Angle fault-tolerant control method and system for forklift steering

Abstract

The application provides an angle fault-tolerant control method and system for forklift steering, and relates to the field of forklift steering control, wherein the method comprises the following steps: setting a desired steering angle sequence of the forklift, obtaining an actual steering angle data stream, performing deviation analysis based on the desired steering angle sequence and the actual steering angle data stream, introducing historical change trend data of a steering bridge to perform fault-tolerant judgment on a real-time angle deviation value, performing steering control analysis of the steering bridge based on a fault-tolerant judgment result to calculate a fault-tolerant compensation angle, and driving a forklift steering actuator to perform self-adaptive fault-tolerant control on the steering angle. The technical problem that the angle deviation is difficult to accurately control when the forklift steers, and there is a lack of effective fault-tolerant processing and self-adaptive compensation mechanism, which affects steering stability, is solved. The technical effects of accurate control and fault-tolerant processing on the forklift steering angle, improved steering stability and safety are achieved.

Description

Technical Field

[0001] This invention relates to the field of forklift steering control technology, specifically to an angle-tolerant control method and system for forklift steering. Background Technology

[0002] Forklifts are commonly used material handling equipment in warehousing, logistics, and industrial production. Their steering performance directly affects operational efficiency and safety. Existing forklift steering systems typically achieve steering angle control based on a preset control model and single sensor feedback. However, in actual operation, factors such as mechanical clearances, hydraulic system response lag, sensor noise, and environmental disturbances can cause deviations between the actual steering angle and the desired steering angle.

[0003] Most current steering control schemes focus on conventional feedback adjustment, and their methods for judging steering angle deviations are relatively crude. They struggle to distinguish between normal dynamic fluctuations and abnormal deviations in a timely manner. When sudden changes or accumulated errors occur in the steering angle, there is a lack of effective fault-tolerant mechanisms, which can easily lead to lag in control response or overcorrection. Furthermore, existing technologies generally lack trend analysis and adaptive compensation mechanisms based on historical operating data, making it difficult to dynamically adjust control strategies according to the actual operating conditions of the forklift. This, in turn, affects the smoothness and stability of the steering process and increases operational risks.

[0004] In summary, existing technologies suffer from technical problems such as difficulty in accurately controlling angle deviations during forklift steering, lack of effective fault tolerance and adaptive compensation mechanisms, and impact on steering stability. Summary of the Invention

[0005] The purpose of this application is to provide an angle-tolerant control method and system for forklift steering, which solves the technical problems in the prior art where angle deviation is difficult to control accurately during forklift steering, lacks effective fault tolerance processing and adaptive compensation mechanisms, and affects steering stability.

[0006] In view of the above problems, this application provides an angle-tolerant control method and system for forklift steering.

[0007] The first aspect of this application provides an angle-tolerant control method for forklift steering, the method comprising: setting a desired steering angle sequence for the forklift based on a forklift steering path planning module; performing real-time sensing of the forklift steering axle based on an angle sensor to obtain an actual steering angle data stream; performing deviation analysis based on the desired steering angle sequence and the actual steering angle data stream to calculate a real-time angle deviation value; introducing historical trend data of the steering axle to perform a fault-tolerant judgment on the real-time angle deviation value to generate a fault-tolerant judgment result; calculating a fault-tolerant compensation angle based on the fault-tolerant judgment result to perform steering control analysis of the steering axle, and generating a steering control command to drive the forklift steering actuator to perform adaptive fault-tolerant control of the steering angle.

[0008] Optionally, the forklift steering path planning module retrieves the forklift's navigation map and performs real-time trajectory calculation according to the forklift's operation task to generate a global reference path; based on the global reference path, steering tracking prediction is performed, and the desired steering angle sequence is set; a magnetic encoder is fixedly connected to the steering knuckle of the steering axle, and an angle sensor is fixed to the axle body opposite to the steering knuckle to generate an angle sensing network; the angle sensing network is used to read data in real time at a preset sampling frequency to generate the actual steering angle data stream.

[0009] Optionally, the expected steering angle sequence and the actual steering angle data stream are time-stamped and synchronized to determine a control timing reference data pair; the control timing reference data pair is traversed, and the expected steering angle minus the actual steering angle is extracted to obtain the original instantaneous angle deviation value, wherein the control timing of the expected steering angle and the actual steering angle is the same; a theoretical response lag threshold range is set, and it is determined whether the original instantaneous angle deviation exceeds the theoretical response lag threshold range; if the original instantaneous angle deviation value does not exceed the theoretical response lag threshold range, the original instantaneous angle deviation value is determined to be reasonable, and the original instantaneous angle deviation value is output as the real-time angle deviation value; if the original instantaneous angle deviation exceeds the theoretical response lag threshold range, the original instantaneous angle deviation value is marked as a suspicious value, and the instantaneous angle deviation value of the previous i-th moment is extracted based on the control timing of the expected steering angle and the actual steering angle as a valid deviation value for iterative determination, until the valid deviation value does not exceed the theoretical response lag threshold range, and then the valid deviation value is output as the real-time angle deviation value, where i is a positive integer less than or equal to 2.

[0010] Optionally, the desired steering angle sequence is traversed for global timing analysis to construct a global clock timestamp; the global clock timestamp is matched with the desired steering angle sequence to generate a first timestamp matching result; the actual steering angle data stream is traversed for timing analysis according to sampling points to construct a local clock timestamp; the local clock timestamp is matched with the actual steering angle data stream to generate a second timestamp matching result; the first timestamp matching result and the second timestamp matching result are clocked for synchronization correction to determine the control timing reference data pair.

[0011] Optionally, a sliding time window is set, and historical angle deviation values ​​are continuously stored according to the sliding time window to generate a historical angle deviation sequence; numerical difference calculation is performed on the historical angle deviation sequence to generate a deviation change rate sequence; periodic analysis is performed according to the deviation change rate sequence and the historical angle deviation sequence to generate the historical change trend data.

[0012] Optionally, the stable operating condition information of the forklift is identified, and historical angle deviation sequences are matched according to the stable operating conditions to construct training samples; time series analysis is performed using the training samples to construct a dynamic behavior benchmark model; the real-time angle deviation value is synchronized to the dynamic behavior benchmark model for consistency comparison to obtain an anomaly score; fault tolerance correlation analysis is performed based on the anomaly score, multi-level fault tolerance triggering conditions are set, and fault tolerance judgment is performed according to the multi-level fault tolerance triggering conditions combined with the anomaly score to generate a fault tolerance judgment result.

