oil pressure inside the cylinder increases accordingly. The high-pressure oil is then supplied to other cylinders with lower pressure through the connecting pipelines, thereby distributing and balancing the load at each suspension point to a certain extent.

[0004] However, under conditions where the center of gravity of the load is offset or the vehicle body is turning and generating overturning moment, the flow response of the oil inside the passive hydraulic system lags behind the instantaneous change of the load, which can lead to continuous overload at some suspension points, posing a certain safety risk. Summary of the Invention

[0005] This application provides a method for load balancing and attitude coordination constraint control of hydraulic floating suspension for heavy-duty omnidirectional AGVs, which is used to avoid safety risks caused by local overload.

[0006] Firstly, this application provides a method for load balancing and attitude coordination constraint control of a heavy-duty omnidirectional AGV with hydraulic floating suspension. The heavy-duty omnidirectional AGV is equipped with multiple sets of differential steering wheels and correspondingly connected suspension cylinders. This method is applied to a control system and includes: acquiring vehicle attitude signals, pressure signals and displacement signals of the suspension cylinders, and determining the actual vertical load and vehicle tilt angle data of each differential steering wheel; acquiring current operating condition characteristics and determining the load weight, attitude weight, and tilt angle constraint boundary corresponding to the operating condition characteristics; inputting the actual vertical load, vehicle tilt angle data, load weight, attitude weight, and tilt angle constraint boundary into a predictive controller, and performing a rolling solution within the tilt angle constraint boundary with the objective of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command of each suspension cylinder; performing a positive correlation conversion of the actual vertical load to determine the coordination torque command of each differential steering wheel; and adjusting the hydraulic pressure of each suspension cylinder and the output torque of each differential steering wheel based on the pressure adjustment command and the coordination torque command to control the operation of the heavy-duty omnidirectional AGV.

[0007] In the above embodiments, the control system integrates pressure regulation commands and coordinated torque commands to synchronously regulate each suspension cylinder and each differential steering wheel, balancing the vertical load of each suspension point within the tilt angle constraint boundary, and avoiding structural damage or operational instability caused by local overload.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the step of acquiring vehicle attitude signals, suspension cylinder pressure signals, and displacement signals, and determining the actual vertical load and vehicle tilt angle data of each differential steering wheel specifically includes: acquiring vehicle attitude signals, suspension cylinder pressure signals, and displacement signals; calculating the instantaneous rate of change of the displacement signal; when the instantaneous rate of change exceeds the excitation threshold, performing dynamic decoupling of the pressure signal using a sliding mode observer to obtain the static off-center load component and the dynamic impact component; and determining the actual vertical load and vehicle tilt angle data of each differential steering wheel based on the static off-center load component and the vehicle attitude signal.

[0009] In the above embodiments, the control system uses the instantaneous rate of change of the displacement signal to determine the disturbance state, and uses a sliding mode observer to decouple the pressure signal into a static off-center load component and a dynamic impact component, so that the load calculation eliminates transient impact interference and improves the accuracy of the actual vertical load data.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, the step of performing a rolling solution within the tilt angle constraint boundary with the objective of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command of each suspension cylinder specifically includes: obtaining the frame stiffness parameters of the heavy-duty omnidirectional AGV, calculating the frame torsional deformation in combination with the actual vertical load; determining the load deviation tolerance of the frame torsional deformation according to a preset torsional fatigue limit; and performing a rolling solution within the tilt angle constraint boundary and the load deviation tolerance to determine the pressure adjustment command of each suspension cylinder.

[0011] In the above embodiments, the control system introduces frame stiffness parameters and torsional fatigue limit to determine the load deviation tolerance. Under the dual constraints of tilt angle constraint and tolerance, rolling solution is performed to prevent the frame from being subjected to excessive torsional stress due to over-balancing load, thus ensuring the safety of the frame structure.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the step of performing a rolling solution within the tilt angle constraint boundary and load deviation tolerance to determine the pressure adjustment command of each suspension cylinder specifically includes: extracting the variance data of the actual vertical load during the rolling solution process; freezing the current rolling solution cycle and outputting a zero-increment adjustment signal when the variance data falls within the load deviation tolerance and the vehicle tilt angle data is within the tilt angle constraint boundary; and using the zero-increment adjustment signal as the pressure adjustment command of each suspension cylinder.

[0013] In the above embodiments, the control system freezes the solution and outputs a zero-increment signal when the load variance falls within the tolerance and the tilt angle meets the constraints, thus avoiding redundant adjustments when the load is already within an acceptable range and reducing the energy consumption of the hydraulic system and the wear of valve components.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, the step of performing a positive correlation calculation of the driving torque on the actual vertical load to determine the cooperative torque command of each differential steering wheel specifically includes: performing a positive correlation calculation of the driving torque on the actual vertical load to obtain an initial torque distribution sequence; obtaining the real-time stator temperature of the motor of each differential steering wheel to determine the torque attenuation coefficient of each differential steering wheel; and determining the cooperative torque command of each differential steering wheel based on the initial torque distribution sequence and the torque attenuation coefficient.

[0015] In the above embodiment, the control system performs positive correlation torque conversion based on the actual vertical load and determines the attenuation coefficient in combination with the stator temperature, so that the output torque of each differential steering wheel matches its load capacity, while protecting the high-temperature motor windings from thermal damage.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, the step of determining the cooperative torque command of each differential steering wheel based on the initial torque distribution sequence and the torque attenuation coefficient specifically includes: calculating an initial torque sequence based on the initial torque distribution sequence and the torque attenuation coefficient; calculating the total vehicle torque deficit based on the initial torque sequence and the initial torque distribution sequence; and proportionally compensating the total vehicle torque deficit into the initial torque of the remaining differential steering wheels whose stator temperature is below a temperature threshold, thereby determining the cooperative torque command of each differential steering wheel.

[0017] In the above embodiment, the control system compensates the vehicle torque loss caused by thermal decay to the low-temperature motor in proportion, so that the total driving torque of the vehicle remains unchanged, thus protecting the high-temperature motor while maintaining the integrity of the vehicle's driving capability.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, the step of obtaining the current operating condition characteristics and determining the load weight, attitude weight, and tilt constraint boundary corresponding to the operating condition characteristics specifically includes: obtaining vehicle tilt angle data and actual vertical load within a continuous time window; extracting the low-frequency drift rate of the vehicle tilt angle data and the step amplitude of the actual vertical load to determine the current operating condition characteristics; and performing interpolation optimization in a preset multidimensional weight matrix based on the operating condition characteristics to determine the load weight, attitude weight, and tilt constraint boundary corresponding to the operating condition characteristics.

[0019] In the above embodiments, the control system identifies the operating conditions by extracting the low-frequency drift rate of the tilt angle and the load step amplitude, and performs interpolation optimization in the multi-dimensional weight matrix so that the load weight, attitude weight and tilt angle constraint boundary are adapted to the operating conditions in real time.

[0020] In a second aspect, embodiments of this application provide a control system comprising: one or more processors and a memory; the memory being coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a control system, cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a control system, cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the control system provided in the second aspect, the computer storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the methods provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. By acquiring vehicle attitude signals and pressure and displacement signals from suspension cylinders to determine the actual vertical load and vehicle tilt angle data of each differential steering wheel, and minimizing vertical load deviation within the tilt angle constraint boundary, the predictive controller executes rolling solution to determine the pressure adjustment command of each suspension cylinder. Simultaneously, it performs positive correlation calculation of the actual vertical load to determine the coordinated torque command of each differential steering wheel. The predictive controller optimizes within the constraint domain based on load weight and attitude weight, so that the two objectives of load balance and attitude stability are simultaneously considered. Therefore, the hydraulic pressure of each suspension cylinder and the output torque of each differential steering wheel can be coordinated, effectively solving the problem of continuous overload of some suspension points due to the lag in oil flow response in the passive hydraulic suspension in the prior art. Thus, it realizes the coordinated control of load balance and attitude stability of heavy-duty omnidirectional AGV under off-center load and steering conditions.

[0026] 2. By employing a technique that combines the acquisition of frame stiffness parameters with the calculation of frame torsional deformation under actual vertical load, and determining the load deviation tolerance based on the torsional fatigue limit, and performing rolling solutions within the tilt angle constraint boundary and load deviation tolerance, the load deviation tolerance incorporates the frame structural strength into the optimization constraint. Therefore, the pressure adjustment command promotes load balance without causing excessive torsional deformation of the frame. This effectively solves the problem in existing technologies where blindly pursuing complete load balance without considering the frame torsional characteristics may cause frame fatigue damage, thereby achieving optimized load distribution while ensuring the safety of the frame structure.

[0027] 3. By employing a technique that uses the actual vertical load to perform a positive correlation calculation of the driving torque to obtain the initial torque distribution sequence, and by obtaining the real-time stator temperature of the motor of each differential steering wheel to determine the torque attenuation coefficient, the torque distribution is proportional to the vertical load, ensuring that the driving force of each wheel matches its normal force. The torque attenuation coefficient allows the high-temperature motor to appropriately reduce its output. Therefore, the output torque of each differential steering wheel takes into account both load matching and motor thermal protection, effectively solving the problems of light-load wheel slippage and high-temperature motor overload operation caused by uniform distribution of driving torque in existing technologies. This achieves the synergy of load-adaptive torque distribution and motor thermal protection. Attached Figure Description

[0028] Figure 1This is a schematic diagram of the hardware architecture of the control system in an embodiment of this application;

[0029] Figure 2 This is a schematic diagram of the control module architecture of the control system in an embodiment of this application;

[0030] Figure 3 This is a flowchart illustrating the load balancing and attitude coordination constraint control method for heavy-duty omnidirectional AGV hydraulic floating suspension in the embodiments of this application.

