Response prediction methods, systems, and control equipment for agricultural machinery steering actuators
By modeling the three components of amplitude correction, time delay, and rate limitation in a series, the dynamic response problem of agricultural machinery steering actuators under high-frequency command update conditions was solved, achieving accurate front wheel angle prediction and low-cost deployment.
Patent Information
- Application Number
- CN202611516364.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-28
- Publication Date
- 2026-11-17
AI Technical Summary
In the existing technology, the dynamic response modeling of agricultural machinery steering actuators is difficult to accurately reflect the real dynamic response under high-frequency command update conditions. The model parameters have poor coupling and are difficult to interpret, resulting in a heavy computational burden and making it difficult to stably deploy on low-cost electronic control units.
A three-stage approach of amplitude correction, time delay, and rate limit is adopted to model the understeering, pure delay, and rate limit effects of the steering actuator, respectively. The model parameters are obtained through offline calibration to generate the predicted actual front wheel angle.
The model accurately predicts the front wheel angle response under high-frequency command update conditions. The model parameters have clear physical meanings, making it suitable for low-cost electronic control unit deployment and improving the accuracy and stability of trajectory prediction.
Smart Images

Figure CN123104769A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of automatic navigation technology for agricultural machinery, specifically to a response prediction method, system, and control equipment for the steering actuator of agricultural machinery. Background Technology
[0002] In agricultural machinery automatic navigation operations, the dynamic response of the steering actuator from the target front wheel angle to the actual front wheel angle directly affects the accuracy of trajectory prediction and is the core sub-module of the agricultural machinery automatic navigation trajectory prediction model.
[0003] In related technologies, for the dynamic response of steering actuators, the industry generally uses a first-order inertial model or a two-parameter combination model of pure time delay and inertial time constant for modeling, and obtains model parameters mostly through online identification or re-excitation identification before each test. However, the above-mentioned modeling methods in related technologies have the following common problems: First, the model is difficult to accurately reflect the real dynamic response of the actuator under high-frequency command update conditions, and the prediction accuracy is limited; Second, the model parameters are physically coupled and difficult to trace back to specific physical links, resulting in poor interpretability and unstable cross-cycle prediction; Third, parameter acquisition depends on online identification or repeated excitation identification, which is computationally burdensome and difficult to stably deploy on low-cost electronic control units. Summary of the Invention
[0004] This application provides a response prediction method, system, and agricultural machinery control equipment for agricultural machinery steering actuators, so as to at least partially solve the above-mentioned technical problems.
[0005] On one hand, this application provides a response prediction method for an agricultural machinery steering actuator, including: acquiring a target front wheel angle; and generating a predicted actual front wheel angle by serially processing the target front wheel angle through an amplitude correction stage, a time delay stage, and a rate limiting stage; wherein, the amplitude correction stage is used to adjust the amplitude of the target front wheel angle according to a first parameter; the time delay stage is used to delay the output time of the amplitude-corrected target front wheel angle according to a second parameter; and the rate limiting stage is used to limit the expected single-step angle change according to a third parameter; the serial processing includes: using... The first parameter is used to scale the target front wheel angle to obtain the scaled target front wheel angle; the scaled target front wheel angle is delayed by the time corresponding to the second parameter to obtain the delayed target front wheel angle; a single-step steering angle limiting threshold is determined according to the third parameter and the simulation step size to limit the expected single-step steering angle change to obtain the single-step actual steering angle increment; wherein, the expected single-step steering angle change is the difference between the delayed target front wheel angle and the predicted actual front wheel angle of the previous simulation step; the predicted actual front wheel angle of the previous simulation step is superimposed with the single-step actual steering angle increment to obtain the predicted actual front wheel angle of the current simulation step.
[0006] In some embodiments, delaying the scaled target front wheel angle by the duration corresponding to the second parameter includes: determining the delay step number based on the second parameter and the simulation step size; the delay step number being greater than or equal to 1; determining a storage location in the current simulation step based on the step number of the current simulation step and the delay step number; reading the historical scaled target front wheel angle from the storage location in the storage space as the delayed target front wheel angle; the storage space being used to store the historical scaled target front wheel angle; and writing the scaled target front wheel angle corresponding to the current simulation step into the storage location in the storage space.
[0007] In some embodiments, the storage space is a circular queue; determining the storage location based on the step number of the current simulation step and the number of delay steps includes: calculating the remainder obtained by dividing the step number of the current simulation step by the number of delay steps; and using the remainder as an index to locate the storage location in the circular queue.
[0008] In some embodiments, determining the single-step angle limiting threshold based on the third parameter and the simulation step size, and limiting the expected single-step angle change to obtain the single-step actual angle increment, includes: when the expected single-step angle change is greater than or equal to the single-step angle limiting threshold, using the single-step angle limiting threshold as the single-step actual angle increment; when the expected single-step angle change is less than or equal to the negative value of the single-step angle limiting threshold, using the negative value of the single-step angle limiting threshold as the single-step actual angle increment; when the expected single-step angle change is between the negative value of the single-step angle limiting threshold and the single-step angle limiting threshold, using the expected single-step angle change as the single-step actual angle increment.
[0009] In some embodiments, the first parameter is an amplitude scaling factor, used to characterize the steady-state following ratio of the predicted actual front wheel angle to the target front wheel angle under high-frequency command update conditions, and the value range of the amplitude scaling factor is greater than 0 and not greater than 1; the second parameter is a pure delay duration, used to characterize the cumulative dead time from the controller issuing the steering command to the steering actuator starting to act; the third parameter is the maximum rotational angular velocity, used to characterize the upper limit of the speed of the steering actuator constrained by physical power.
[0010] In some embodiments, the amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity are predetermined through the following steps: acquiring calibration log data, which includes measured values of the target front wheel angle and actual front wheel angle at multiple times; constructing an error function with the amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity as variables to be optimized, and minimizing the root mean square error between the predicted actual front wheel angle and the measured actual front wheel angle as the optimization objective; performing joint optimization on the variables to be optimized, and minimizing the error function based on constraints to obtain the amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity; wherein the constraints include: the amplitude scaling factor is greater than 0 and not greater than 1, and the pure delay duration and the maximum rotational angular velocity are both positive numbers.
[0011] In some embodiments, the joint optimization solution employs a gradient-free optimization algorithm.