[0013] Optionally, the predicted deviation value of the dynamic behavior benchmark model is extracted according to the timestamp of the real-time angle deviation value; the residual parameter is calculated by comparing the real-time angle deviation value with the predicted deviation value; a confidence interval is set based on the dynamic behavior benchmark model; the excess value of the residual parameter and the confidence interval is calculated, and a first anomaly index is set based on the excess value; the change of the real-time angle deviation value is analyzed by the dynamic behavior benchmark model according to a fixed time period to determine the short-term change pattern; a normal change pattern is extracted based on the historical change trend data, and the matching degree of the normal change pattern and the short-term change pattern is calculated to set a second anomaly index; the first anomaly index and the second anomaly index are weighted and calculated to generate the anomaly score.

[0014] Optionally, fluctuation analysis is performed on the historical trend data based on the anomaly score, adaptive scaling is performed according to the fluctuation results, and multi-level fault tolerance triggering conditions are set. The multi-level fault tolerance triggering conditions include a first dynamic threshold and a second dynamic threshold, where the first dynamic threshold is lower than the second dynamic threshold. When the anomaly score is lower than the first dynamic threshold, the steering angle state is normal, and backtracking is performed according to the anomaly score to generate a first fault tolerance judgment result. When the anomaly score is between the first dynamic threshold and the second dynamic threshold, the steering angle state deviates, and continuous monitoring is performed according to the anomaly score to generate a second fault tolerance judgment result. When the anomaly score is higher than the second dynamic threshold, the steering angle state is abnormal, and strong fault tolerance intervention is performed according to the anomaly score to generate a third fault tolerance judgment result.

[0015] Optionally, when the fault tolerance judgment result is the first fault tolerance judgment result, feedforward compensation is performed based on the steering angle state to generate a first fault tolerance compensation angle; when the fault tolerance judgment result is the second fault tolerance judgment result, feedback-feedforward composite compensation is performed based on the steering angle state to generate a second fault tolerance compensation angle; when the fault tolerance judgment result is the third fault tolerance judgment result, a deep fault tolerance mode is triggered: S1: Construct a sliding surface based on the real-time angle deviation value, perform gain analysis on the sliding surface according to the third fault tolerance judgment result, and calculate the equivalent control quantity; S2: Set the third fault tolerance compensation angle according to the equivalent control quantity.

[0016] A second aspect of this application provides an angle-tolerant control system for forklift steering, the system comprising: a data acquisition component for setting a desired steering angle sequence for the forklift based on a forklift steering path planning module, and for real-time sensing of the forklift steering axle based on an angle sensor to obtain an actual steering angle data stream; a deviation analysis component for performing deviation analysis based on the desired steering angle sequence and the actual steering angle data stream, and calculating a real-time angle deviation value; a fault-tolerant judgment component for introducing historical change trend data of the steering axle to perform fault-tolerant judgment on the real-time angle deviation value, and generating a fault-tolerant judgment result; and a fault-tolerant control component for calculating a fault-tolerant compensation angle based on the fault-tolerant judgment result, performing steering control analysis of the steering axle, and generating steering control commands to drive the forklift steering actuator to perform adaptive fault-tolerant control of the steering angle.

[0017] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0018] The method provided in this application embodiment sets the desired steering angle sequence of the forklift based on the forklift steering path planning module, and obtains the actual steering angle data stream by real-time sensing of the forklift steering axle based on angle sensors; it performs deviation analysis based on the desired steering angle sequence and the actual steering angle data stream to calculate the real-time angle deviation value; it introduces historical change trend data of the steering axle to perform fault tolerance judgment on the real-time angle deviation value and generates a fault tolerance judgment result; it calculates the fault tolerance compensation angle based on the fault tolerance judgment result to perform steering control analysis of the steering axle, and generates steering control commands to drive the forklift steering actuator to perform adaptive fault tolerance control of the steering angle. This achieves the technical effect of precise control and fault tolerance processing of the forklift steering angle, improving steering stability and safety.

[0019] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the angle-tolerant control method for forklift steering provided in this application.

[0022] Figure 2 This is a schematic diagram of the structure of the angle-tolerant control system for forklift steering provided in this application.

[0023] Figure labeling: Data acquisition component 11, deviation analysis component 12, fault tolerance judgment component 13, fault tolerance control component 14. Detailed Implementation

[0024] This application provides a fault-tolerant control method and system for forklift steering angles, addressing the technical problems in existing technologies where it is difficult to accurately control angle deviations during forklift steering, and the lack of effective fault-tolerant processing and adaptive compensation mechanisms affects steering stability. It achieves precise control and fault-tolerant processing of the forklift steering angle, improving steering stability and safety.

[0025] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0026] Example 1, as Figure 1As shown, this application provides an angle-tolerant control method for forklift steering, the angle-tolerant control method for forklift steering includes: The desired steering angle sequence of the forklift is set based on the forklift steering path planning module, and the actual steering angle data stream is obtained by real-time sensing of the forklift steering axle based on the angle sensor.

[0027] Furthermore, the method for setting the desired steering angle sequence of the forklift based on the forklift steering path planning module and obtaining the actual steering angle data stream by real-time sensing of the forklift steering axle based on the angle sensor includes: retrieving the forklift's navigation map based on the forklift steering path planning module to perform real-time trajectory calculation according to the forklift's operation task and generate a global reference path; performing steering tracking prediction based on the global reference path and setting the desired steering angle sequence; fixing a magnetic encoder to the steering knuckle of the steering axle and fixing the angle sensor to the axle body opposite the steering knuckle to generate an angle sensing network; and generating the actual steering angle data stream by real-time reading through the angle sensing network at a preset sampling frequency.

[0028] Specifically, the forklift steering path planning module consists of a path planning algorithm and a navigation system integrated into the forklift. The navigation system includes map data, environmental information, and the forklift's real-time position and motion status. Environmental information includes obstacles and terrain, and the path planning algorithm includes A / B. Path planning algorithms, such as Algorithm A and Dijkstra's algorithm, primarily function to calculate an optimal trajectory for the forklift based on its task, working environment, and target location, and adjust the steering angle in real time. The forklift steering path planning module first retrieves a navigation map of the forklift's environment based on its current and target positions. This map contains key information such as the terrain and obstacle distribution of the forklift's work area. Combined with the task received by the forklift, which includes, but is not limited to, the direction of movement, the shape of the work area, and the work route (e.g., moving goods from point A to point B in a warehouse), the forklift steering path planning module uses a path planning algorithm such as Algorithm A... Algorithms such as Dijkstra's algorithm are used to perform real-time trajectory calculations and generate a safe and efficient global reference path. The global reference path is the ideal path that the forklift needs to follow during the entire operation task. It includes every turning point and direction of movement of the forklift from the starting point to the target point. The global reference path is a discrete set of trajectory points, and each point corresponds to a certain time point and position during the movement of the forklift.