[0031] Figure 4 This is another flowchart illustrating the method for load balancing and attitude coordination constraint control of heavy-duty omnidirectional AGV hydraulic floating suspension in the embodiments of this application;

[0032] Figure 5 This is a schematic diagram of the physical device structure of a control system in an embodiment of this application. Detailed Implementation

[0033] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0034] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0035] Heavy-duty omnidirectional AGVs refer to automated guided vehicles with omnidirectional mobility and rated load capacity ranging from several tons to tens of tons. These types of equipment undertake flexible logistics tasks for large-sized workpieces in scenarios such as aerospace component transfer, heavy mold handling, and ship section docking.

[0036] A differential steering wheel is a drive unit in which two coaxially arranged drive motors drive the same steering wheel group through a differential mechanism. This drive unit achieves steering function by adjusting the speed difference between the two motors, and at the same time provides driving force with the average speed of the two motors. Each differential steering wheel acts as a drive actuator to output horizontal driving force, and also acts as an independent suspension support point to bear the vertical load from the frame and cargo.

[0037] Hydraulic floating suspension refers to a suspension structure that uses hydraulic cylinders as elastic elements to connect the vehicle frame and the differential steering wheel mounting base. This structure changes the support force and height of each suspension point by adjusting the pressure and flow rate of the hydraulic oil in the cylinder cavity.

[0038] In traditional passive hydraulic suspension systems, a specific oil chamber in each cylinder is interconnected via pipelines and connected to an accumulator. The passive flow of hydraulic fluid between the cylinders achieves a limited balance of load. However, under conditions such as cargo center of gravity shift, uneven road surfaces, or overturning moments generated by vehicle steering, the response speed of this passive flow method is insufficient to keep up with instantaneous load changes. This can cause some suspension points to experience continuous overload, posing a risk to the vehicle frame structure and transportation safety.

[0039] A predictive controller is an optimization module built on the principle of model predictive control. This module predicts the system state over a finite time domain in each control cycle and solves for the optimal control sequence that minimizes the objective function under given constraints. The first element of this sequence is used as the control output for the current cycle, and the prediction and solution are repeated in the next cycle, forming a rolling optimization mechanism. Load weights and attitude weights are used to quantify the relative importance of load balancing and attitude stabilization objectives in the optimization objective function, respectively. Their ratio determines the bias of the control system between load balancing and attitude maintenance. The roll and pitch constraint boundaries refer to the allowable range of variation for the vehicle's roll and pitch angles. This range is determined based on operating conditions and safety criteria and serves as a hard constraint for the predictive controller in the rolling solution.

[0040] Given the aforementioned equipment configuration and control requirements, a control method is needed that can integrate multi-source sensor signals and coordinate the joint actions of the hydraulic suspension system and drive system within the constraint domain, so as to achieve coordinated protection of load balance and attitude stability under different operating conditions.

[0041] Please see Figure 1 This is a schematic diagram of the hardware architecture of the control system in an embodiment of this application. Figure 1 The diagram shows a first suspension group 10 and a second suspension group 11 connected to a controller 12 via a signal path. A hydraulic pump station 13 provides system oil supply to each suspension group. The first suspension group 10 includes a first pressure sensor 101, a first displacement sensor 102, a first proportional valve 103, and a first suspension cylinder 104. The first pressure sensor 101 detects the chamber pressure of the first suspension cylinder 104 and transmits the pressure signal to the controller 12. The first displacement sensor 102 detects the extension and retraction of the piston rod of the first suspension cylinder 104 and transmits the displacement signal to the controller 12.

[0042] The controller 12 generates a pressure regulation command based on the received pressure and displacement signals and transmits the command to the first proportional valve 103. The first proportional valve 103 receives system oil from the hydraulic pump station 13 and regulates the flow rate and pressure of the hydraulic oil flowing to the first suspension cylinder 104 according to the pressure regulation command. The second suspension group 11 is symmetrical in structure and connection to the first suspension group 10, including a second displacement sensor 111, a second pressure sensor 112, a second proportional valve 113, and a second suspension cylinder 114. The signal transmission relationships between each sensor and the controller 12, as well as the oil supply and command relationships between each proportional valve and the hydraulic pump station 13 and the controller 12, are consistent with those of the first suspension group 10. In practical applications, heavy-duty omnidirectional AGVs can be configured with more than two suspension groups. Figure 1 The following diagram illustrates two sets of suspension groups.

[0043] Please see Figure 2 This is a schematic diagram of the control module architecture of the control system in an embodiment of this application. Figure 2 The diagram shows a signal acquisition module 201, a state decoupling module 202, a working condition feature extraction module 203, a predictive control core module 204, a torque collaborative distribution module 205, and a hardware execution module 206.

[0044] The signal acquisition module 201 is used to acquire basic signals such as pressure signals, displacement signals, and vehicle attitude signals from each suspension cylinder, and simultaneously acquires the stator temperature of the motors of each differential steering wheel. The state decoupling module 202 receives the basic signals output by the signal acquisition module 201, separates the load and tilt angle data, and transmits them to the operating condition feature extraction module 203 and the predictive control core module 204, respectively. The operating condition feature extraction module 203 determines the current operating condition characteristics based on the load and tilt angle data and outputs the corresponding weights and constraint boundaries to the predictive control core module 204. The predictive control core module 204 internally contains three sub-units: a stiffness tolerance calculation unit 2041, a rolling solution unit 2042, and a zero-increment freeze unit 2043. This module performs a rolling solution on the load deviation under the constraints of the weights and constraint boundaries, and outputs pressure adjustment commands to the hardware execution module 206. The torque coordination distribution module 205 internally comprises two sub-units: an initial torque conversion unit 2051 and a thermal derating and compensation unit 2052. This module receives load data from the state decoupling module 202 and stator temperature data from the signal acquisition module 201, and outputs a coordinated torque command to the hardware execution module 206. The hardware execution module 206 drives each proportional valve to adjust the hydraulic pressure of the suspension cylinders according to the pressure adjustment command, and simultaneously drives the motors of each differential steering wheel to output the corresponding torque according to the coordinated torque command.

[0045] The functional logic and data processing procedures of each of the above modules will be explained in detail in the subsequent steps.

[0046] The method provided in this embodiment is described in detail below. Please refer to [link / reference]. Figure 3 This is a flowchart illustrating a method for load balancing and attitude coordination constraint control of a heavy-duty omnidirectional AGV hydraulic floating suspension in an embodiment of this application.

[0047] S301. Acquire vehicle body attitude signals, suspension cylinder pressure signals and displacement signals to determine the actual vertical load and vehicle body tilt angle data of each differential steering wheel.

[0048] Among them, vehicle attitude signal refers to multi-axis attitude angle data collected by inertial measurement unit mounted on the chassis. Pressure signal refers to the real-time hydraulic pressure value output by the pressure sensor of each suspension cylinder. Displacement signal refers to the piston rod extension / retraction amount output by the displacement sensor of each suspension cylinder. Actual vertical load refers to the ground normal reaction force at each differential steering wheel, calculated based on the pressure signal and the effective pressure area of ​​the cylinder. Vehicle roll angle data refers to the roll angle and pitch angle extracted from the vehicle attitude signal.

[0049] Specifically, the control system synchronously acquires pressure and displacement signals from each suspension cylinder, as well as vehicle attitude signals, at a preset sampling period. The control system multiplies the pressure value of each suspension cylinder by its effective pressure-bearing area and, after correcting for the cosine component of the cylinder's mounting angle, calculates the actual vertical load on each differential steering wheel. Simultaneously, the control system extracts roll and pitch angles from the vehicle attitude signals as vehicle tilt angle data. This step corresponds to... Figure 2 The functions of the signal acquisition module 201 and the state decoupling module 202.

[0050] In some embodiments, the above signals can be acquired and load data converted in multiple ways: Optionally, the control system polls each sensor node at a fixed period via the CAN bus, reads the pressure digital quantity converted by A / D and the LVDT displacement output signal, applies a low-pass filter to the pressure value and multiplies it by the effective area of ​​the cylinder to obtain the vertical load, and reads the attitude angle filtered by Kalman from the inertial measurement unit as the tilt angle data; Optionally, the control system acquires the raw data of all sensors in parallel with microsecond-level synchronous precision via the EtherCAT bus, applies a Butterworth filter to the pressure signal to eliminate pipeline pulsation interference and performs load conversion, and outputs the tilt angle data after fusing the accelerometer and gyroscope data through complementary filtering of the attitude signal. It is understood that other communication protocols or filtering methods can also be used to acquire and process the above signals, which are not limited here.

[0051] S302. Obtain the current operating condition characteristics and determine the load weight, attitude weight, and tilt constraint boundary corresponding to the operating condition characteristics.