[0012] In some embodiments, the gradient-free optimization algorithm is the Nelder-Mead simplex optimization algorithm.
[0013] In some embodiments, the method further includes: predicting the trajectory of the agricultural machinery based on the predicted actual front wheel angle and a preset Ackerman kinematic model.
[0014] On the other hand, embodiments of this application provide a response prediction system for an agricultural machinery steering actuator, comprising: a data acquisition module for acquiring a target front wheel angle; an amplitude correction module for adjusting the amplitude of the target front wheel angle according to a first parameter; a time delay module for delaying the output time of the amplitude-corrected target front wheel angle according to a second parameter; a rate limiting module for limiting the expected single-step angle change of the steering actuator according to a third parameter; and a prediction calculation module for generating a predicted actual front wheel angle based on the output result after the series processing of the amplitude correction module, the time delay module, and the rate limiting module.
[0015] On the other hand, embodiments of this application provide an agricultural machinery control device, including a memory and a processor. The memory stores computer programs or instructions, and the processor executes the computer programs or instructions to implement any of the response prediction methods for agricultural machinery steering actuators provided in embodiments of this application.
[0016] On the other hand, embodiments of this application provide a computer-readable storage medium storing a computer program or instructions, which, when executed by a processor, implement the steps in any of the response prediction methods for agricultural machinery steering actuators provided in embodiments of this application.
[0017] On the other hand, embodiments of this application also provide a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps in the response prediction method for any of the agricultural machinery steering actuators provided in embodiments of this application.
[0018] This application discloses a method, system, and control device for predicting the response of an agricultural machinery steering actuator. The method acquires the target front wheel angle and processes it through a series of amplitude correction, time delay, and rate limiting steps to generate a predicted actual front wheel angle. The amplitude correction step adjusts the amplitude of the target front wheel angle according to a first parameter; the time delay step delays the output time of the amplitude-corrected target front wheel angle according to a second parameter; and the rate limiting step limits the expected single-step angle change according to a third parameter. This application replaces the first-order inertial response with a speed- and amplitude-limited linear ramp response, decomposing the dynamic response of the steering actuator into three independent physical effect steps: understeering, pure delay, and rate limit. These are modeled separately and recursively cascaded, enabling accurate prediction of the front wheel angle response under high-frequency command update conditions. The model parameters have clear and independent physical meanings. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 A flowchart illustrating a response prediction method for an agricultural machinery steering actuator provided in an embodiment of this application; Figure 2 A schematic diagram of a three-stage serial signal flow provided in an embodiment of this application; Figure 3 A schematic diagram of a single-beat step response provided in an embodiment of this application; Figure 4 A comparative schematic diagram of a speed-limiting delay model and a first-order inertial model provided for embodiments of this application; Figure 5 This application provides a real-vehicle log verification diagram of steering response under high-frequency command update conditions. Figure 6 This is a schematic diagram of the structure of a response prediction system for an agricultural machinery steering actuator provided in an embodiment of this application. Detailed Implementation
[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0022] In the following description, specific embodiments of this application will be illustrated with reference to steps and symbols performed by one or more computers, unless otherwise stated. Therefore, these steps and operations will be referred to several times as being performed by a computer, and computer execution as referred to herein includes operations by a computer processing unit representing electronic signals of data in a structured format. This operation transforms the data or maintains it at a location in the computer's memory system, which can be reconfigured or otherwise alter the operation of the computer in a manner well known to those skilled in the art. The data structure maintained by the data is the physical location of the memory, which has specific characteristics defined by the data format. However, the principles of this application are described in the foregoing text, which is not intended to be limiting, and those skilled in the art will understand that many of the steps and operations described below can also be implemented in hardware.
[0023] The terms "module" or "unit" as used herein can be considered as software objects executing on the computing system. The various components, modules, engines, and services described herein can be considered as implementations on the computing system. While the apparatuses and methods described herein are implemented in software in some embodiments, they can also be implemented in hardware, all of which are within the scope of this application.
[0024] Those skilled in the art will understand that, unless explicitly stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when an element is “connected” or “coupled” to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein may include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.
[0025] In automated navigation operations such as tilling, sowing, plant protection, and harvesting, agricultural machinery needs to travel precisely along preset paths. Trajectory prediction is a key technical aspect: starting from the machinery's current position, the system continuously extrapolates the vehicle's trajectory over a future period based on the control strategy, and then determines steering commands accordingly. The trajectory prediction model needs to accurately predict the vehicle's trajectory over this period, and the steering response model is the core submodule of the trajectory prediction model—it is responsible for predicting the actual response of the vehicle's front wheel angles after a given desired front wheel angle command, and then predicting the vehicle's trajectory through kinematic extrapolation.
[0026] The dynamic response of agricultural machinery steering actuators (including steering motors / hydraulic steering axles and transmission mechanisms) from the target front wheel angle to the actual front wheel angle involves multiple physical factors, including understeer (the actual front wheel angle is smaller than the target front wheel angle under high-frequency command update conditions), command transmission delay (dead time for communication and drive response), and rate limitation (the motor / hydraulic cylinder is constrained by maximum power and safety speed limits, resulting in a maximum rotational angular velocity). The accuracy of front wheel angle prediction directly determines the accuracy of trajectory prediction, and thus the alignment quality of the entry line.
[0027] Industry-related technologies often model the above response process using first-order inertial models or online identification methods. The former does not explicitly characterize the upper limit of the actuator's speed, while the latter involves large computational loads and time-varying and unstable parameters, making it difficult to operate stably on low-cost agricultural machinery controllers.
[0028] Based on this, embodiments of this application provide a method for predicting the response of an agricultural machinery steering actuator. Please refer to... Figure 1 The response prediction method includes the following steps S110 and S120.
[0029] S110: Obtain the target front wheel angle. Wherein, the target front wheel angle... The desired front wheel angle (unit: °) is output by the controller.
[0030] S120: The target front wheel angle is processed in series with amplitude correction, time delay, and speed limiting stages to generate the predicted actual front wheel angle. The predicted actual front wheel angle... The actual front wheel angle prediction value obtained by the model provided in the embodiments of this application in the current simulation step.