[0029] For example, using A The algorithm performs real-time trajectory calculation, discretizing the navigation map of the forklift's environment into a grid, for example, using a two-dimensional grid map to represent the environment, where each grid cell represents a region. On this two-dimensional grid map, passable areas and obstacle areas are marked, and each grid cell is represented by coordinates (x, y), indicating the forklift's position. Combined with the forklift's task, the current position and target position of the forklift are marked as the start and end points, respectively. The algorithm calculates the cost for each grid cell. The total cost f(x,y) for each node consists of two parts: g(x,y) is the actual cost from the starting point to the current node (x,y), typically the number of steps or distance from the starting point to that node; h(x,y) is the heuristic cost from the current node (x,y) to the target node, using Manhattan distance or Euclidean distance to estimate the straight-line distance. The algorithm formula is: f(x,y)=g(x,y)+h(x,y), where g(x,y) is the actual cost of the current path, and h(x,y) is the estimated cost from the current position to the target position. The algorithm selects the node with the lowest total cost f(x,y) for expansion and adds it to the list of expanded nodes. Then, it explores the surrounding neighboring nodes starting from this node, calculating the cost of each neighboring node. The calculation process considers obstacles, walls, and forklift turning constraints. It iterates through the nodes until the target location is found. When the target location is found, A... The algorithm starts from the target node and backtracks along the parent node of each node, i.e. the best path from the starting point to the target, to finally obtain an optimal path. The optimal path consists of a series of path points, including (x1, y1), (x2, y2)...(xn, yn), which represent the movement route of the forklift from the starting point to the target. This optimal path is used as the global reference path for the forklift.

[0030] Steering tracking prediction is performed based on a global reference path. Specifically, the steering angle can be calculated using path points (x1, y1), (x2, y2) and the forklift's current position (x, y): θ = atan2(y2-y2, x2-x1) - atan2(y-y1, x-x1). The angle difference between the forklift's current position and the next reference point is calculated using the steering angle formula, which is taken as the current expected steering angle. By updating step by step, a sequence of expected steering angles is formed for the entire operation task. The expected steering angle sequence is also presented in the form of discrete data, with each data point corresponding to a specific position on the global reference path, which clarifies the ideal steering angle that the forklift should achieve at that position.

[0031] Simultaneously, a magnetic encoder is fixedly connected to the steering knuckle of the forklift's steering axle. The magnetic encoder is a high-precision angle measurement device based on the principle of magnetoelectric conversion, capable of accurately measuring the rotation angle of the steering knuckle by detecting changes in the magnetic field. The steering knuckle is a key component of the forklift's steering system, and its rotation angle directly reflects the steering angle of the forklift wheels. An angle sensor is fixed to the axle body opposite the steering knuckle. The angle sensor measures the relative angle change between the axle body and the steering knuckle. Working in conjunction with the magnetic encoder, it can more comprehensively and accurately obtain the actual steering information of the steering axle. The magnetic encoder and angle sensor together constitute an angle sensing network, connected to the forklift's control system via wired or wireless means to achieve real-time data transmission. Through the constructed angle sensing network, real-time reading is performed according to a preset sampling frequency. The preset sampling frequency is determined by factors such as the forklift's travel speed, steering frequency, and actual needs; for example, a sampling frequency of 50Hz per second is set to ensure timely and accurate capture of changes in steering angle. By reading the data from the angle sensing network in real time and processing it according to the acquisition timestamp, an actual steering angle data stream is generated.

[0032] By generating a global reference path and a desired steering angle sequence, ideal target values ​​are provided for the forklift's steering control. Real-time feedback is provided by constructing an angle sensing network and acquiring the actual steering angle data stream. The combination of these two methods allows for timely understanding of the deviation between the forklift's steering angle and the desired value, providing the necessary data foundation for subsequent fault-tolerant control. This ensures the forklift can travel safely and stably along the predetermined path, improving the efficiency and accuracy of forklift operations.

[0033] Based on the deviation analysis between the expected steering angle sequence and the actual steering angle data stream, the real-time angle deviation value is calculated.

[0034] Furthermore, based on the deviation analysis between the expected steering angle sequence and the actual steering angle data stream, a real-time angle deviation value is calculated. The method includes: performing timestamp synchronization calibration on the expected steering angle sequence and the actual steering angle data stream to determine a control timing reference data pair; traversing the control timing reference data pair, extracting the expected steering angle minus the actual steering angle to obtain the original instantaneous angle deviation value, wherein the control timing of the expected steering angle and the actual steering angle is the same; setting a theoretical response lag threshold range, and determining whether the original instantaneous angle deviation exceeds the theoretical response lag threshold range; if the original instantaneous angle deviation value... If the original instantaneous angle deviation value does not exceed the theoretical response lag threshold range, it is determined that the original instantaneous angle deviation value is reasonable, and the original instantaneous angle deviation value is output as the real-time angle deviation value. If the original instantaneous angle deviation value exceeds the theoretical response lag threshold range, the original instantaneous angle deviation value is marked as a suspicious value. Based on the control timing of the expected steering angle and the actual steering angle, the instantaneous angle deviation value of the previous i-th moment is extracted as the effective deviation value and cyclically judged until the effective deviation value does not exceed the theoretical response lag threshold range. Then, the effective deviation value is output as the real-time angle deviation value, where i is a positive integer less than or equal to 2.

[0035] Specifically, since the expected steering angle sequence is generated in real time by the path planning module, and the actual steering angle data stream is measured by the angle sensor, these two data streams may be out of sync in time. By synchronizing the timestamp of the expected steering angle sequence to the timestamp of the actual steering angle data stream, it is ensured that the two have corresponding angle values ​​at the same time, generating a one-to-one control timing reference data pair to eliminate the error caused by the time difference.