[0052] Among them, operating condition characteristics refer to the feature descriptions reflecting the current operating state type of the heavy-duty omnidirectional AGV, such as straight-line constant speed travel, cornering, stationary rotation, or slope travel. Load weight refers to the numerical coefficients assigned to the load balance objective in the objective function of the predictive controller. Attitude weight refers to the numerical coefficients assigned to the attitude stability objective. Tilt angle constraint boundaries refer to the upper and lower limits of the allowable roll angle and pitch angle of the vehicle body.

[0053] Specifically, the control system determines the characteristics of the operating condition based on the currently collected sensor data and operating commands, and looks up the corresponding load weights, attitude weights, and tilt angle constraint boundaries in a pre-calibrated parameter table. Different operating conditions have different requirements for load balance and attitude stability: for example, in cornering, the vehicle body is prone to tilting due to centrifugal force, requiring increased attitude weights and narrower tilt angle constraint boundaries; in straight-line constant-speed driving, the load weights can be relatively increased to emphasize load balance. This step corresponds to... Figure 2 Functions of the medium-duty condition feature extraction module 203.

[0054] In some embodiments, the operating condition characteristics and corresponding parameters can be determined in multiple ways: Optionally, the control system, based on parameters such as curvature and speed in the path planning instructions issued by the upper-level scheduling system, combined with the current vehicle attitude signal and actual vertical load, matches the operating condition type using preset rule criteria, and then reads the corresponding weights and constraint boundaries from the parameter lookup table; Optionally, the control system performs frequency domain analysis on the tilt angle data and load data within the continuous sampling window, determines the operating condition type based on the distribution pattern of the spectral characteristics using a pattern recognition algorithm, and then looks up the corresponding value in the preset parameter table. It is understood that other operating condition identification methods and parameter determination methods can also be used, and are not limited here.

[0055] S303. Input the actual vertical load, vehicle tilt angle data, load weight, attitude weight and tilt angle constraint boundary into the predictive controller. Within the tilt angle constraint boundary, perform a rolling solution with the goal of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command of each suspension cylinder.

[0056] The predictive controller refers to the optimization solution module established based on the principle of model predictive control. Vertical load deviation refers to the difference between the actual vertical load of each differential steering wheel and the target mean.

[0057] Specifically, the control system uses the actual vertical load and vehicle tilt angle data obtained in step S301, as well as the load weights, attitude weights, and tilt angle constraint boundaries obtained in step S302, as inputs to the predictive controller. The predictive controller constructs an objective function using the sum of squared vertical load deviations weighted by the load weights and the sum of squared tilt angle deviations weighted by the attitude weights, and uses the tilt angle constraint boundaries as hard constraints. It performs rolling optimization on the pressure adjustment of each suspension cylinder within the prediction time domain, taking the first element of the optimal control sequence as the pressure adjustment command for the current cycle. This step corresponds to... Figure 2 The function of the rolling solver unit 2042 within the predictive control core module 204.

[0058] When performing rolling solution within the tilt angle constraint boundary, the control system constructs a rolling optimization objective function with hard constraints of inequalities, the specific mathematical expression of which is as follows:

[0059]

[0060] The predictive controller performs rolling optimization within each control cycle with the objective of minimizing the cost function J. In the formula, i represents the numbers of the n sets of differential steering wheels and suspension cylinders (i=1,2,...,n); This represents the actual vertical load on the currently observed i-th differential steering wheel; This represents the average current vertical load of all n steering wheels; and These are the load weights and attitude weights issued by the pre-condition identification module, respectively. and Represents the actual roll and pitch angles of the vehicle body.

[0061] Meanwhile, the controller solution process is strictly limited to (subject to) the feasible region of the three types of inequality hard constraints:

[0062] The first type is attitude safety boundaries, which require the prediction of the roll angle. With pitch angle The absolute values ​​must be less than or equal to the allowable tilt angle constraint boundary under the current working condition. and ;

[0063] The second type is the hardware load limit boundary, which requires the load distribution of any output to be... It must be within the minimum physical limit load allowed by the suspension system. With the maximum physical limit load Between these points, prevent the hydraulic cylinder from bottoming out or bursting;

[0064] The third category is the actuator adjustment limit boundary, which requires the control input quantity to be sent to the proportional pressure reducing valve of the i-th suspension cylinder. The valve must be at its minimum permissible opening. With maximum opening between.

[0065] When all the above physical and attitude hard constraints are met, the controller uses a sequential solution algorithm to find the condition that allows... (Absolute load deviation) and The optimal control sequence that minimizes the weighted sum of (absolute attitude deviation) given the weights. The first sequence value is then used as the current pressure adjustment command for issuance.

[0066] It should be noted that before constructing the rolling optimization problem, the predictive controller needs to first establish a discrete state-space model describing the dynamic behavior of the hydraulic suspension system. The control system uses the chamber pressure values ​​of each suspension cylinder and the roll and pitch angles of the vehicle body to form a state vector, and uses the proportional valve control inputs corresponding to each suspension cylinder to form a control vector.

[0067] Within each discrete sampling period, the system's state at the next moment is determined by both the current state and the current control input. For the cavity pressure of each suspension cylinder, its dynamic evolution follows the volumetric elastic modulus equation of the hydraulic fluid: when the proportional valve opening increases, allowing more high-pressure fluid to flow into the cylinder cavity, the effective fluid volume within the cavity increases. Under the volume constraint of the cylinder's sealed cavity, the cavity pressure rises accordingly, with the rate of increase being directly proportional to the volumetric elastic modulus of the fluid and inversely proportional to the current remaining volume of the cavity. When the proportional valve opening decreases or closes, the fluid compressed by the supporting load in the cylinder cavity flows back to the low-pressure side, and the cavity pressure decreases accordingly.

[0068] The control system discretizes the aforementioned continuous-time-domain hydraulic dynamics equations at the control cycle using the forward Euler method or a zero-order hold, obtaining a discrete recursive relationship for the pressure state. For the vehicle body attitude angle, its variation depends on the spatial distribution of the support forces at each suspension point: when the vertical support force formed by multiplying the cavity pressure of each suspension cylinder by the effective pressure area is unevenly distributed on the frame, the torque imbalance will drive the vehicle body to rotate around the roll and pitch axes. The angular acceleration of this rotation is determined by dividing the resultant torque by the moment of inertia of the vehicle body about the corresponding axis. Based on the geometric position of each suspension point in the frame coordinate system and the corresponding vertical support force, the control system calculates the resultant torque around the roll and pitch axes, and establishes a second-order dynamic equation for the attitude angle by combining it with the vehicle body's moment of inertia. After discretization, this equation is incorporated into the attitude state recursive relationship of the state-space model.

[0069] After establishing the state-space model, the predictive controller uses the currently sampled actual state as the initial value in each control cycle. Using this model, it progressively extrapolates the state trajectory at each time step along the prediction time domain. The objective function values ​​for each time step are accumulated along this extrapolated trajectory, and the constraints at each time step are checked. This transforms the rolling optimization problem into a constrained numerical optimization problem in a finite prediction time domain. The length of the prediction time domain determines the controller's look-ahead capability, and its value must balance the response time constant of the hydraulic system and the real-time computation requirements.

[0070] For hydraulic suspension systems, the pressure build-up and release process in the cylinder chamber is limited by the flow characteristics of the proportional valve and the compressibility of the hydraulic fluid. The typical pressure response time constant is on the order of tens to hundreds of milliseconds. Therefore, the prediction time domain is usually taken as several times the length of this response time constant. The length of the control time domain does not exceed the prediction time domain, which is used to limit the dimensionality of the optimization variables to reduce the computational load.

[0071] For example, with a control cycle of 20 milliseconds, a prediction time domain of 10 steps (i.e., predicting forward for 200 milliseconds), and a control time domain of 3 steps, the predictive controller solves for the proportional valve control input sequence of 3 time steps in each cycle, so that the state prediction trajectory in the next 10 time steps satisfies the constraints and the cumulative value of the objective function is minimized. Then, only the first element of the sequence is used as the actual control output of the current cycle. In the next cycle, the state is resampled and the solution is re-executed to form a rolling optimization closed loop.

[0072] S304. Perform positive correlation conversion of drive torque on the actual vertical load to determine the coordinated torque command of each differential steering wheel.

[0073] Among them, the positive correlation conversion of drive torque refers to setting the drive torque distribution value of each differential steering wheel to a mapping relationship that is proportional to its corresponding actual vertical load. The cooperative torque command refers to the target torque value output by the control system to the drive motor controller of each differential steering wheel.

[0074] Specifically, after determining the actual vertical load in step S301, the control system uses the ratio of the actual vertical load of each differential wheel to the total vertical load of the vehicle as a torque distribution coefficient. The total required torque of the vehicle is then distributed to each differential wheel according to this ratio, resulting in a coordinated torque command for each differential wheel. This distribution method assigns a larger drive torque to the differential wheel bearing a larger vertical load, matching the upper limit of tire friction and reducing the risk of wheel slippage under light load. This step corresponds to... Figure 2 The function of the initial torque conversion unit 2051 within the torque coordination distribution module 205.