[0031] The amplitude correction stage is used to adjust the amplitude of the target front wheel angle according to the first parameter; the time delay stage is used to delay the output time of the amplitude-corrected target front wheel angle according to the second parameter; and the rate limiting stage is used to limit the expected single-step angle change according to the third parameter. The amplitude correction stage, time delay stage, and rate limiting stage are three physical effect stages connected in series, and their functions are respectively: scaling the amplitude, shifting and delaying the output time, and symmetrically limiting the single-step change. The three stages are connected in series, with the output of the previous stage serving as the input of the next stage. In some embodiments, the series order of the three stages can also be time delay stage, amplitude correction stage, and rate limiting stage, that is, shifting and delaying the output time first, followed by scaling the amplitude. This application does not limit this.
[0032] In some embodiments, the serial processing in step S120 above includes the following sub-steps S121 to S124.
[0033] S121: The target front wheel angle is scaled using the first parameter to obtain the scaled target front wheel angle. Let the first parameter be... The target front wheel angle is Then, the scaled target front wheel angle can be expressed as: .
[0034] Among them, the first parameter Used to adjust the magnitude of the target front wheel angle.
[0035] S122: Delay the scaled target front wheel angle by the time corresponding to the second parameter to obtain the delayed target front wheel angle. Let the second parameter be... The scaled target front wheel angle will be Delay the second parameter After the corresponding time, the delayed target front wheel angle can be expressed as: .
[0036] Among them, the second parameter This is used to delay the output time of the target front wheel angle after amplitude correction. In discrete simulation implementation, the delay can be achieved through buffering, queuing, or other discretization methods that can achieve pure delay, as long as the scaled target front wheel angle is output after the time corresponding to the second parameter.
[0037] S123: Determine the single-step steering angle limiting threshold based on the third parameter and the simulation step size, and limit the expected single-step steering angle change to obtain the actual single-step steering angle increment; where the expected single-step steering angle change is the difference between the delayed target front wheel angle and the predicted actual front wheel angle of the previous simulation step. Let the third parameter be... The simulation step size is The single-step rotation angle limiting threshold L can then be determined based on the third parameter and the simulation step size. For example, multiplying the third parameter by the simulation step size yields the single-step rotation angle limiting threshold: .
[0038] Expected single-step rotation angle change The delayed target front wheel angle Compared with the predicted actual front wheel angle in the previous simulation step The difference can be expressed as: .
[0039] For the expected single-step rotation angle change By limiting the amplitude, the actual rotation angle increment per step is obtained. For example, the desired single-step rotation angle change can be limited to between positive and negative single-step rotation angle limiting thresholds using a symmetrical saturation limiting method. .
[0040] in, This indicates a limiting operation. When the expected single-step angle change exceeds the single-step angle limiting threshold range, the corresponding positive or negative single-step angle limiting threshold is used as the actual single-step angle increment; when the expected single-step angle change does not exceed the single-step angle limiting threshold range, the expected single-step angle change is used as the actual single-step angle increment.
[0041] S124: The predicted actual front wheel angle of the previous simulation step is superimposed with the single-step actual steering angle increment to obtain the predicted actual front wheel angle of the current simulation step, which can be expressed as: .
[0042] Connecting S121 to S124 in series, the overall discrete recurrence relation after connecting the three links in series can be expressed as: .
[0043] in, As the first parameter, For the second parameter, As the third parameter, For simulating step size, For the target front wheel angle, To predict the actual front wheel angle.
[0044] Through the above-described cascaded processing, each simulation step can recursively generate the predicted actual front wheel angle for the current simulation step based on the target front wheel angle, the first parameter, the second parameter, the third parameter, and the simulation step size. Specifically, the amplitude correction stage reflects the adjustment of the target front wheel angle in amplitude, the time delay stage reflects the delay in the command output time, and the rate limiting stage reflects the limiting effect of the single-step angle change constrained by the third parameter. These three stages are sequentially connected to generate the predicted actual front wheel angle, thus providing a basis for predicting the response of the agricultural machinery steering actuator.
[0045] In some embodiments, the first parameter is an amplitude scaling factor. The first parameter is used to characterize the steady-state following ratio of the predicted actual front wheel angle to the target front wheel angle under high-frequency command update conditions. The amplitude scaling factor ranges from greater than 0 to less than 1. The second parameter is the pure delay duration. This parameter characterizes the cumulative dead time from when the controller issues a steering command to when the steering actuator begins to move, including communication delay, drive response, and mechanical dead-zone start-up. The third parameter is the maximum rotational angular velocity. , is used to characterize the upper limit of the rate of a steering actuator constrained by physical power.
[0046] In control theory, the amplitude correction element is called a proportional element. Its transfer function is a constant and dimensionless function, which can be expressed as: Regardless of changes in the input signal, the output is always equal to the input multiplied by the first parameter. Its physical meaning is understeering. In the scenario of automatic navigation entry into a lane for agricultural machinery, the controller sends target front wheel angle commands at a fixed interval (e.g., 50ms) at high frequency. Due to the constraints of the actuator's speed limit V and transmission delay τ, the next command arrives before the previous command has finished executing—each target front wheel angle cannot actually be fully executed. The first parameter is the amplitude scaling factor. This describes the stable proportional reduction of the actual front wheel angle relative to the target front wheel angle under such high-frequency command updates. For example, when the first parameter is 0.99, an input of 10° will output 9.9°, meaning the amplitude is subject to a fixed "discount," independent of time. It should be noted that if only a single command is sent and a sufficiently long wait is allowed, the actual front wheel angle can be nearly achieved (scale ≈ 1); scale < 1 is the result of the combined effect of high-frequency command updates and a limited execution rate, rather than an inherent static deviation of the actuator. Because the control frequency and vehicle conditions are relatively stable, this understeer is calibrable.
[0047] In control theory, a time delay element can be called a pure time delay element, and its transfer function can be expressed as follows: , where s is the Laplace operator and τ is the delay time; its function is to make the output signal equal to the entire input signal shifted backward by τ time, with the waveform remaining unchanged. Its physical meaning is the command transmission dead time, that is, the cumulative delay from when the controller issues a steering command to when the actuator begins to move, including communication delays, driver response, and mechanical dead-zone start-up. For example, if the input changes at t=0, the output will not begin to change until t=τ, and the output remains unchanged for the time τ.