[0036] Iterate through all control timing reference data pairs. For each moment, calculate the difference between the desired steering angle and the actual steering angle to obtain the original instantaneous angle deviation value. The original instantaneous angle deviation value = desired steering angle - actual steering angle. The control timing for the desired and actual steering angles is the same. By subtracting, the deviation of the forklift's steering at the same moment can be accurately reflected. For example, at a certain moment, if the desired steering angle is 30° and the actual steering angle is 28°, then the original instantaneous angle deviation value is 30 - 28 = 2°. Based on the dynamic response characteristics of the forklift steering system, the theoretical response lag threshold range is determined. The theoretical response lag threshold range refers to the maximum reasonable deviation range between the expected steering angle and the actual steering angle under normal control lag conditions. For example, the step response time parameters of the forklift steering system are obtained through bench tests or vehicle calibration, including the system response delay time Td from the issuance of the control command to the start of the steering angle change and the rise time Tr from 10% to 90% of the steering angle. At the same time, the maximum rate of change of the steering angle θmax during this process is recorded in ° / s. Based on the sampling period Ts of the control system, the upper limit of the angle error that will inevitably be generated due to physical lag within one or more sampling periods is calculated. The theoretical angle lag can be expressed as Δθ=θmax×(Td). At the same time, a safety margin coefficient k is introduced in combination with the sensor measurement error and the mechanical clearance error of the actuator, which is taken as 1.2~1.5. The theoretical response lag threshold range is determined to be ±k×Δθ. For example, when the maximum angular velocity of the steering system is 30° / s, the response delay is 80ms, and the sampling period is 20ms, Δθ≈3° is calculated, and the theoretical response hysteresis threshold range is set to ±3.6°.

[0037] If the original instantaneous angle deviation does not exceed the theoretical response lag threshold range, the original instantaneous angle deviation value is determined to be reasonable and output as the real-time angle deviation value. If the original instantaneous angle deviation exceeds the theoretical response lag threshold range, the original instantaneous angle deviation value is marked as a suspicious value. Based on the control timing of the expected steering angle and the actual steering angle at the current moment, the instantaneous angle deviation value within the previous i-th moment is extracted as the valid deviation value for iterative judgment until the valid deviation value does not exceed the theoretical response lag threshold range, where i is a positive integer less than or equal to 2. For example, if the original instantaneous angle deviation value calculated at a certain moment is 4°, which exceeds the theoretical response lag threshold range of ±3.6°, then the instantaneous angle deviation value of the previous 1-th moment is extracted, assuming it is 2°. This instantaneous angle deviation value is within the theoretical response lag threshold range, so 2° is output as the real-time angle deviation value at that moment. If the value of the previous 1-th moment also exceeds the range, then the value of the previous 2-th moment is extracted for judgment, until the valid deviation value does not exceed the theoretical response lag threshold range, and the valid deviation value is output as the real-time angle deviation value.

[0038] By analyzing the deviation between the expected steering angle sequence and the actual steering angle data stream, the steering accuracy of the forklift can be monitored in real time. By introducing a timestamp synchronization calibration and a lag threshold judgment mechanism, meaningless deviations caused by sensor noise or short-term fluctuations can be effectively eliminated. Furthermore, by judging the deviation value at the previous time step i, the tolerance for lag and error is further improved, making the steering of the forklift more stable and reliable. This ensures that the forklift travels stably according to the global reference path, thereby improving the safety and efficiency of forklift operations.

[0039] Furthermore, the desired steering angle sequence and the actual steering angle data stream are time-stamped and synchronized to determine the control timing reference data pair. The method includes: traversing the desired steering angle sequence to perform global timing analysis and constructing a global clock timestamp; matching the global clock timestamp with the desired steering angle sequence to generate a first timestamp matching result; traversing the actual steering angle data stream according to sampling points to perform timing analysis and construct a local clock timestamp; matching the local clock timestamp with the actual steering angle data stream to generate a second timestamp matching result; and performing clock synchronization correction between the first timestamp matching result and the second timestamp matching result to determine the control timing reference data pair.

[0040] Specifically, the process of traversing the desired steering angle sequence and constructing a global clock timestamp is completed using the main control cycle of the forklift steering controller as the time base. First, the steering controller provides a unified system clock, generated by a high-precision timer or a real-time operating system (RTOS), whose period is the steering control cycle. Based on this, the desired steering angle sequence is traversed point-by-point according to its generation order. When the desired steering angle value is output in the k-th control cycle, the current clock count value or the control cycle count value k is synchronously read and mapped to the corresponding global clock timestamp, completing the global timing analysis. Then, each desired steering angle value is bound to its global clock timestamp at the time of generation, constructing a pair of data [desired steering angle, timestamp], forming a desired steering angle sequence with a unified timestamp, i.e., the first timestamp matching result.

[0041] By employing control cycle counting and the system master clock, the timing attribute of the desired steering angle can be ensured to be strictly consistent with the actual effective time of the control command. Simultaneously, the actual steering angle data stream is independently traversed, and a local clock timestamp is constructed based on sensor sampling interrupts or local counters. This local timestamp is then matched with the actual steering angle value at each sampling point to generate a second timestamp matching result. After obtaining two sets of time-stamped data, a clock synchronization correction mechanism aligns the first and second timestamp matching results. Specifically, time interpolation, nearest neighbor matching, or linear resampling can be used to map the actual steering angle data onto the global clock time axis. This eliminates timing deviations caused by inconsistencies between the control cycle and sensor sampling frequency, forming control timing reference data pairs under the same time base. Each control timing reference data pair corresponds one-to-one with the desired steering angle and the actual steering angle.

[0042] By constructing global and local clock timestamps and performing precise clock synchronization correction, the expected steering angle sequence and the actual steering angle data stream are accurately correlated in the time dimension, improving the accuracy and reliability of forklift steering angle fault-tolerant control.

[0043] Historical trend data of the steering axle is introduced to make a fault tolerance judgment on the real-time angle deviation value, and a fault tolerance judgment result is generated.

[0044] Furthermore, the process of constructing historical trend data includes: setting a sliding time window, continuously storing historical angle deviation values ​​according to the sliding time window to generate a historical angle deviation sequence; performing numerical difference calculation on the historical angle deviation sequence to generate a deviation change rate sequence; and performing periodic analysis based on the deviation change rate sequence and the historical angle deviation sequence to generate the historical trend data.

[0045] Specifically, a sliding time window is set to limit the range of data participating in the statistical analysis. The window length can be adjusted according to the actual needs of forklift steering control and the data update frequency. For example, the sliding time window length can be set to 3 seconds, with a sliding step of 1 second. This means that the data range observed each time is within 3 seconds, and the window moves backward every second to obtain a new 5-second data segment. Based on the sliding time window, historical angle deviation values ​​are continuously stored in a first-in, first-out (FIFO) manner. When a new deviation value enters the window, the oldest deviation value is automatically removed, thus forming a historical angle deviation sequence that shifts over time. Numerical difference calculation is performed on the historical angle deviation sequence, that is, the difference between the angle deviation values ​​at adjacent times is calculated and divided by the corresponding time interval to obtain the rate of change of the deviation over time, forming a deviation change rate sequence. For example, assuming a time interval of 1 second, for the historical angle deviation sequence [0.2, 0.3, 0.1, 0.4, 0.2], the first deviation change rate is 0.3 - 0.2 = 0.1° / s, the second deviation change rate is 0.1 - 0.3 = -0.2° / s, and so on. These deviation change rates are stored in order to obtain the deviation change rate sequence [0.1, -0.2, 0.3, -0.2]. The deviation change rate sequence is used to reflect the dynamic change intensity and direction of the forklift steering deviation.