[0075] In some embodiments, the above-mentioned positive correlation conversion can be performed in several ways: Optionally, the control system calculates the percentage of the actual vertical load of each differential steering wheel to the total load, multiplies the total required torque of the vehicle by this percentage to obtain the torque distribution value for each wheel, and then applies a torque change rate limit to prevent sudden output changes; Optionally, the control system estimates the maximum available driving force of each wheel based on the actual vertical load of each differential steering wheel and a preset tire-ground friction coefficient, and distributes the total required torque proportionally with this maximum available driving force as the upper limit, ensuring that the distribution value does not exceed the adhesion limit of each wheel. It is understood that other positive correlation mapping methods can also be used to achieve torque distribution, which are not limited here.

[0076] S305, based on pressure regulation commands and coordinated torque commands, adjusts the hydraulic pressure of each suspension cylinder and the output torque of each differential steering wheel to control the operation of the heavy-duty omnidirectional AGV.

[0077] Specifically, after generating pressure regulation and coordinated torque commands in steps S303 and S304 respectively, the control system converts the pressure regulation command into control current signals for each proportional valve, driving each proportional valve to adjust the chamber pressure of the corresponding suspension cylinder, thus redistributing the support force at each suspension point according to the optimization result. Simultaneously, the control system sends the coordinated torque command to the motor controllers of each differential steering wheel via the communication bus, and each motor controller adjusts the drive current according to the received torque target value. Pressure regulation and torque regulation are executed in parallel, jointly controlling the operating status of the heavy-duty omnidirectional AGV. This step corresponds to... Figure 2 The function of hardware execution module 206.

[0078] In some embodiments, the above-mentioned parallel regulation can be performed in several ways: Optionally, the control system synchronously sends pressure regulation commands and coordinated torque commands along the clock edge of the same control cycle. After receiving the commands, each proportional valve and each motor controller independently executes its own closed-loop regulation. The proportional valve tracks the target pressure value through its built-in pressure closed-loop controller, and the motor controller tracks the target torque value through a vector control algorithm. Optionally, the control system adopts a time-sharing execution strategy, first sending the pressure regulation command and waiting for the pressure feedback of each suspension cylinder to approach the target value before sending the coordinated torque command to each motor controller. This method is suitable for situations where the response speed of the hydraulic system differs significantly from that of the drive system. It is understood that other regulation strategies can also be used to achieve the above-mentioned parallel or time-sharing control, which are not limited here.

[0079] In the method flow described in steps S301 to S305 of the above embodiment, the predictive controller performs a rolling solution within the tilt angle constraint boundary with the goal of minimizing the vertical load deviation, but the solution process does not take into account the structural strength characteristics of the frame itself within the constraint range.

[0080] Under heavy load conditions, the adjustment of load differences at each suspension point will cause the redistribution of support reaction forces at different positions of the frame. If the pressure adjustment command causes the torsional deformation of the frame to exceed the material fatigue limit, structural damage may accumulate during long-term operation.

[0081] Meanwhile, in the above embodiments, the torque distribution of each differential steering wheel is calculated based only on the proportional relationship of the actual vertical load, without taking into account the real-time thermal state of the drive motor. When the stator temperature of a certain motor rises due to continuous high load operation, continuing to distribute torque according to the original ratio may cause the motor to be in a thermal overload state.

[0082] To address the shortcomings mentioned above, the following embodiments introduce frame stiffness parameters and torsional fatigue limits during the generation of pressure regulation commands to determine the load deviation tolerance, enabling the rolling solution to be executed under the dual constraints of tilt angle and structural constraints; and introduce motor stator temperature during the generation of coordinated torque commands to determine the torque attenuation coefficient, so that torque distribution takes into account both load matching and motor thermal protection.

[0083] The method provided in this embodiment will now be described in more detail. Please refer to [link / reference]. Figure 4 This is another flowchart illustrating the load balancing and attitude coordination constraint control method for heavy-duty omnidirectional AGV hydraulic floating suspension in this application embodiment.

[0084] S401. Acquire vehicle body attitude signals, suspension cylinder pressure signals and displacement signals to determine the actual vertical load and vehicle body tilt angle data of each differential steering wheel.

[0085] Refer to step S301, which will not be repeated here.

[0086] In some embodiments, the control system acquires vehicle attitude signals, suspension cylinder pressure signals, and displacement signals, and calculates the instantaneous rate of change of the displacement signals. When the instantaneous rate of change exceeds the excitation threshold, the pressure signals are dynamically decoupled using a sliding mode observer to obtain static off-center load components and dynamic impact components. Based on the static off-center load components and vehicle attitude signals, the actual vertical load and vehicle tilt angle data of each differential steering wheel are determined.

[0087] The instantaneous rate of change refers to the velocity dimension value obtained by dividing the difference in displacement signal between adjacent sampling periods by the sampling period, used to represent the extension and retraction speed of the suspension cylinder piston rod. The excitation threshold is a pre-calibrated critical value for the rate of change of displacement; when the instantaneous rate of change exceeds this value, the suspension cylinder is determined to be subjected to road surface excitation. The sliding mode observer is a state observer constructed based on sliding mode variable structure control theory. This observer uses a switching term generated by a sign function to forcibly constrain the system state to the sliding surface. On this sliding surface, the pressure signal is frequency-divided to separate the low-frequency static off-center load component from the high-frequency dynamic impact component. The static off-center load component refers to the pressure component reflecting the continuous load level of each suspension point after removing transient impacts. The dynamic impact component refers to the transient pressure fluctuation component caused by road surface excitation.

[0088] Specifically, the control system calculates the instantaneous rate of change of the displacement signal of each suspension cylinder in each sampling period and compares this rate of change with a preset excitation threshold. When the instantaneous rate of change of all suspension cylinders is lower than the excitation threshold, the control system determines that the current road surface condition is stable and directly calculates the vertical load according to the conventional method in step S401. When the instantaneous rate of change of any suspension cylinder exceeds the excitation threshold, the control system determines that there is road excitation and activates the sliding mode observer to perform dynamic decoupling for the pressure signals of the suspension cylinder and the adjacent suspension cylinders affected by coupling. The sliding mode observer uses the pressure signal and displacement signal as input, constructs a system model using the force balance equation of the hydraulic cylinder and the volumetric elastic modulus equation of the hydraulic fluid, and adjusts the switching gain to make the observer state converge to the sliding surface within a finite time, extracting the low-frequency static off-center load component on the sliding surface. The control system uses this static off-center load component to replace the original pressure signal in the vertical load conversion, while retaining the dynamic impact component for subsequent road surface condition assessment.

[0089] It should be noted that the system model of the sliding mode observer is constructed based on the mechanical and hydraulic coupling dynamics of a single suspension cylinder. The control system uses the cavity pressure of the suspension cylinder and the piston rod speed as the state variables of the observer, and the measured pressure value of the pressure sensor as the output of the observer.

[0090] The first state equation of the system comes from the force balance relationship of the cylinder piston: the resultant force on the piston is equal to the pressure in the cavity multiplied by the hydraulic thrust generated by the effective pressure area, minus the vertical load reaction force from the frame and cargo, and then minus the viscous frictional resistance when the piston moves. The resultant force divided by the equivalent mass of the piston and connecting parts is the acceleration of the piston.

[0091] The second state equation of the system is derived from the oil volumetric elastic modulus equation: the rate of change of pressure in the cavity is equal to the oil volumetric elastic modulus divided by the current effective volume of the cavity, and then multiplied by the difference between the net oil flow rate into the cavity and the rate of change of volume caused by the piston movement.

[0092] Based on the two state equations mentioned above, the sliding mode observer constructs a set of state estimation equations with the same structure as the system model, and introduces a switching correction term based on the pressure estimation error into the estimation equations. The structure of this switching correction term is as follows: take the sign function of the pressure observation error (i.e., the difference between the measured pressure and the estimated pressure), and multiply it by a switching gain. When the observation error is positive, the sign function outputs a positive one; when the observation error is negative, it outputs a negative one. This discontinuous switching action forces the observer's state trajectory to the pre-designed sliding surface.

[0093] The sliding surface is designed using a linear combination that makes both the pressure observation error and its rate of change zero. Specifically, the pressure observation error multiplied by a positive design factor plus the rate of change of the pressure observation error equals zero. The switching gain must be greater than the upper bound of all unmodeled disturbances and parameter uncertainties in the system to ensure that the observer state is driven onto the sliding surface within a finite time and remains there.

[0094] Once the observer state reaches and remains on the sliding surface, the pressure observation error is forced to zero. At this point, the equivalent control quantity contained in the switching correction term precisely reflects the combined effect of all external disturbances not described by the system's nominal model.

[0095] The control system applies a low-pass filter to the equivalent control quantity to extract its gradual and rapid components: the gradual component corresponds to the continuous load offset caused by cargo eccentricity or ground tilt, i.e., the static off-center load component; the rapid component corresponds to the transient impact force caused by uneven road surface, i.e., the dynamic impact component.

[0096] The control system uses only the pressure value obtained by converting the static off-center load component for subsequent actual vertical load calculations, while isolating the dynamic impact component from the load control loop, thereby avoiding over-adjustment of the predictive controller caused by transient impact interference.

[0097] For example, when a heavy-duty omnidirectional AGV traverses a road surface paved with metal speed bumps, the piston in the suspension cylinder is rapidly compressed the instantaneously upon passing the speed bump. The instantaneous rate of change of the displacement signal exceeds the excitation threshold, triggering the sliding mode observer to intervene. The observer converges to the sliding surface within several sampling cycles and separates the pressure signal: the static off-center load component remains at a stable value reflecting the current eccentricity of the cargo, while the dynamic impact component presents as a pulse waveform lasting approximately tens of milliseconds. The control system calculates the vertical load based on the static off-center load component, avoiding misjudging overload at the suspension point due to transient pressure spikes caused by speed bumps, thus preventing the predictive controller from generating unnecessary hydraulic pressure adjustment actions due to brief disturbances.