[0048] The rate limiting mechanism is a nonlinear saturated mechanism and cannot be described by a transfer function; it symmetrically limits the difference between the delayed target value and the current actual front wheel angle, and is a nonlinear operator. The actuator is constrained by the rated speed of the motor or the flow rate at the hydraulic valve orifice, and has a maximum angular velocity upper limit V (° / s), with each step change not exceeding [a certain value]. When the deviation is large, the front wheel angle approaches the target at a constant speed V (linear slope); when the deviation is small, it reaches the target in one step.
[0049] The reason why the above three parameters are independent is that their scopes do not overlap: It acts on the amplitude to characterize understeering under high-frequency command updates; It acts on time shift and is a pure transmission delay, which is independent of the amplitude. Acting on the upper bound of the time derivative, constraining The norm of the three parameters belongs to rate saturation nonlinearity. The three parameters have different dimensions: dimensionless, second, and degree per second, respectively. Mathematically, they are uncoupled and can therefore be interpreted independently. Compared to the industry practice of combining amplitude, delay, and rate into a single time constant for fitting, this application describes the three physical effects separately. During calibration, the three parameters do not interfere with each other, and the fitting is stable. Under the condition that the controller command update frequency and hardware configuration remain unchanged, the three parameters are stable throughout their entire lifecycle.
[0050] Substituting the three parameters into the discrete recursive formula, the overall discrete recursive formula after the three links are connected in series can be expressed by the following formula (1): Formula (1): .
[0051] Please see Figure 2 Target front wheel angle After sequential processing through amplitude correction, time delay, and rate limiting stages, the predicted actual front wheel angle is finally obtained. Each step first retrieves the delayed instruction (from the circular queue). The target front wheel angle value before the time step is calculated, then the "delayed target after understeer minus the actual value from the previous step" is calculated to get the desired change. Finally, the change is truncated using a clip. Within the range (meaning this step can only change this much at most), add it to the actual value of the previous step to get the actual front wheel angle of this step.
[0052] Please see Figure 3 , Figure 3 This diagram illustrates the single-beat step response, showcasing the interaction of the first, second, and third parameters. The horizontal axis represents time, and the vertical axis represents the front wheel angle. The target front wheel angle jumps from 0° to 20° in a single-beat step manner, with the predicted actual front wheel angle sequentially experiencing a pure delay phase, a linear climb phase, and a saturation-to-reach phase. During the pure delay phase, the predicted actual front wheel angle is affected by the second parameter... The time interval remains constant, reflecting the cumulative dead time from the controller issuing the steering command to the steering actuator starting to move. During the linear climb phase, the predicted actual front wheel angle approaches the target at a constant speed with the maximum rotational angular velocity corresponding to the third parameter V, reflecting the upper limit of the actuator's rate constrained by physical power. During the saturation positioning phase, the predicted actual front wheel angle reaches the product of the first parameter scale and the target front wheel angle and remains stable, reflecting the amplitude scaling caused by understeering under high-frequency command update conditions. This figure intuitively illustrates that the first, second, and third parameters act on the three dimensions of amplitude, time shift, and time derivative, respectively, and their physical meanings are independent and interpretable.
[0053] For example, assume scale = 0.99, V = 20° / s, τ = 90ms. =50ms (i.e., sampling period is 50ms), the target front wheel angle jumps from 0° to 20° step, where = 20° / s × 0.05s = 1° / step, meaning that each step can only rotate a maximum of 1°. That is, when the controller commands the front wheels to rotate 20°, there is no response for the first 90ms (pure delay), then it climbs at a constant speed of 1° per step (rate limiting), eventually stopping at 19.8° (amplitude scaling). It should be noted that... Figure 3 The diagram shows a single-cycle operation where only one command is sent and execution is awaited. In the high-frequency command update operation, due to continuous command updates, the first parameter, scale, will be less than 1, which characterizes stable understeering. The model parameters provided in this application have clear, traceable, and interpretable physical meanings, overcoming the shortcomings of two-parameter models where the physical meanings are ambiguous and uninterpretable, facilitating independent verification and engineering debugging.
[0054] The definitions of the three parameters and the variables used in the recursive formula are shown in Table 1 below.
[0055] Table 1. Definitions of the three parameters and variables in the recursive formula
[0056] In some embodiments, the amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity are predetermined by the following steps S201 to S203.
[0057] S201: Obtain calibration log data. The calibration log data includes the measured values of the target front wheel angle and the actual front wheel angle at multiple time points. The calibration log data can be obtained from the actual vehicle logs collected during the engineering phase, and its sample set can be represented as: ,in, k=1,...,N 0, N 0 represents the number of sample logs.
[0058] S202: Using the amplitude scaling factor, pure delay duration, and maximum rotational angular velocity as variables to be optimized, and minimizing the root mean square error between the predicted actual front wheel angle and the measured actual front wheel angle as the optimization objective, an error function is constructed. The predicted actual front wheel angle is generated recursively based on the measured target front wheel angle value in the calibration log data, according to the response prediction method of any of the aforementioned embodiments. This error function can be expressed as: .
[0059] in, Indicates Using the above discrete recursive formula (1) as input, the predicted actual front wheel angle is generated recursively, and the three constant calibration is transformed into least squares optimization. Minimizing the above squared error is equivalent to minimizing the root mean square error.
[0060] S203: Perform joint optimization on the variables to be optimized, minimizing the error function based on constraints, to obtain the magnitude scaling factor, pure delay duration, and maximum rotational angular velocity; where the constraints include: the magnitude scaling factor is greater than 0 and not greater than 1, and the pure delay duration and maximum rotational angular velocity are both positive numbers, i.e. , , .
[0061] In some embodiments, As a parameter vector, joint optimization is performed by minimizing the root mean square error between the model's predicted values and the measured values. The advantage of joint optimization is that the three parameters are already coupled in the same recursive formula in the physical model, and joint optimization can search for a better solution, avoiding the error accumulation that may be introduced by step-by-step fitting.
[0062] Assuming the controller command update frequency and hardware configuration remain unchanged, the three parameters are determined by the system characteristics: Controller command update frequency and execution rate The ratio relationship determines this; From the maximum power of the motor With reduction ratio Decision, for example ; This is determined by the communication link and the driver response. Assuming the system configuration remains constant, then: , , .
[0063] That is, under the condition that the controller command update frequency and hardware configuration remain unchanged, the three parameters are time-invariant constants and do not require online identification. The time-varying nature of online identification stems from parameter drift caused by model structure mismatch; when the model structure is correct, the parameters themselves do not exhibit time variation.