[0046] Using the historical angle deviation sequence and corresponding deviation change rate sequence within a sliding time window as input signals, frequency domain analysis or time domain periodic feature extraction is performed on the historical angle deviation sequence and corresponding deviation change rate sequence. For example, the main frequency components are calculated using Fast Fourier Transform (FFT) to identify the periodic oscillation characteristics of the steering deviation. Simultaneously, the sliding mean, variance, peak spacing, and cyclical patterns of the upper and lower limits of the deviation are calculated in the time domain to capture the amplitude variation law and short-period fluctuation characteristics of the deviation. The frequency domain and time domain analysis results are fused to form a comprehensive feature vector containing deviation amplitude, rate of change, main period, and fluctuation pattern, i.e., historical change trend data.

[0047] By setting a sliding time window, calculating the deviation change rate sequence, and performing periodic analysis to generate historical trend data, we can comprehensively and accurately understand the dynamic characteristics of forklift steering deviation, providing a scientific basis for forklift steering angle fault-tolerant control, thereby improving the safety and stability of forklift operation.

[0048] Furthermore, historical trend data of the steering axle is introduced to perform fault tolerance judgment on the real-time angle deviation value, generating a fault tolerance judgment result. The method includes: identifying the stable operating condition information of the forklift; matching the historical angle deviation sequence according to the stable operating condition to construct a training sample; using the training sample to perform time series analysis to construct a dynamic behavior benchmark model; synchronizing the real-time angle deviation value to the dynamic behavior benchmark model for consistency comparison to obtain an anomaly score; performing fault tolerance correlation analysis based on the anomaly score, setting multi-level fault tolerance trigger conditions, and performing fault tolerance judgment according to the multi-level fault tolerance trigger conditions combined with the anomaly score to generate a fault tolerance judgment result.

[0049] Specifically, a stable operating condition is defined as a forklift's operation during normal operation where parameters such as travel speed and steering angle change rate remain relatively stable. For example, if a forklift's speed fluctuation is within ±5% and its steering angle change rate is close to 0 during straight-line constant-speed travel, this can be considered a stable operating condition. By installing speed and steering angle sensors on the forklift, real-time data on travel speed and steering angle is collected. Data analysis algorithms, such as moving average filtering algorithms, are used to smooth the collected data to accurately identify stable operating conditions. The moving average filtering algorithm averages the collected data over a certain time window (e.g., using a window length of 3) to effectively eliminate random noise in the collected data and improve the accuracy of operating condition identification. The stable operating conditions are then matched with corresponding historical angle deviation sequences to construct training samples.

[0050] Time series analysis is performed using training samples. For example, the Autoregressive Integral Moving Average (ARIMA) model is used to transform the time series data into a stationary series through differencing. Then, autoregressive and moving average models are established to analyze the dynamic changes of forklifts. First, the parameters (p, d, q) of the ARIMA model are determined, where p represents the autoregressive order, d represents the differencing order, and q represents the moving average order. These parameters are determined by analyzing the autocorrelation function (ACF) and partial autocorrelation function (PACF) plots of the time series. For example, for a historical deviation series, by plotting the ACF and PACF plots, it is found that the ACF plot is truncated after a lag of 2, and the PACF plot is truncated after a lag of 1. Initially, p=1 and q=2 are determined. By using differencing operations to make the series stationary, d=1 is determined, and finally, the ARIMA(1,1,2) model is obtained.

[0051] The ARIMA model is trained using matched training samples, and the model parameters, i.e., the autoregressive coefficients, are estimated using methods such as maximum likelihood estimation. 1 and the moving average coefficients θ1, θ2, and the variance σ of the error term. 2 Specifically, it refers to the likelihood function L( ) used to construct the ARIMA model. 1,θ1,θ2,σ 2 The training data represents the probability of observing a training sample sequence under given parameters. The optimal parameter estimates are then obtained by maximizing the likelihood function using optimization algorithms such as BFGS or the Newton-Raphson method. During training, the ARIMA model can be continuously updated using a rolling window or iterative approach to ensure it captures the dynamic changes and time dependencies of historical angle deviation sequences. After training, the ARIMA model is validated using a subset of historical data not used in the training. Prediction error metrics, such as mean squared error and mean absolute error, are calculated. If the errors are within acceptable ranges (e.g., mean squared error less than 0.05 and mean absolute error less than 0.2), the trained ARIMA model is used as a benchmark model reflecting the dynamic behavior of forklift steering under stable operating conditions. This benchmark model can predict future short-term deviation trends based on the current real-time deviation sequence and provides a reliable reference for anomaly scoring and multi-level fault tolerance assessment.

[0052] Real-time angle deviation values ​​are synchronized chronologically to a pre-constructed dynamic behavior baseline model for analysis. The predicted deviation values ​​of the dynamic behavior baseline model are obtained, and the real-time angle deviation values ​​are compared with the predicted deviation values. The consistency comparison result is obtained by calculating the residual parameter between the real-time angle deviation values ​​and the predicted values. Based on the consistency comparison result, an anomaly score is calculated. Through consistency comparison and anomaly score, the degree of deviation between the real-time angle deviation value and normal behavior can be quantified. Fault tolerance correlation analysis is performed based on the anomaly score, setting multi-level fault tolerance trigger conditions. Fault tolerance judgment is then performed according to these multi-level fault tolerance trigger conditions combined with the anomaly score, generating a fault tolerance judgment result.

[0053] By introducing historical trend data of the steering axle to make fault-tolerant judgments on real-time angle deviations, we can make full use of the experience information of historical data, improve the accuracy and reliability of fault-tolerant judgments, and thus provide a reliable basis for fault-tolerant compensation control, thereby effectively ensuring the safe and stable operation of forklift steering.