[0098] S402. Obtain the current operating condition characteristics and determine the load weight, attitude weight, and tilt angle constraint boundary corresponding to the operating condition characteristics.

[0099] Refer to step S302, which will not be repeated here.

[0100] In some embodiments, the control system acquires vehicle tilt angle data and actual vertical load within a continuous time window; extracts the low-frequency drift rate of the vehicle tilt angle data and the step amplitude of the actual vertical load to determine the current operating condition characteristics; and performs interpolation optimization in a preset multidimensional weight matrix based on the operating condition characteristics to determine the load weight, attitude weight, and tilt angle constraint boundary corresponding to the operating condition characteristics.

[0101] Among them, the low-frequency drift rate refers to the rate of change of the low-frequency component of the vehicle tilt angle data within a continuous time window, used to characterize the slow deviation trend of the vehicle attitude. A larger value usually corresponds to driving on a slope or turning in a curve. The step amplitude refers to the maximum amplitude of the step change that occurs in the actual vertical load signal within a continuous time window, used to characterize the degree of load abrupt change. A larger value usually corresponds to loading and unloading goods or passing through abrupt road surface changes.

[0102] Specifically, the control system maintains a sliding time window within each control cycle. The vehicle tilt angle data within the window is low-frequency components extracted using a low-pass filter. The change in this low-frequency component between the start and end points of the window is calculated and divided by the window duration to obtain the low-frequency drift rate. Simultaneously, the control system performs step detection on the actual vertical load signal within the window and calculates the maximum step amplitude of the load change within the window. Using the low-frequency drift rate and step amplitude as two-dimensional coordinates, the control system locates the region of the current operating condition characteristic within a preset multi-dimensional weight matrix and calculates the load weight, attitude weight, and tilt angle constraint boundary corresponding to that coordinate point using bilinear interpolation. For example, when the low-frequency drift rate is high and the step amplitude is low, corresponding to a constant speed turning curve, the attitude weight should be high to emphasize tilt angle control; when the low-frequency drift rate is low and the step amplitude is high, corresponding to loading / unloading or sudden road surface changes, the load weight should be high to emphasize load balance; when both indicators are low, corresponding to a constant speed straight driving condition, the load weight and attitude weight are both moderately valued.

[0103] It should be noted that the construction of the multi-dimensional weight matrix was completed offline during the system calibration phase. The control system calibration engineers used the tilt angle low-frequency drift rate and vertical load step amplitude as two independent dimensions, dividing the value range of each dimension into several discrete nodes at equal intervals or non-equal intervals determined by engineering experience. The nodes for the low-frequency drift rate dimension cover from zero (representing a straight-line driving condition where the vehicle is completely stationary) to the maximum drift rate value corresponding to the minimum turning radius of this type of AGV; the nodes for the step amplitude dimension cover from zero (representing a completely stable load) to the maximum step amplitude value corresponding to the sudden application or removal of the maximum rated load of this type of AGV.

[0104] At each grid point formed by the intersection of all discrete nodes in the two dimensions, calibration engineers determine the optimal load weight, attitude weight, and tilt constraint boundary values ​​for achieving comprehensive optimal load balance and attitude stability under specific operating conditions through simulation experiments or real-vehicle tests. These three sets of parameters are stored in three two-dimensional lookup tables that correspond one-to-one with the grid points. These three lookup tables together constitute a multi-dimensional weight matrix. During online system operation, the control system calculates the current actual values ​​of the low-frequency drift rate and step amplitude in each control cycle, using these two actual values ​​as coordinates to locate the grid cell in the aforementioned two-dimensional grid.

[0105] Since the actual values ​​typically do not fall exactly on a discrete node, the control system employs a bilinear interpolation algorithm to weighted average the parameter values ​​at the four vertices of the cell: first, linear interpolation is performed along the low-frequency drift rate dimension on two pairs of adjacent vertices of the cell to obtain two intermediate values; then, linear interpolation is performed along the step amplitude dimension on these two intermediate values ​​to obtain the final interpolation result. This process is performed once for each of the three parameters: load weight, attitude weight, and tilt constraint boundary, thereby obtaining continuous parameter values ​​that accurately adapt to the characteristics of the current operating condition.

[0106] For example, in a certain calibration scheme, the low-frequency drift rate dimension is divided into 5 nodes: 0, 0.05, 0.10, 0.15, and 0.20 degrees / second, and the step amplitude dimension is divided into 5 nodes: 0, 5, 10, 15, and 20 kN, forming a total of 25 grid points. When the AGV's actual low-frequency drift rate is 0.07 degrees / second and the actual step amplitude is 8 kN at a certain moment, the coordinate falls within the grid cell composed of nodes (0.05, 5), (0.10, 5), (0.05, 10), and (0.10, 10). The control system performs bilinear interpolation on the pre-stored load weight values ​​at these four vertices to obtain the load weight used in the current cycle. The attitude weight and tilt angle constraint boundary are also bilinearly interpolated to obtain the corresponding values.

[0107] In addition, under different operating conditions, the control system adjusts the numerical ratio of load weight, attitude weight and tilt angle constraint boundary to make the optimization bias of the predictive controller adapt to the changes in operating conditions in real time. For example, in the case of cornering, the attitude weight is increased and the tilt angle constraint boundary is narrowed, while in the case of straight driving, the load weight is appropriately increased to focus on load balance.

[0108] S403. Obtain the frame stiffness parameters of the heavy-duty omnidirectional AGV and calculate the frame torsional deformation based on the actual vertical load.

[0109] After determining the load weights, attitude weights, and tilt constraint boundaries in step S402, the control system needs to assess the structural bearing capacity of the chassis before generating pressure adjustment commands to avoid excessive deformation of the chassis caused by the pressure adjustment commands output by the subsequent rolling solution. To this end, the control system first obtains the chassis stiffness parameters and calculates the current torsional deformation of the chassis.

[0110] Among them, frame stiffness parameters refer to physical quantities characterizing the frame's ability to resist torsional deformation. These typically include the frame's torsional stiffness coefficient, which is determined by parameters such as the frame's material elastic modulus, moment of inertia, and structural length. This coefficient is calibrated during the frame design phase through finite element analysis or torsional tests and stored in the control system. Frame torsional deformation refers to the torsional angle or deformation caused by uneven support reactions along the frame's diagonal direction under the current vertical load distribution at each suspension point.

[0111] Specifically, the control system reads pre-calibrated frame stiffness parameters from internal storage, including the frame's torsional stiffness coefficient and the geometric coordinates of each suspension point. Based on the actual vertical loads of each differential steering wheel determined in step S401, the control system calculates the torsional moment generated by the uneven distribution of vertical loads at each suspension point. This torsional moment is calculated by multiplying the difference in vertical loads at two sets of suspension points diagonally along the frame by the corresponding lever arm length, obtaining the torsional moment component acting on the frame. The control system divides the torsional moment by the frame's torsional stiffness coefficient to obtain the current frame torsional deformation. The magnitude of this torsional deformation reflects the level of torsional stress on the frame structure caused by the current load distribution. For example, when the left front wheel and right rear wheel of a four-wheeled AGV bear a larger load, while the right front wheel and left rear wheel bear a smaller load, the frame undergoes torsional deformation along the diagonal direction; this deformation is the frame torsional deformation.

[0112] In some embodiments, the frame stiffness parameters and torsional deformation can be obtained and calculated in several ways: Optionally, during the initialization phase, the control system loads a stiffness matrix exported by finite element analysis software from a configuration file. This stiffness matrix contains the coupling stiffness coefficients between each suspension point. In each control cycle, the control system multiplies the actual vertical load vector with the inverse of the stiffness matrix to obtain the displacement vector of each suspension point. The difference in the diagonal direction is then extracted from the displacement vector as the torsional deformation. Optionally, the control system treats the frame as an equivalent torsion spring based on a simplified beam model. The torsional deformation is obtained by multiplying the load difference between the diagonal suspension points at both ends of the frame by the lever arm and dividing by the equivalent torsional stiffness. This method involves less computation and is suitable for frames with relatively regular structures. It is understood that other structural mechanics calculation methods can also be used to evaluate the frame torsional deformation; this is not limited here. It should be noted that the frame stiffness parameters may change slowly during the frame's service life due to weld fatigue or structural relaxation. The control system can periodically perform calibration updates under no-load conditions to maintain the accuracy of the stiffness parameters.

[0113] S404. Determine the load deviation tolerance of the frame torsional deformation based on the preset torsional fatigue limit.

[0114] After the frame torsional deformation is calculated in step S403, the control system needs to determine the acceptable range of load deviation in order to set the structural safety boundary in the subsequent rolling solution.

[0115] Torsional fatigue limit refers to the maximum torsional deformation or torsional stress value that the frame material and welded structure can withstand without fatigue failure under long-term cyclic torsional loads. This value is predetermined based on the frame material's SN curve and safety factor and stored in the control system. Load deviation tolerance refers to the maximum allowable range of vertical load deviation between each differential steering wheel, provided that the frame structural safety requirements are met. This range is expressed in the form of load variance or load range.