[0064] Therefore, the computational load during operation is extremely low, there is no identifier, only forward simulation is required, making it suitable for low-cost deployment; there are no risks of drift, convergence, and divergence in online identification; recalibration is only required when changing vehicles or steering mechanisms, making it particularly suitable for batch deployment scenarios such as agricultural machinery.
[0065] The definitions of the variables used in the offline calibration process are shown in Table 2 below.
[0066] Table 2 Definition of Variables in Offline Calibration Process
[0067] In some embodiments, the joint optimization solution employs a gradient-free optimization algorithm. A gradient-free optimization algorithm refers to a direct search algorithm that does not rely on the gradient information of the objective function, and is suitable for situations where the objective function is not smooth or the gradient is difficult to obtain analytically. In this embodiment, the objective function is the root mean square error generated recursively by formula (1), which is related to... The gradient is difficult to express explicitly, so a gradient-free search is used to directly minimize the root mean square error of the complete model on all training data. Therefore, no gradient calculation is required, making it simple to implement, with low requirements for the smoothness of the objective function, and applicable to parameter search of recursive models with three coupled parameters.
[0068] In some embodiments, the gradient-free optimization algorithm is the Nelder-Mead simplex optimization algorithm. The Nelder-Mead simplex optimization algorithm is a classic gradient-free direct search algorithm: it maintains a simplex in the parameter space, i.e., n+1 vertices, and iteratively approximates the optimal solution through geometric operations such as reflection, expansion, contraction, and edge shrinking. The embodiments of this application use... For vertex coordinates, under constraints , The inner search finds the three parameters that minimize the root mean square error. The Nelder-Mead simplex optimization algorithm requires no gradients, is easy to implement in low-cost controller calibration tools, and can obtain optimal estimates of the three parameters.
[0069] In some embodiments, the step of delaying the second parameter of the scaled target front wheel angle in step S122 includes the following sub-steps S1221 to S1224.
[0070] S1221: Determine the number of delay steps based on the second parameter and the simulation step size; the number of delay steps is greater than or equal to 1. Let the second parameter be... The simulation step size is The controller every Executing one step per second, i.e., running in discrete simulation steps, means it cannot be directly calculated in continuous time. Therefore, it is necessary to add a "delay" to the continuous time. The concept of "seconds" is transformed into "delay N steps" in discrete operations. Here, N represents the number of delay steps, indicating the duration of the continuous delay. Translated as an integer representing the number of discrete control steps.
[0071] According to the second parameter and simulation step size The number of delay steps N can be expressed as: ,and .
[0072] in, This indicates rounding to the nearest integer, because the controller can only execute in integer steps. Lower bound Ensure at least a one-step delay, even if It is very small and zero latency is not allowed, thus avoiding the pointless indexing of subsequent storage locations.
[0073] S1222: In the current simulation step, determine the storage location based on the step number and delay step number of the current simulation step. Let the step number of the current simulation step be i, and the delay step number be N. Determine the storage location based on the step number i and the delay step number N. The storage location is used to read the historical scaled target front wheel angle from the storage space and to write the scaled target front wheel angle corresponding to the current simulation step.
[0074] In some embodiments, the storage space is a circular queue; determining the storage location based on the step number of the current simulation step and the delay step number includes: calculating the remainder obtained by dividing the step number of the current simulation step by the delay step number; and using the remainder as an index to locate the storage location in the circular queue. That is: .in, For storage location index, This is for the modulo operation. A circular queue is a fixed-length array of length N, used to implement "looking back at the input N steps ago" with minimal memory. For example, a circular queue can be represented as: It should be noted that the most direct way to achieve "looking back at the input N steps ago" is to maintain an array of length N and shift the old value one position to the right each time it is written, which has a time complexity of O(N). The embodiment of this application uses a circular queue, and both writing and reading only require one array operation, with a time complexity of O(1), thus achieving pure delay with minimal memory and minimal computation.
[0075] S1223: Read the historically scaled target front wheel angle from the storage location in the storage space as the delayed target front wheel angle; the storage space is used to store the historically scaled target front wheel angle. In the current simulation step, reading the historically scaled target front wheel angle from the storage location in the storage space as the delayed target front wheel angle can be expressed as: .
[0076] in, This represents the historical scaled target front wheel angle read from the circular queue and written N steps ago.
[0077] According to the discrete delay relation, the output of the i-th step should be equal to the input of the (iN)-th step, which can be expressed as: .
[0078] For example, when N=2, if the input jumps from 0° to 20° in the second step, the output will only become 20° in the fourth step because the output needs to take the value from two steps ago, always lagging behind the input by two steps.
[0079] S1224: Write the scaled target front wheel angle corresponding to the current simulation step to the storage location in the storage space. After reading the historical scaled target front wheel angle from the storage location, writing the scaled target front wheel angle corresponding to the current simulation step to the same storage location can be represented as: .
[0080] It should be noted that the above read and write operations refer to the same storage location, and reading should precede writing. That is, read first, then write. The system reads the historical scaled target front wheel angle written N steps ago and uses it as the delayed target front wheel angle for the current simulation step; then it writes the scaled target front wheel angle corresponding to the current simulation step back into the system. This is done to overwrite the historical value for subsequent reading in step (i+N).
[0081] The key lies in This circular index. By algebraic identity: .
[0082] It can be seen that step i and step (i+N) access the same slot. The current scaled target front wheel angle written in step i will not be overwritten by a write at the same position until step (i+N); before that, the read operations from step (i+1) to step (i+N-1) access other slots and will not touch this value. Therefore, what is read is always the value written N steps ago, which can be represented as: .
[0083] In some embodiments, at the first simulation step, i.e., when i=0, the circular queue is not yet filled with historical data. To avoid false transients caused by reading zero values, each slot in the circular queue can be initialized to the scaled target front wheel angle corresponding to the current simulation step, assuming the system is in a steady state before the delay begins, thus ensuring a smooth start. For example, at i=0, the... All values are filled with the current command value.