[0054] Furthermore, the real-time angle deviation value is synchronized to the dynamic behavior benchmark model for consistency comparison to obtain an anomaly score. The method includes: extracting the prediction deviation value of the dynamic behavior benchmark model according to the timestamp of the real-time angle deviation value; calculating the residual parameter based on the comparison between the real-time angle deviation value and the prediction deviation value; performing data confidence analysis based on the dynamic behavior benchmark model and setting a confidence interval; performing exceedance calculation on the residual parameter and the confidence interval, extracting the parameter exceedance value, and setting a first anomaly index based on the parameter exceedance value; performing change analysis on the real-time angle deviation value according to a fixed time period through the dynamic behavior benchmark model to determine the short-term change pattern; extracting the normal change pattern based on the historical change trend data, calculating the matching degree between the normal change pattern and the short-term change pattern, and setting a second anomaly index; and performing a weighted calculation on the first anomaly index and the second anomaly index to generate the anomaly score.

[0055] Specifically, based on the timestamp of the real-time angle deviation value, the prediction deviation value at the same moment or corresponding prediction step size is extracted from the dynamic behavior benchmark model, and the model output is aligned with the measured data under a unified time benchmark. The real-time angle deviation value and the prediction deviation value are differentially calculated to obtain the residual parameter, which is calculated as: Real-time angle deviation value - Prediction deviation value. The residual parameter is used to characterize the degree of deviation of the real-time behavior from the normal dynamic benchmark.

[0056] Based on the statistical characteristics of the dynamic behavioral baseline model, a confidence analysis is performed on the residual distribution during the training phase. A confidence interval is defined, which refers to the possible range of predicted values ​​from the dynamic behavioral baseline model at a certain confidence level. Statistical methods are used, such as calculating the standard deviation of the prediction deviation, and then determining the confidence interval based on the properties of the normal distribution. For example, a 95% confidence interval is the prediction deviation ± 1.96 × standard deviation. The calculated residual parameters are then compared to the defined confidence intervals for exceedance calculation. If the residual parameters exceed the confidence interval, it indicates a significant difference between the real-time angle deviation and the model's predicted value, potentially indicating an anomaly. The parameter exceedance value is calculated as follows: if the residual parameter is greater than the upper limit of the confidence interval, then the parameter exceedance value = residual parameter - upper limit of the confidence interval; if the residual parameter is less than the lower limit of the confidence interval, then the parameter exceedance value = lower limit of the confidence interval - residual parameter; if the residual parameter is within the confidence interval, then the parameter exceedance value is 0. The extracted parameter exceedance values ​​are normalized and used as the first anomaly index to reflect whether the instantaneous deviation amplitude is abnormal.

[0057] By analyzing the changes in real-time angle deviation values ​​over fixed time intervals, such as 5 seconds, using a dynamic behavioral benchmark model, short-term variation patterns are determined. These patterns reflect the trend of real-time angle deviation values ​​over a short period, such as gradual increase, gradual decrease, or stability. Specifically, time series differencing or trend analysis methods are used to determine these short-term variation patterns. For example, first-order differencing is performed on the real-time angle deviation values ​​at five consecutive time points to obtain a difference sequence. The sign and magnitude of the difference sequence are used to determine the short-term variation pattern. Simultaneously, normal variation patterns under the same operating conditions are extracted from historical trend data. These normal variation patterns represent common trends in historical angle deviation values ​​under stable operating conditions. The matching degree between the normal variation pattern and the short-term variation pattern is calculated using methods such as cosine similarity or Euclidean distance. For example, cosine similarity is used to calculate the matching degree. The normal change pattern and the short-term change pattern are treated as two vectors, and the cosine of the angle between them is calculated. The closer the cosine of the angle is to 1, the higher the matching degree. A second anomaly index is set based on the matching degree. The lower the matching degree, the more significant the deviation of the change pattern from normal, and the higher the second anomaly index. The second anomaly index is mapped to the [0,1] interval. The first and second anomaly indices are then weighted and fused according to preset weights. The anomaly score = first anomaly index × first anomaly weight + second anomaly index × second anomaly weight. The preset weights are set according to actual needs and experience. For example, if the residual parameter is considered to more directly reflect the degree of anomaly of the real-time angle deviation value, a higher weight can be assigned to the first anomaly index, such as 0.6, and a weight of 0.4 can be assigned to the second anomaly index. Then, the corresponding anomaly score is calculated using the weighted fusion formula: anomaly score = first anomaly index × 0.7 + second anomaly index × 0.3.

[0058] By combining residual analysis and change pattern matching, real-time angle deviation values ​​are evaluated from different perspectives, enabling a more comprehensive and accurate assessment of the degree of deviation anomaly. Residual analysis directly reflects the difference between real-time deviation and model prediction deviation, while change pattern matching considers the short-term trend of deviation changes, avoiding the limitations of a single method. By generating anomaly scores, a quantitative basis is provided for forklift steering angle fault-tolerant control, allowing for appropriate fault-tolerant measures to be taken based on the severity of the deviation. This further improves the pertinence and effectiveness of forklift steering angle fault-tolerant control, thereby enhancing the reliability and safety of forklift steering.

[0059] Furthermore, based on the anomaly score, a fault-tolerance correlation analysis is performed, and multi-level fault-tolerance trigger conditions are set. Fault-tolerance judgment is then performed according to these multi-level trigger conditions combined with the anomaly score, generating a fault-tolerance judgment result. The method includes: performing fluctuation analysis on the historical trend data based on the anomaly score; adaptive scaling based on the fluctuation results; setting multi-level fault-tolerance trigger conditions, which include a first dynamic threshold and a second dynamic threshold, where the first dynamic threshold is lower than the second dynamic threshold; when the anomaly score is lower than the first dynamic threshold, the steering angle state is normal, and backtracking is performed according to the anomaly score to generate a first fault-tolerance judgment result; when the anomaly score is between the first and second dynamic thresholds, the steering angle state deviates, and continuous monitoring is performed according to the anomaly score to generate a second fault-tolerance judgment result; when the anomaly score is higher than the second dynamic threshold, the steering angle state is abnormal, and strong fault-tolerance intervention is performed according to the anomaly score to generate a third fault-tolerance judgment result.

[0060] Specifically, the anomaly score sequence within a sliding time window is used as input, and fluctuation analysis is performed on its corresponding historical trend data. By calculating the mean and standard deviation of the anomaly score under stable operating conditions, the natural fluctuation range under normal operating conditions is reflected. The threshold is adaptively scaled according to the fluctuation amplitude, dynamically setting multi-level trigger boundaries with the mean as the center and the standard deviation as the scale. The multi-level fault-tolerant trigger conditions include a first dynamic threshold and a second dynamic threshold. The first dynamic threshold is lower than the second dynamic threshold. The first dynamic threshold can be set as mean + α × standard deviation, where α is 1~1.5, and the second dynamic threshold can be set as mean + β × standard deviation, where β is 2~3. The first dynamic threshold is always lower than the second dynamic threshold. For example, under a certain stable operating condition, the anomaly score mean is 0.25 and the standard deviation is 0.1, then the first dynamic threshold can be set to 0.40, and the second dynamic threshold can be set to 0.55.