[0116] Specifically, the control system compares the preset torsional fatigue limit with the current frame torsional deformation calculated in step S403. Based on the torsional fatigue limit value, the control system deduces the maximum torsional moment the frame can withstand, and then converts this maximum torsional moment into the upper limit of the allowable vertical load difference between each suspension point. This upper limit is the load deviation tolerance. When the current frame torsional deformation has a large margin from the torsional fatigue limit, the load deviation tolerance is correspondingly wider, indicating that the control system can operate within a larger load deviation range without causing frame structural fatigue. When the margin is small, the load deviation tolerance narrows, requiring the control system to more actively promote load balancing to reduce frame torsional stress. Determining the load deviation tolerance provides structural safety constraints for subsequent rolling solutions, avoiding the introduction of adjustments exceeding the frame's bearing capacity during load balancing.

[0117] It should be noted that when the control system performs load deviation tolerance calculation, it can first read the torsional fatigue limit value, which has been calibrated through fatigue tests or finite element simulations, from its internal storage. This value represents the maximum torsional angle that the frame can withstand over a long period under a given safety factor. The control system subtracts the current frame torsional deformation calculated in step S403 from this torsional fatigue limit value to obtain the remaining available torsional deformation margin of the frame under the current operating condition. The control system multiplies this torsional deformation margin by the frame's torsional stiffness coefficient to obtain the additional torsional moment margin that the frame can currently withstand.

[0118] Since the torsional moment of the chassis is generated by the uneven distribution of vertical loads at each suspension point, the control system needs to inversely map the torsional moment margin to the allowable range of vertical load deviation at each suspension point. For a multi-suspension-point chassis with a rectangular layout, the torsional moment is mainly driven by the load difference between two pairs of suspension point groups along the diagonal direction. Its magnitude is equal to the load difference between a pair of suspension points on the diagonal multiplied by the projected distance (i.e., lever arm) of the line connecting the pair of suspension points in the direction perpendicular to the torsional axis. Based on this, the control system divides the torsional moment margin by the diagonal lever arm length to obtain the maximum allowable load difference between the pairs of diagonal suspension points, which is the upper limit of the load deviation in the diagonal direction.

[0119] The control system can convert the upper limit of the diagonal load deviation into the variance threshold of the vertical load of all suspension points through geometric relationships: under the assumption that each suspension point is evenly distributed on the frame, there is a fixed proportional relationship between the upper limit of the diagonal load deviation and the global load variance, which is determined by the number of suspension points and the geometric layout. The control system uses this proportional relationship to convert the upper limit of the diagonal deviation into a load deviation tolerance in the form of variance. Specifically, after obtaining the maximum allowable load difference in the diagonal direction, the control system linearly distributes this maximum load difference to each suspension point based on the preset assumption of uniform distribution of the rigid flat plate of the vehicle body to obtain the single-point limit deviation tolerance. That is, the maximum diagonal load difference is directly divided by the total number of suspension points currently configured in the heavy-duty omnidirectional AGV to obtain the maximum allowable load amount of each differential steering wheel that deviates from the mean. Then, the control system performs a square operation on the single-point limit deviation tolerance of each suspension point and sums all the square values ​​to obtain the arithmetic mean, thereby generating the variance threshold data used to constrain the rolling solution range of the predictive controller. For example, in a system equipped with four sets of differential steering wheels, when the calculated maximum load difference in the diagonal direction is 500N, the control system will first divide it by 4 to obtain the single-point limit deviation tolerance of 125N, and then sum the squares of all four 125N values ​​and divide them by 4 to calculate the variance threshold of 15625, which is used as the load deviation tolerance input to the predictive controller.

[0120] The tolerance value decreases as the current frame torsional deformation increases: when the frame is already in a high torsional deformation state, the remaining margin is limited, the tolerance is correspondingly narrowed, and the predictive controller will be required to operate within a smaller load deviation range; when the frame torsional deformation is small, the tolerance is wider, and the control system has greater adjustment freedom while meeting the load balance requirements.

[0121] For example, the frame torsional fatigue limit of a certain heavy-duty omnidirectional AGV is calibrated as 0.8 degrees, the torsional stiffness coefficient is 5000 N·m / degree, and the diagonal lever arm length is 3 meters. Under the current load distribution, the frame torsional deformation has reached 0.5 degrees, then the remaining torsional deformation margin is 0.3 degrees, the corresponding torsional moment margin is 1500 N·m, and the maximum allowable load difference in the diagonal direction is 500 N. The variance-form load deviation tolerance calculated by the control system based on this is the structural safety constraint boundary for subsequent rolling solutions.

[0122] S405. Perform rolling solution within the tilt angle constraint boundary and load deviation tolerance to determine the pressure adjustment command of each suspension cylinder.

[0123] After determining the tilt angle constraint boundary in step S402 and the load deviation tolerance in step S404, the control system has two types of boundary conditions: attitude constraint and structural constraint, and can perform rolling solution under dual constraints.

[0124] Specifically, the control system uses the actual vertical load and vehicle tilt angle data obtained in step S401, and the load weight and attitude weight obtained in step S402 as inputs to the predictive controller, and simultaneously uses the tilt angle constraint boundary and load deviation tolerance as hard constraints for the predictive controller. The predictive controller performs rolling optimization on the pressure adjustment of each suspension cylinder within the prediction time domain. The constraints include: the vehicle tilt angle does not exceed the tilt angle constraint boundary throughout the entire prediction time domain, and the variance of the vertical load deviation at each suspension point does not exceed the load deviation tolerance. When both types of constraints are satisfied, the predictive controller selects the optimal control sequence that minimizes the objective function value from the feasible solutions that satisfy the constraints, and takes the first element of this sequence as the pressure adjustment command for the current control cycle. This step corresponds to... Figure 2 The rolling solver 2042 within the predictive control core module 204 performs the solution process after receiving the output from the stiffness tolerance calculation unit 2041. Compared to step S303, this step adds load deviation tolerance as a constraint, ensuring that the solution results promote load balance without causing excessive torsional deformation of the chassis.

[0125] To achieve white-box design of the predictive controller and incorporate the vehicle frame's own physical characteristics, the control system, while performing rolling solutions within the tilt angle constraint boundary and load deviation tolerance, constructed a data processing logic that integrates incremental control penalty terms and structural stiffness constraints. The mathematical expression for the rolling optimization cost function established by the control system is as follows: Among them, the newly added To control the incremental penalty weight, This is the difference between the command output to the proportional pressure reducing valve in the current control cycle and the previous cycle (i.e., the adjustment increment).

[0126] In solving this cost function, in addition to setting physical limiting constraints, the control system directly converts the load deviation tolerance calculated in step S404 into a hard constraint of the load difference inequality between the diagonal suspension points (i.e., ||). |≤Δ , where Δ (This refers to the most forgiving pressure difference obtained by mapping the frame torsional stiffness and fatigue limit). In the prediction time domain, the solver searches for the optimal control sequence that minimizes both load and attitude deviations while simultaneously adjusting the action amplitude most smoothly. For example, when transporting a large transformer with a severely offset center of gravity, the controller, within the solver's evaluation, finds that if the pressure of the cylinders at the four corners is forcibly equalized, the load difference on the diagonal will exceed the fatigue threshold limit corresponding to the frame stiffness. In this case, under the constraint of the white-box hard constraint of load deviation tolerance, the MPC algorithm will proactively abandon absolute pressure equalization and utilize an incremental penalty term. By constraining each valve to smoothly transition to the safety tolerance boundary, physical-level defense against metal fatigue of heavy-duty vehicle frames is achieved at the software algorithm level.

[0127] In some embodiments, the control system extracts the variance data of the actual vertical load during the rolling solution process; when the variance data falls within the load deviation tolerance and the vehicle tilt angle data is within the tilt angle constraint boundary, the current rolling solution cycle is frozen and a zero-increment adjustment signal is output; the zero-increment adjustment signal is used as the pressure adjustment command for each suspension cylinder.

[0128] Among them, variance data refers to the variance of the actual vertical load of each differential steering wheel relative to the mean, used to quantify the unevenness of the current load distribution. Zero-increment adjustment signal refers to a control signal with zero pressure adjustment, meaning that each suspension cylinder maintains the hydraulic pressure of the previous cycle within the current control cycle. Freeze rolling solution cycle means that no new round of optimization calculation is performed within the current control cycle, and the zero-increment signal is directly used as the output.

[0129] Specifically, before executing the rolling solution in each control cycle, the control system first extracts the variance data of the actual vertical load of each differential steering wheel, which was recently determined in step S401. The control system compares this variance data with the load deviation tolerance determined in step S404, and simultaneously compares the current vehicle tilt angle data with the tilt angle constraint boundary determined in step S402. When the variance data is less than or equal to the upper limit of the load deviation tolerance, and both the roll and pitch angles of the vehicle are within the tilt angle constraint boundary, the control system determines that the current load distribution and vehicle attitude are acceptable, freezes the rolling solution for this control cycle, and outputs a zero-increment adjustment signal. Each proportional valve maintains its current opening after receiving the zero-increment adjustment signal, and the hydraulic pressure of each suspension cylinder remains at the level of the previous cycle. When the variance data exceeds the load deviation tolerance or the tilt angle data exceeds the tilt angle constraint boundary in subsequent sampling cycles, the control system unfreezes the process and resumes normal rolling solution. This step corresponds to... Figure 2 The function of the zero-incremental freeze unit 2043 within the predictive control core module 204.