[0084] This application embodiment precisely implements a fixed-length circular queue and a modular index. With pure delay, each step requires only one array read and one array write, resulting in a time complexity of O(n log n). Its z-transfer function is Its amplitude is always 1, and its phase is strictly linear, differing from a continuous pure delay only by a factor of 1. This method reduces discrete quantization errors. Compared to the Padé approximation or all-pole approximation, this implementation has no amplitude attenuation or phase nonlinearity errors, which is particularly advantageous when there are many delay steps, making it suitable for real-time operation of low-cost ECUs. It should be noted that the related technology uses the all-pole approximation. Using this approximation to simulate pure delay introduces amplitude attenuation and phase nonlinearity errors; the embodiments of this application do not use this approximation, but instead use a fixed-length circular queue to accurately realize the above-mentioned N-step pure delay.
[0085] The definitions of the variables used in the pure delay stage and the circular queue are shown in Table 3 below.
[0086] Table 3 Definitions of variables for pure delay elements and circular queues
[0087] In some embodiments, the step S123 above, which determines the single-step rotation angle limiting threshold based on the third parameter and the simulation step size, limits the expected single-step rotation angle change, and obtains the single-step actual rotation angle increment, includes the following sub-steps S1231 to S1233.
[0088] S1231: When the expected single-step angle change is greater than or equal to the single-step angle limiting threshold, the single-step angle limiting threshold is used as the actual single-step angle increment. If the desired rotation speed is too high, but the actuator is constrained by the third parameter V and cannot rotate that fast, then this simulation step can only rotate at full speed in the positive direction, i.e.: .
[0089] S1232: When the expected single-step angle change is less than or equal to the negative value of the single-step angle limiting threshold, the negative value of the single-step angle limiting threshold is used as the actual single-step angle increment. If the desired reverse rotation is excessive, and the actuator is constrained by the third parameter V, then this simulation step can only rotate at full speed in the negative direction, i.e.: .
[0090] S1233: When the expected single-step angle change is between the negative value of the single-step angle limiting threshold and the single-step angle limiting threshold, the expected single-step angle change is used as the actual single-step angle increment. If the expected value is achievable within the current simulation step, then the simulation step will directly jump to the target value, i.e.: .
[0091] The above three branches can be uniformly represented as a symmetrical limiting operation, that is: .
[0092] Then, the predicted actual front wheel angle of the previous simulation step is superimposed with the single-step actual steering angle increment to obtain the predicted actual front wheel angle of the current simulation step: .
[0093] Its equivalent continuous form, a rate-saturated system with pure delay, can be expressed as: .
[0094] in, It is a symmetric saturation operator.
[0095] The aforementioned continuous form is a rate-saturated nonlinear element with pure delay, belonging to the composite nonlinearity of saturation operator and dead time in control theory.
[0096] The definitions of the variables used in the rate limiting stage are shown in Table 4 below.
[0097] Table 4 Definition of Variables for Rate Limiting Components
[0098] Let the target front wheel angle step input be... ,and The step response of the response prediction method provided in this application embodiment can be analytically obtained as a piecewise function: .
[0099] Wherein, 0≤t<τ is the dead time phase, in which the delayed target front wheel angle is 0, and the predicted actual front wheel angle remains 0; τ≤t<t_sat is the linear ramp phase, in which the expected single-step angle change amount is greater than the single-step angle limiting threshold, and the predicted actual front wheel angle rises at a constant rate V, that is ; t≥t_sat is the saturation in-position phase, in which the predicted actual front wheel angle reaches and then enters a steady state; wherein .
[0100] Please refer to Figure 4 , Figure 4 for comparing the step response shapes of the response prediction model of the embodiment of the present application and the conventional first-order inertia model. In the figure, the abscissa is time, and the ordinate is the front wheel angle; the target front wheel angle jumps from 0° to 20° in a step manner.
[0101] The red curve in the figure is the response of the conventional first-order inertia model, which is characterized by the maximum initial change rate, approximately , then decelerates gradually, asymptotically approaches the target value in an exponential form, and the final steady state is approximately 19.8°. The blue curve in the figure is the response prediction model of the embodiment of the present application, which first undergoes a pure delay of approximately , then climbs linearly at a constant maximum rotational angular velocity , and stops after reaching approximately 19.8°. The example parameters in the figure are: for the embodiment of the present application , , ; for the first-order inertia model , . The two have the same final steady-state amplitude, but different dynamic response shapes.
[0102] It can be seen from the comparison of Figure 4 that the response shape of the embodiment of the present application is "pure delay + linear ramp + step saturation", which is more consistent with the real physical behavior of the steering actuator under the constraint of maximum power, that is, "approaching the target at a constant maximum rate and stopping quickly after reaching the position"; while the exponential asymptotic response of the first-order inertia model is too fast in the initial stage, which easily leads to overestimation of the short-term front wheel angle prediction under the high-frequency command update working condition. Therefore, the embodiment of the present application is beneficial to improving the prediction accuracy of the front wheel angle.
[0103] In some embodiments, the above single-step limiting process can be implemented by the following code:
[0104] In some embodiments, the above-described response prediction method further includes: predicting the driving trajectory of the agricultural machinery based on the predicted actual front wheel angle and a preset Ackerman kinematic model. The Ackerman kinematic model refers to a kinematic model that extrapolates the trajectory of the vehicle's rear axle center from the front wheel steering angle, and is the core sub-model for predicting the trajectory of the agricultural machinery's automatic navigation. In this embodiment, the predicted actual front wheel angle, rather than the target front wheel angle, is used as the input to the Ackerman kinematic model, ensuring that the trajectory extrapolation is based on the "actual steering response that will occur" rather than the "ideal expected steering," thereby reducing trajectory deviations caused by understeer and delay.
[0105] After obtaining the predicted actual front wheel angle for the current simulation step Then, the data is substituted into a preset Ackerman kinematics model, combined with the current pose of the agricultural machinery, including its position and heading, to continuously predict the vehicle's trajectory over a future period, providing accurate trajectory prediction input for the control strategy. Through the Ackerman kinematics model, the pose for the next simulation step can be recursively obtained based on the predicted actual front wheel angle of the current simulation step, thus continuously predicting the vehicle's trajectory over a future period.
[0106] Therefore, the embodiments of this application connect the response prediction model with Ackerman kinematics, so that the trajectory prediction is based on the actual front wheel angle response, avoiding the problem of short-term overestimation of first-order inertial models under high-frequency command updates, thereby improving the alignment quality of agricultural machinery entering the line and the accuracy of automatic navigation.