[0061] The real-time anomaly score is compared with the aforementioned multi-level dynamic thresholds to determine the fault tolerance status: When the anomaly score is below the first dynamic threshold, the current steering angle is determined to be normal. A short-term backtracking verification is performed based on the anomaly score's changing trend. The purpose of the backtracking is to analyze whether this normal state has continuity and stability. By observing the changes in the anomaly score over a period of time, the normal state of the steering angle is further confirmed, generating a first fault tolerance judgment result. This first fault tolerance judgment result indicates a normal state, requiring no fault tolerance. When the anomaly score is between the first and second dynamic thresholds, it indicates a slight deviation in the steering angle, but not yet reaching the risk of loss of control. At this time, continuous monitoring is performed according to the anomaly score. The frequency of continuous monitoring can be adjusted according to the actual situation, for example, monitoring once every 3 seconds. Through continuous monitoring, the changing trend of the anomaly score is observed, triggering small-amplitude feedforward or limiting compensation, generating a second fault tolerance judgment result. This second fault tolerance judgment result indicates a slight deviation, suggesting observation and micro-compensation. When the anomaly score is greater than the second dynamic threshold, it indicates that the real-time deviation has significantly deviated from the dynamic behavior benchmark. The steering angle state is judged to be abnormal, and a strong fault tolerance strategy is immediately triggered to generate a third fault tolerance judgment result. That is, the third fault tolerance judgment result is significantly abnormal and requires strong fault tolerance intervention.

[0062] By setting dynamic thresholds based on historical data fluctuation analysis using anomaly scores, adaptive adjustments to fault-tolerance trigger conditions are achieved, enabling more precise adaptation to changes in forklift steering angle deviations under different operating conditions. The multi-level fault-tolerance judgment mechanism employs different strategies—normal backtracking, continuous monitoring with suggested compensation, and strong intervention—based on different ranges of anomaly scores. This allows for timely and effective responses according to the severity of the deviation, avoiding over-intervention or under-intervention issues caused by single threshold judgments. This improves the fault tolerance and reliability of forklift steering control, further enhancing forklift operational safety.

[0063] Based on the fault tolerance judgment result, the fault tolerance compensation angle is calculated to perform steering control analysis of the steering axle, and steering control commands are generated to drive the forklift steering actuator to perform adaptive fault tolerance control of the steering angle.

[0064] Furthermore, the method for calculating the fault tolerance compensation angle based on the fault tolerance judgment result includes: when the fault tolerance judgment result is the first fault tolerance judgment result, performing feedforward compensation based on the steering angle state to generate a first fault tolerance compensation angle; when the fault tolerance judgment result is the second fault tolerance judgment result, performing feedback-feedforward composite compensation based on the steering angle state to generate a second fault tolerance compensation angle; when the fault tolerance judgment result is the third fault tolerance judgment result, triggering a deep fault tolerance mode: S1: constructing a sliding surface based on the real-time angle deviation value, performing gain analysis on the sliding surface according to the third fault tolerance judgment result, and calculating the equivalent control quantity; S2: setting the third fault tolerance compensation angle according to the equivalent control quantity.

[0065] Specifically, when the fault tolerance judgment result is the first fault tolerance judgment result, the steering angle state is determined to be normal. At this time, based on the expected steering angle sequence output by the steering path planning module, the equivalent time lag corresponding to the actual response of the steering system from the command input to the steering axle is estimated according to the steering geometric parameters such as the current speed of the forklift, wheelbase and steering mechanism transmission ratio. Then, within the equivalent time lag range, the expected steering angle sequence is time-shifted to extract the expected steering angle change corresponding to the future time. This change is used as the feedforward compensation amount to compensate for the mechanical and hydraulic response lag that may occur in the actual execution of the steering system.

[0066] When the fault tolerance judgment result is the second fault tolerance judgment result, a slight deviation in the steering angle is detected. At this point, a feedback-feedforward composite compensation strategy is adopted: on top of the feedforward compensation angle, a feedback compensation amount proportional to the real-time angle deviation value and positively correlated with the fault tolerance level coefficient is superimposed. The fault tolerance level coefficient is a coefficient set according to the severity of the current deviation, used to adjust the intensity of the feedback compensation. For example, if the real-time angle deviation value is 0.5°, the fault tolerance level coefficient is set to 0.8, and the proportionality coefficient between the feedback compensation amount and the real-time angle deviation value is set to 0.6, then the feedback compensation amount is 0.5 × 0.6 × 0.8 = 0.24°. Assuming the feedforward compensation angle is 0.3°, then the second fault tolerance compensation angle is 0.3 + 0.24 = 0.54°. This composite compensation method combines the anticipation of feedforward compensation with the accuracy of feedback compensation, more effectively correcting slight deviations in the steering angle.

[0067] When the fault tolerance judgment result is the third fault tolerance judgment result, it indicates that the steering angle has become significantly abnormal, and the deep fault tolerance mode is immediately triggered: First, a sliding surface is constructed based on the real-time angle deviation value. The constructed sliding surface is represented as follows: Where e is the real-time angle deviation value, λ is the rate of change of the angular deviation, and λ is the sliding mode coefficient, a positive constant used to adjust the shape and dynamic performance of the sliding surface. Gain analysis is performed on the sliding surface based on the third tolerance judgment result to calculate the equivalent control quantity, used to quickly suppress deviation propagation. Gain analysis determines the gain parameters in sliding mode control based on the forklift's steering dynamic characteristics and control requirements to ensure the stability and speed of forklift steering angle tolerance control. The calculated equivalent control quantity can be expressed as ueq = -K. sgn(s), where K is adaptively adjusted according to the anomaly level, and a third fault-tolerant compensation angle is generated based on the equivalent control quantity to ensure that the steering quickly returns to the safe range.

[0068] After calculating the fault-tolerant compensation angle, it is superimposed on the forklift's target steering angle to obtain the final steering control angle. Steering control commands are generated based on this angle and transmitted via the electronic control unit (ECU) to the forklift's steering actuator, such as the electric power steering (EPS) system. The steering actuator adjusts the steering axle angle according to the received steering control commands, achieving adaptive fault-tolerant control of the steering angle.