[0130] The predictive controller performs rolling solution and handles optimization conflicts within the feasible region with hard constraints of inequalities as follows: The control system constructs a rolling optimization objective function with minimizing the vertical load deviation of each differential steering wheel as the main optimization objective and the limits of vehicle roll angle and pitch angle as hard boundary constraints of inequalities. In the conventional iterative solution, the algorithm minimizes the deviation of the load of each steering wheel within the feasible region without exceeding the limit of vehicle tilt angle. When extreme load imbalance causes objective conflict in the optimization solution, i.e., the algorithm determines that further reducing the load difference will inevitably cause the vehicle tilt angle to exceed the safety limit, the control system will not sacrifice attitude stability indefinitely to achieve absolute load balance. Instead, it automatically triggers a degenerate optimization mechanism. When the vehicle frame attitude is just against the constraint surface of the allowable tilt angle boundary condition (such as the roll angle being at the maximum allowable critical value), it reverses to obtain the optimal load distribution solution that can be achieved under the extreme attitude and uses it as the pressure adjustment output command of each proportional pressure reducing valve. For example, when the eccentric load of the cargo causes the controller to be unable to control the roll angle within the tilt constraint boundary while meeting the load balance target, the controller will constrain the roll angle at the boundary critical value and solve the pressure regulation scheme with the minimum load deviation under this attitude condition.

[0131] S406. Perform a positive correlation calculation of the driving torque on the actual vertical load to obtain the initial torque distribution sequence.

[0132] After determining the pressure regulation command in step S405, the control system needs to synchronously generate torque commands for each differential steering wheel. Unlike step S304, this step splits the torque conversion process into two stages: first, the initial torque distribution sequence is obtained, and then further corrections are made in subsequent steps S407 and S408 based on the motor's thermal state.

[0133] The initial torque distribution sequence refers to the ordered set of torque values ​​for each wheel, obtained by allocating the total required torque of the vehicle based solely on the actual vertical load proportion of each differential wheel, without considering the thermal state of the motor. Each element in this sequence is proportional to the actual vertical load of the corresponding differential wheel.

[0134] Specifically, the control system obtains the actual vertical load of each differential steering wheel determined in step S401, and calculates the proportional coefficient of the actual vertical load of each differential steering wheel to the total vertical load of the vehicle. The control system multiplies the total required torque of the vehicle (determined by the upper-level motion controller based on the target speed and acceleration) by the proportional coefficient of each differential steering wheel to obtain the initial torque distribution sequence. For example, in a heavy-duty omnidirectional AGV with four sets of differential steering wheels, if the actual vertical loads of each differential steering wheel are 30kN, 25kN, 20kN, and 25kN respectively, the total vertical load of the vehicle is 100kN, and the total required torque of the vehicle is 2000N·m, then the initial torque distribution sequence is 600N·m, 500N·m, 400N·m, and 500N·m. This sequence ensures that the wheel with the larger load receives a larger driving force distribution, matching its tire adhesion limit. This step corresponds to... Figure 2 The function of the initial torque conversion unit 2051 within the torque coordination distribution module 205.

[0135] In some embodiments, the positive correlation conversion of drive torque can be performed in several ways: Optionally, the control system directly uses the vertical load percentage of each differential wheel as the distribution coefficient to linearly distribute the total required torque to each wheel, and applies an upper limit protection to the torque of each wheel to ensure that it does not exceed the rated torque of the corresponding motor; Optionally, the control system multiplies the vertical load of each differential wheel by a preset ground friction coefficient to obtain the maximum available driving force of each wheel, and uses the proportion of the maximum available driving force of each wheel to the total available driving force as the distribution coefficient. This method is closer to the physical constraints of tire adhesion. It is understood that other positive correlation mapping methods can also be used to perform torque conversion, which are not limited here.

[0136] S407. Obtain the real-time stator temperature of the motor of each differential steering wheel and determine the torque attenuation coefficient of each differential steering wheel.

[0137] After obtaining the initial torque distribution sequence in step S406, the control system needs to evaluate the thermal state of each drive motor to determine whether each motor can continuously output the initially distributed torque value under the current temperature conditions.

[0138] The real-time stator temperature of the motor refers to the current temperature value collected by the temperature sensor embedded in the stator winding of each differential steering wheel drive motor. This temperature value reflects the real-time thermal state of the motor winding. The torque attenuation coefficient is a proportional coefficient determined based on the stator temperature and used to reduce the output torque of the corresponding motor. Its value ranges from 0 to 1. When the stator temperature is within the safe range, the coefficient is 1 (i.e., no attenuation), and when the stator temperature approaches or exceeds the thermal protection threshold, the coefficient is less than 1.

[0139] Specifically, the control system reads real-time stator temperature data from the motor controllers of each differential steering wheel via the signal acquisition module 201. The control system compares the stator temperature of each motor with a preset temperature-attenuation coefficient mapping relationship to determine the torque attenuation coefficient of each differential steering wheel. This mapping relationship is typically a piecewise linear function: when the stator temperature is below the attenuation initiation temperature, the torque attenuation coefficient is 1; when the stator temperature is between the attenuation initiation temperature and the maximum allowable temperature, the torque attenuation coefficient decreases linearly with increasing temperature; when the stator temperature reaches the maximum allowable temperature, the torque attenuation coefficient drops to a preset minimum value (e.g., 0.3). For example, if a motor has an attenuation initiation temperature of 120°C, a maximum allowable temperature of 160°C, and a minimum attenuation coefficient of 0.3, when the stator temperature of that motor is 140°C, the torque attenuation coefficient is 1 - (140 - 120) / (160 - 120) × (1 - 0.3) = 0.65. This step corresponds to... Figure 2 The process of the torque coordination distribution module 205 receiving stator temperature data from the signal acquisition module 201 and calculating the attenuation coefficient.

[0140] In some embodiments, the torque attenuation coefficient can be determined in several ways: Optionally, the control system pre-stores a piecewise linear mapping table in a configuration file, and uses the stator temperature as input to look up the table and interpolate to obtain the corresponding attenuation coefficient. This method is simple to calculate and easy to calibrate. Optionally, the control system predicts the future trend of the winding temperature in real time based on the motor's thermal impedance model. When it is predicted that the winding temperature will exceed the maximum allowable value within a few seconds, the attenuation coefficient is reduced in advance. This method has feedforward characteristics and can start derating before the temperature actually rises. It is understood that other temperature-attenuation mapping methods or thermal prediction algorithms can also be used to determine the torque attenuation coefficient, which is not limited here.

[0141] S408. Determine the cooperative torque command for each differential steering wheel based on the initial torque distribution sequence and torque attenuation coefficient.

[0142] After obtaining the initial torque distribution sequence in step S406 and determining the torque attenuation coefficient of each differential steering wheel in step S407, the control system combines the two to generate the final cooperative torque command.

[0143] Specifically, the control system multiplies each element of the initial torque distribution sequence obtained in step S406 by the torque attenuation coefficient of the corresponding differential wheel determined in step S407 to obtain the attenuated torque value of each differential wheel. The control system compares the sum of the attenuated torque values ​​of each differential wheel with the sum of the initial torque distribution sequence; the difference between the two is the amount of torque loss in the entire vehicle caused by thermal attenuation. The control system allocates this torque loss according to the proportion of the remaining torque margin of each low-temperature differential wheel and adds it to the attenuated torque value of each low-temperature differential wheel to obtain the final coordinated torque command. For example, following the example of step S406, if the torque attenuation coefficients of the four sets of differential steering wheels are 1.0, 0.65, 1.0, and 1.0 respectively, then the attenuated torque sequence is 600 N·m, 325 N·m, 400 N·m, and 500 N·m, and the total vehicle torque loss is 2000 - (600 + 325 + 400 + 500) = 175 N·m. This 175 N·m will be proportionally distributed to the first, third, and fourth sets of differential steering wheels. This step corresponds to... Figure 2 The function of the internal heat derating and compensation unit 2052 of the medium torque collaborative distribution module 205.

[0144] In some embodiments, the control system calculates an initial torque sequence based on the initial torque distribution sequence and the torque attenuation coefficient; calculates the vehicle torque deficit based on the initial torque sequence and the initial torque distribution sequence; and proportionally compensates the vehicle torque deficit to the initial torque of the remaining differential steering wheels whose stator temperature is below the temperature threshold, thereby determining the cooperative torque command for each differential steering wheel.

[0145] The initial torque sequence refers to the set of intermediate torque values ​​for each differential steering wheel obtained by multiplying each element in the initial torque distribution sequence by its corresponding torque attenuation coefficient. The total vehicle torque loss is the difference between the sum of the initial torque distribution sequence and the sum of the initial torque sequence; this difference characterizes the loss of overall vehicle driving capability due to thermal attenuation. The temperature threshold is the stator temperature critical value used to determine whether the differential steering wheel motor is in a low-temperature state; motors below this threshold are considered low-temperature motors with torque compensation capabilities.