[0107] Figure 5 This is a real-vehicle log verification diagram of the steering response under high-frequency command update conditions, used to illustrate the prediction effect of the response prediction method described in this application embodiment in actual agricultural machinery operation. The horizontal axis in the diagram represents time, and the vertical axis represents the front wheel angle. The black dashed line in the diagram represents the target front wheel angle. This refers to the desired front wheel angle output by the controller at high frequency; the solid blue line represents the actual front wheel angle. The measured front wheel angle of the actuator; the solid green line represents the predicted value of the three-constant model, i.e., based on the first parameter. Second parameter and the third parameter The predicted actual front wheel angle is obtained through recursion. Below is the model prediction error curve, which is the difference between the model prediction value and the measured value. The root mean square error (RMSE) is indicated in the figure as approximately 1.260°. In the example, the sampling period for the real vehicle log is approximately... , First parameter It is used to characterize understeering under high-frequency command update conditions; the third parameter V The first parameter is used to characterize the maximum rotational angular velocity of the actuator; the second parameter... , used to characterize the pure delay of command transmission; the target front wheel angle is There were 39 internal jumps.
[0108] Depend on Figure 5 It is evident that under high-frequency command update conditions, the target front wheel angle changes frequently, while the actual front wheel angle is always in a chasing state, indicating calibrable understeer. Figure 5 The annotation "The target has bounced back, but it is still catching up with the previous step" intuitively illustrates the lag catch-up phenomenon caused by the pure delay corresponding to the second parameter and the rate limit corresponding to the third parameter. The predicted curve and the measured curve of the three-constant model in this application embodiment are highly consistent in terms of steady-state amplitude, delay start point, and ramp slope, verifying the accuracy and interpretability of the three-constant physical separation modeling under high-frequency command update conditions.
[0109] This application discloses a method, system, and control device for predicting the response of an agricultural machinery steering actuator. The method acquires the target front wheel angle and processes it through a series of amplitude correction, time delay, and rate limiting steps to generate a predicted actual front wheel angle. The amplitude correction step adjusts the amplitude of the target front wheel angle according to a first parameter; the time delay step delays the output time of the amplitude-corrected target front wheel angle according to a second parameter; and the rate limiting step limits the expected single-step angle change according to a third parameter. This application replaces the first-order inertial response with a speed- and amplitude-limited linear ramp response, decomposing the dynamic response of the steering actuator into three independent physical effect steps: understeering, pure delay, and rate limit. These are modeled separately and recursively cascaded, enabling accurate prediction of the front wheel angle response under high-frequency command update conditions. The model parameters have clear and independent physical meanings.
[0110] The response prediction method for agricultural machinery steering actuators provided in this application also has the following beneficial effects: I. Improved accuracy of front wheel angle prediction. The embodiments of this application replace the exponential asymptotic response of first-order inertia with a clip-limited linear ramp response, which is closer to the physical behavior of the actual actuator under high-frequency command update conditions, which is "limited by the maximum speed and approaches the target at a constant speed". At the same time, the understeer effect is independently modeled as a constant scale, so that the part of the front wheel angle of each target that is "not executed in place" can be accurately quantified, eliminating the prediction bias caused by the scale being implicit in the time constant.
[0111] Second, the physical meaning of the model parameters is clearly defined. The first, second, and third parameters correspond to three independent physical effects: scale (understood steering), V (upper limit of actuator angular velocity), and τ (pure delay). The calibration results can be traced back to specific physical components, facilitating independent verification and engineering debugging, and overcoming the shortcomings of the two-parameter model in that its physical meaning is ambiguous and uninterpretable.
[0112] Third, the pure delay has small implementation error and extremely low computational cost. The N-step pure delay is accurately implemented with a fixed-length circular queue and modular indexing. Each step only requires one array read and one array write, with a time complexity of O(1). Compared with the Padé approximation or all-pole approximation, there is no polynomial operation overhead and no high-frequency phase error. Its advantages are particularly obvious when there are many delay steps, making it suitable for low-cost ECU real-time operation.
[0113] IV. Offline calibration once, stable throughout the entire lifecycle, and no need for online identification. Under the condition that the controller command update frequency and hardware configuration remain unchanged, the three parameters are essentially stable system constants; calibration is completed once through offline logs during the engineering phase and is solidified into constants during operation, eliminating the drift, convergence, and divergence risks associated with online identification; recalibration is only required when changing vehicles or steering mechanisms, making it particularly suitable for batch deployment scenarios such as agricultural machinery.
[0114] To facilitate better implementation of the response prediction method for agricultural machinery steering actuators provided in this application embodiment, this application embodiment also provides a response prediction system 100 for agricultural machinery steering actuators based on the above-described response prediction method. The meanings of the terms used are the same as in the above-described response prediction method for agricultural machinery steering actuators, and specific implementation details can be found in the descriptions in the method embodiments.
[0115] Please see Figure 6 The response prediction system 100 for the agricultural machinery steering actuator may include: Data acquisition module 101 is used to acquire the target front wheel angle; The amplitude correction module 102 is used to adjust the amplitude of the target front wheel angle according to the first parameter; The time delay module 103 is used to delay the output time of the target front wheel angle after amplitude correction according to the second parameter; Rate limiting module 104 is used to limit the expected single-step angle change of the steering actuator according to the third parameter; The prediction calculation module 105 is used to generate the predicted actual front wheel angle based on the output result after the series processing of the amplitude correction module 102, the time delay module 103 and the rate limiting module 104.
[0116] Since the above-mentioned response prediction system for agricultural machinery steering actuators can be used to execute the above-mentioned response prediction method for agricultural machinery steering actuators, for further descriptions of the various modules in the above-mentioned response prediction system for agricultural machinery steering actuators and their beneficial effects, please refer to the embodiments of the above-mentioned response prediction method for agricultural machinery steering actuators, which will not be elaborated here.
[0117] This application also provides an agricultural machinery control device, which can be used to implement the methods described in the above method embodiments. The agricultural machinery control device may be a chip, a terminal device, or a server.
[0118] Agricultural machinery control equipment may include one or more processors. The processor can support the implementation of the methods described in the preceding method embodiments. The processor can be a general-purpose processor or a special-purpose processor. For example, the processor can be a Central Processing Unit (CPU). Alternatively, the processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.
[0119] Agricultural machinery control equipment may also include one or more memories. The memories store computer programs. The memories can be independent of the processor or integrated into the processor.