[0146] Specifically, the control system multiplies each element of the initial torque distribution sequence obtained in step S406 by the corresponding torque attenuation coefficient determined in step S407 to obtain the initial calculated torque sequence. The control system subtracts the sum of the initial calculated torque sequences from the sum of the initial torque distribution sequences to obtain the total vehicle torque deficit. The control system iterates through the stator temperatures of each differential steering wheel's motor, filters out the set of differential steering wheels whose stator temperatures are below the temperature threshold, and calculates the remaining torque margin of each differential steering wheel in this set, i.e., the difference between the rated torque of each low-temperature motor and its initial calculated torque. The control system distributes the total vehicle torque deficit according to the proportion of the remaining torque margin of each low-temperature differential steering wheel to the total remaining margin, and adds the distributed amount to the initial calculated torque of the corresponding low-temperature differential steering wheel to obtain the cooperative torque command for each differential steering wheel. For high-temperature differential steering wheels (i.e., differential steering wheels with stator temperatures higher than or equal to the temperature threshold), their cooperative torque command is the initial calculated torque value and does not participate in compensation reception. This step corresponds to... Figure 2 The function of the internal thermal derating and compensation unit 2052 of the medium torque collaborative distribution module 205.

[0147] In some embodiments, there may be a situation where the stator temperature of all differential steering wheels is higher than the temperature threshold, resulting in no available cryogenic motor to perform compensation. In response, the control system will directly set the cooperative torque command of each differential steering wheel to the values ​​in the initial torque sequence, accept the shortfall in the total drive torque of the vehicle, and send a torque limitation signal to the upper-level motion controller. The upper-level motion controller will then reduce the target speed or acceleration to match the currently available drive capability, and simultaneously trigger a vehicle deceleration alarm.

[0148] S409, based on pressure regulation commands and coordinated torque commands, adjusts the hydraulic pressure of each suspension cylinder and the output torque of each differential steering wheel to control the operation of the heavy-duty omnidirectional AGV.

[0149] Refer to step S305, which will not be repeated here.

[0150] In this embodiment, the technical means of generating pressure adjustment commands for each suspension cylinder by performing rolling solution based on the predictive controller under the dual constraints of tilt angle constraint boundary and load deviation tolerance, and simultaneously performing positive correlation conversion of drive torque on the actual vertical load and thermal attenuation compensation combined with motor stator temperature to generate coordinated torque commands for each differential steering wheel, allows the pressure adjustment of the hydraulic suspension system and the torque distribution of the drive system to coordinate under a unified control framework. This effectively solves the problems of continuous overload of some suspension points due to lag in oil flow response in passive hydraulic suspension, as well as wheel slippage and high-temperature motor overload operation caused by uniform distribution of drive torque. It achieves coordinated protection of load balance, attitude stability, frame structure safety and motor thermal protection for heavy-duty omnidirectional AGV under various operating conditions.

[0151] The control system in the embodiments of this application is described below from the perspective of hardware processing. Please refer to [link / reference needed]. Figure 5 This is a schematic diagram of the physical device structure of the control system in an embodiment of this application.

[0152] It should be noted that, Figure 5 The structure of the control system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0153] like Figure 5 As shown, the control system includes a CPU 501, which can perform various appropriate actions and processes according to a program stored in ROM 502 or a program loaded from storage section 508 into RAM 503, such as executing the methods described in the above embodiments. RAM 503 also stores various programs and data required for system operation. The CPU 501, ROM 502, and RAM 503 are interconnected via bus 504. I / O interface 505 is also connected to bus 504.

[0154] The following components are connected to I / O interface 505: input section 506 including audio input devices, push-button switches, etc.; output section 507 including a liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 508 including a hard disk, etc.; and communication section 509 including a network interface card such as a LAN (Local Area Network) card, modem, etc. Communication section 509 performs communication processing via a network such as the Internet. Drive 510 is also connected to I / O interface 505 as needed. Removable media 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 510 as needed so that computer programs read from them can be installed into storage section 508 as needed.

[0155] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by CPU 501, it performs the various functions defined in this application.

[0156] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0157] Specifically, the control system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the heavy-duty omnidirectional AGV hydraulic floating suspension load balancing and attitude cooperative constraint control method provided in the above embodiment.

[0158] In another aspect, this application also provides a computer-readable storage medium, which may be included in the control system described in the above embodiments; or it may exist independently and not assembled into the control system. The storage medium carries one or more computer programs that, when executed by a processor of the control system, cause the control system to implement the heavy-duty omnidirectional AGV hydraulic floating suspension load balancing and attitude cooperative constraint control method provided in the above embodiments.

[0159] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0160] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

Claims

1. A method for load balancing and attitude coordination constraint control of a heavy-duty omnidirectional AGV with hydraulic floating suspension, characterized in that, The heavy-duty omnidirectional AGV is equipped with multiple sets of differential steering wheels and correspondingly connected suspension cylinders; the method is applied to a control system, and the method includes: Acquire vehicle body attitude signals, suspension cylinder pressure signals and displacement signals, and determine the actual vertical load and vehicle body tilt angle data of each differential steering wheel; Obtain the current operating condition characteristics, and determine the load weight, attitude weight, and tilt angle constraint boundary corresponding to the operating condition characteristics; The actual vertical load, the vehicle body tilt angle data, the load weight, the attitude weight, and the tilt angle constraint boundary are input into the predictive controller. Within the tilt angle constraint boundary, a rolling solution is performed with the goal of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command of each suspension cylinder. Perform a positive correlation calculation of the driving torque on the actual vertical load to determine the cooperative torque command for each differential steering wheel; Based on the pressure regulation command and the coordinated torque command, the hydraulic pressure of each of the suspension cylinders and the output torque of each of the differential steering wheels are adjusted to control the operation of the heavy-duty omnidirectional AGV.

2. The method according to claim 1, characterized in that, The steps of acquiring vehicle attitude signals, suspension cylinder pressure signals, and displacement signals, and determining the actual vertical load and vehicle tilt angle data of each differential steering wheel specifically include: Acquire vehicle body attitude signals, suspension cylinder pressure signals, and displacement signals, and calculate the instantaneous rate of change of the displacement signals; When the instantaneous rate of change exceeds the excitation threshold, the pressure signal is dynamically decoupled using a sliding mode observer to obtain the static off-center load component and the dynamic impact component. Based on the static off-center load component and the vehicle attitude signal, the actual vertical load and vehicle tilt angle data of each differential steering wheel are determined.

3. The method according to claim 1, characterized in that, The step of performing a rolling solution within the tilt angle constraint boundary with the objective of minimizing the vertical load deviation of each differential steering wheel to determine the pressure adjustment command of each suspension cylinder specifically includes: Obtain the frame stiffness parameters of the heavy-duty omnidirectional AGV, and calculate the frame torsional deformation based on the actual vertical load; Based on the preset torsional fatigue limit, the load deviation tolerance of the frame torsional deformation is determined; A rolling solution is performed within the tilt angle constraint boundary and the load deviation tolerance to determine the pressure adjustment command for each suspension cylinder.

4. The method according to claim 3, characterized in that, The step of performing a rolling solution within the tilt angle constraint boundary and the load deviation tolerance to determine the pressure adjustment command for each suspension cylinder specifically includes: During the rolling solution process, the variance data of the actual vertical load is extracted; When the variance data falls within the load deviation tolerance and the vehicle tilt angle data is within the tilt angle constraint boundary, freeze the current rolling solution cycle and output a zero-increment adjustment signal. The zero-increment adjustment signal is used as the pressure adjustment command for each suspension cylinder.

5. The method according to claim 1, characterized in that, The step of performing a positive correlation conversion of the actual vertical load to determine the cooperative torque command for each differential steering wheel specifically includes: Perform a positive correlation calculation of the driving torque on the actual vertical load to obtain the initial torque distribution sequence; The real-time stator temperature of the motor of each differential steering wheel is obtained, and the torque attenuation coefficient of each differential steering wheel is determined. Based on the initial torque distribution sequence and the torque attenuation coefficient, the cooperative torque command for each differential steering wheel is determined.

6. The method according to claim 5, characterized in that, The step of determining the cooperative torque command for each differential steering wheel based on the initial torque distribution sequence and the torque attenuation coefficient specifically includes: Calculate the initial torque sequence based on the initial torque distribution sequence and the torque attenuation coefficient; Calculate the vehicle torque loss based on the initial torque sequence and the initial torque distribution sequence; The total torque deficit of the vehicle is proportionally compensated into the initial calculated torque of the remaining differential steering wheels whose stator temperature is below the temperature threshold, thereby determining the cooperative torque command for each differential steering wheel.

7. The method according to claim 1, characterized in that, The steps of obtaining the current operating condition characteristics and determining the load weights, attitude weights, and tilt constraint boundaries corresponding to the operating condition characteristics specifically include: Acquire the vehicle body tilt angle data and the actual vertical load within a continuous time window; Extract the low-frequency drift rate of the vehicle body tilt angle data and the step amplitude of the actual vertical load to determine the current operating condition characteristics; Based on the operating condition characteristics, interpolation optimization is performed in a preset multidimensional weight matrix to determine the load weight, attitude weight, and tilt constraint boundary corresponding to the operating condition characteristics.

8. A control system, characterized in that, The control system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the control system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the control system, it causes the control system to perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the control system, the control system performs the method as described in any one of claims 1-7.

Images