[0120] Agricultural machinery control equipment may also include transceivers. Processors can communicate with other devices or chips via transceivers. For example, processors can send and receive data with other devices or chips via transceivers.
[0121] The computer program in memory can be executed by a processor, causing the processor to perform the steps in any of the above-mentioned response prediction methods for agricultural machinery steering actuators.
[0122] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.
[0123] Therefore, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which is loaded by a processor to execute the steps described in the above-described method embodiments of this application. For example, the computer program, when loaded by a processor, can execute the steps in any of the above-described methods for predicting the response of an agricultural machinery steering actuator.
[0124] For details on the implementation of each of the above operations / steps, please refer to the previous examples, which will not be repeated here.
[0125] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.
[0126] Since the computer program stored in the computer-readable storage medium can execute the steps in any of the above method embodiments provided in the embodiments of this application, the beneficial effects that the methods described in any of the above method embodiments can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.
[0127] This application also provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in the various embodiments described above.
[0128] The above provides a detailed description of the response prediction method, system, and agricultural machinery control equipment for an agricultural machinery steering actuator provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for predicting the response of a steering actuator in agricultural machinery, characterized in that, include: Obtain the target front wheel angle; The target front wheel angle is processed in series with amplitude correction, time delay and rate limiting to generate the predicted actual front wheel angle; The amplitude correction stage is used to adjust the amplitude of the target front wheel angle according to the first parameter; the time delay stage is used to delay the output time of the amplitude-corrected target front wheel angle according to the second parameter; and the rate limiting stage is used to limit the expected single-step angle change according to the third parameter. The serial processing includes: scaling the target front wheel angle using the first parameter to obtain a scaled target front wheel angle; delaying the scaled target front wheel angle by the duration corresponding to the second parameter to obtain a delayed target front wheel angle; determining a single-step angle limiting threshold based on the third parameter and the simulation step size, and limiting the expected single-step angle change to obtain a single-step actual angle increment; wherein the expected single-step angle change is the difference between the delayed target front wheel angle and the predicted actual front wheel angle of the previous simulation step; and superimposing the predicted actual front wheel angle of the previous simulation step with the single-step actual angle increment to obtain the predicted actual front wheel angle of the current simulation step.
2. The method according to claim 1, characterized in that, The step of delaying the scaled target front wheel angle by the duration corresponding to the second parameter includes: The number of delay steps is determined based on the second parameter and the simulation step size; the delay step is greater than or equal to 1. In the current simulation step, the storage location is determined based on the step number of the current simulation step and the number of delay steps; The historical scaled target front wheel angle is read from the storage location in the storage space and used as the delayed target front wheel angle; the storage space is used to store the historical scaled target front wheel angle. Write the scaled target front wheel angle corresponding to the current simulation step to the storage location in the storage space.
3. The method according to claim 2, characterized in that, The storage space is a circular queue; determining the storage location based on the step number of the current simulation step and the delay step number includes: Calculate the remainder obtained by dividing the current simulation step number by the number of delay steps; The remainder is used as an index to locate the storage location in the circular queue.
4. The method according to claim 1, characterized in that, The step of determining the single-step rotation angle limiting threshold based on the third parameter and the simulation step size, and limiting the expected single-step rotation angle change to obtain the actual single-step rotation angle increment, includes: When the expected single-step angle change is greater than or equal to the single-step angle limiting threshold, the single-step angle limiting threshold is used as the single-step actual angle increment. When the expected single-step angle change is less than or equal to the negative value of the single-step angle limiting threshold, the negative value of the single-step angle limiting threshold is taken as the single-step actual angle increment. When the expected single-step angle change is between the negative value of the single-step angle limiting threshold and the single-step angle limiting threshold, the expected single-step angle change is used as the single-step actual angle increment.
5. The method according to any one of claims 1 to 4, characterized in that, The first parameter is an amplitude scaling factor, which is used to characterize the steady-state following ratio of the predicted actual front wheel angle to the target front wheel angle under high-frequency command update conditions. The value range of the amplitude scaling factor is greater than 0 and not greater than 1. The second parameter is the pure delay duration, which is used to characterize the cumulative dead time from when the controller issues a steering command to when the steering actuator begins to move; The third parameter is the maximum rotational angular velocity, which is used to characterize the upper limit of the rate of the steering actuator constrained by physical power.
6. The method according to claim 5, characterized in that, The amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity are predetermined through the following steps: Obtain calibration log data, which includes the measured values of the target front wheel angle and the actual measured values of the front wheel angle at multiple times; Using the amplitude scaling factor, the pure delay duration, and the maximum rotational angular velocity as variables to be optimized, and minimizing the root mean square error between the predicted actual front wheel angle and the measured actual front wheel angle as the optimization objective, an error function is constructed. The variables to be optimized are jointly optimized and the error function is minimized based on the constraints to obtain the magnitude scaling factor, the pure delay duration and the maximum rotational angular velocity; The constraints include: the amplitude scaling factor is greater than 0 and not greater than 1; the pure delay duration and the maximum rotational angular velocity are both positive numbers.
7. The method according to claim 6, characterized in that, The joint optimization solution employs a gradient-free optimization algorithm.
8. The method according to claim 7, characterized in that, The gradient-free optimization algorithm is the Nelder-Mead simplex optimization algorithm.
9. The method according to claim 1, characterized in that, The method further includes: Based on the predicted actual front wheel angle and the preset Ackerman kinematic model, the driving trajectory of the agricultural machinery is predicted.
10. A response prediction system for an agricultural machinery steering actuator, characterized in that, include: The data acquisition module is used to obtain the target front wheel angle; An amplitude correction module is used to adjust the amplitude of the target front wheel angle according to a first parameter; The time delay module is used to delay the output time of the target front wheel angle after amplitude correction according to the second parameter; A rate limiting module is used to limit the expected single-step angle change of the steering actuator according to a third parameter; The prediction calculation module is used to generate a predicted actual front wheel angle based on the output result after the series processing of the amplitude correction module, the time delay module and the rate limiting module.
11. An agricultural machinery control device, characterized in that, It includes a memory and a processor, the memory storing a computer program or instructions, and the processor executing the computer program or instructions to implement the response prediction method for the agricultural machinery steering actuator as described in any one of claims 1 to 9.
12. A computer-readable storage medium storing a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the response prediction method for the agricultural machinery steering actuator as described in any one of claims 1 to 9.