[0069] By using different compensation angle calculation methods based on different fault tolerance judgment results, it is possible to accurately compensate for different degrees of deviation in the forklift steering angle. Feedforward compensation can eliminate potential interference in advance, feedback compensation can correct actual deviations in a timely manner, and deep fault tolerance mode can provide strong fault tolerance capability when significant anomalies occur. Through adaptive fault tolerance control, the stability and reliability of forklift steering can be effectively improved, ensuring that the forklift can turn accurately and safely under various working conditions.

[0070] Example 2, based on the same inventive concept as the angle-tolerant control method for forklift steering in the aforementioned examples, such as... Figure 2 As shown, this application provides an angle-tolerant control system for forklift steering, wherein the angle-tolerant control system for forklift steering includes: The data acquisition component 11 is used to set the desired steering angle sequence of the forklift based on the forklift steering path planning module, and to obtain the actual steering angle data stream by real-time sensing of the forklift steering axle based on the angle sensor; the deviation analysis component 12 is used to perform deviation analysis based on the desired steering angle sequence and the actual steering angle data stream, and to calculate the real-time angle deviation value; the fault tolerance judgment component 13 is used to introduce the historical change trend data of the steering axle to perform fault tolerance judgment on the real-time angle deviation value, and to generate a fault tolerance judgment result; the fault tolerance control component 14 is used to calculate the fault tolerance compensation angle based on the fault tolerance judgment result, to perform steering control analysis of the steering axle, and to generate steering control commands to drive the forklift steering actuator to perform adaptive fault tolerance control of the steering angle.

[0071] Furthermore, the data acquisition component 11 is also used for: retrieving the forklift's navigation map based on the forklift steering path planning module, performing real-time trajectory calculation according to the forklift's operation task, and generating a global reference path; performing steering tracking prediction based on the global reference path, and setting the desired steering angle sequence; fixing the magnetic encoder to the steering knuckle of the steering axle, and fixing the angle sensor to the axle body opposite to the steering knuckle, thereby generating an angle sensing network; and performing real-time reading through the angle sensing network at a preset sampling frequency to generate the actual steering angle data stream.

[0072] Furthermore, the deviation analysis component 12 is also used for: performing timestamp synchronization calibration on the expected steering angle sequence and the actual steering angle data stream to determine a control timing reference data pair; traversing the control timing reference data pair, extracting the expected steering angle minus the actual steering angle to obtain the original instantaneous angle deviation value, wherein the control timing of the expected steering angle and the actual steering angle is the same; setting a theoretical response lag threshold range, and determining whether the original instantaneous angle deviation exceeds the theoretical response lag threshold range; if the original instantaneous angle deviation value does not exceed the theoretical response lag threshold range, If the original instantaneous angle deviation value is deemed reasonable, it is output as the real-time angle deviation value. If the original instantaneous angle deviation value exceeds the theoretical response lag threshold range, it is marked as a suspicious value. Based on the control timing of the expected steering angle and the actual steering angle, the instantaneous angle deviation value at the previous i-th moment is extracted as the effective deviation value and iteratively judged until the effective deviation value does not exceed the theoretical response lag threshold range. Then, the effective deviation value is output as the real-time angle deviation value, where i is a positive integer less than or equal to 2.

[0073] Furthermore, the deviation analysis component 12 is also used for: traversing the desired steering angle sequence to perform global timing analysis and constructing a global clock timestamp; matching the global clock timestamp with the desired steering angle sequence to generate a first timestamp matching result; traversing the actual steering angle data stream to perform timing analysis according to sampling points and constructing a local clock timestamp; matching the local clock timestamp with the actual steering angle data stream to generate a second timestamp matching result; and performing clock synchronization correction between the first timestamp matching result and the second timestamp matching result to determine the control timing reference data pair.

[0074] Furthermore, the fault tolerance judgment component 13 is also used to: set a sliding time window, continuously store historical angle deviation values ​​according to the sliding time window, and generate a historical angle deviation sequence; perform numerical difference calculation on the historical angle deviation sequence to generate a deviation change rate sequence; and perform periodic analysis according to the deviation change rate sequence and the historical angle deviation sequence to generate the historical change trend data.

[0075] Furthermore, the fault tolerance judgment component 13 is also used to: identify the stable operating condition information of the forklift, match the historical angle deviation sequence according to the stable operating condition, and construct training samples; use the training samples to perform time series analysis and construct a dynamic behavior benchmark model; synchronize the real-time angle deviation value to the dynamic behavior benchmark model for consistency comparison and obtain an anomaly score; perform fault tolerance correlation analysis based on the anomaly score, set multi-level fault tolerance triggering conditions, and perform fault tolerance judgment according to the multi-level fault tolerance triggering conditions combined with the anomaly score to generate a fault tolerance judgment result.

[0076] Furthermore, the fault-tolerant judgment component 13 is also used to: extract the prediction deviation value of the dynamic behavior benchmark model according to the timestamp of the real-time angle deviation value; calculate the residual parameter based on the comparison between the real-time angle deviation value and the prediction deviation value; perform data confidence analysis based on the dynamic behavior benchmark model and set a confidence interval; perform exceedance calculation on the residual parameter and the confidence interval, extract the parameter exceedance value, and set a first anomaly index based on the parameter exceedance value; perform change analysis on the real-time angle deviation value according to a fixed time period through the dynamic behavior benchmark model to determine the short-term change pattern; extract the normal change pattern based on the historical change trend data, calculate the matching degree between the normal change pattern and the short-term change pattern, and set a second anomaly index; and perform weighted calculation on the first anomaly index and the second anomaly index to generate the anomaly score.

[0077] Furthermore, the fault tolerance judgment component 13 is also used for: performing fluctuation analysis on the historical trend data based on the anomaly score, adaptive scaling according to the fluctuation results, and setting multi-level fault tolerance trigger conditions, wherein the multi-level fault tolerance trigger conditions include a first dynamic threshold and a second dynamic threshold, and the first dynamic threshold is lower than the second dynamic threshold; when the anomaly score is lower than the first dynamic threshold, the steering angle state is normal, and backtracking is performed according to the anomaly score to generate a first fault tolerance judgment result; when the anomaly score is between the first dynamic threshold and the second dynamic threshold, the steering angle state deviates, and continuous monitoring is performed according to the anomaly score to generate a second fault tolerance judgment result; when the anomaly score is higher than the second dynamic threshold, the steering angle state is abnormal, and strong fault tolerance intervention is performed according to the anomaly score to generate a third fault tolerance judgment result.

[0078] Furthermore, the fault-tolerant control component 14 is also used to: when the fault tolerance judgment result is the first fault tolerance judgment result, perform feedforward compensation based on the steering angle state to generate a first fault tolerance compensation angle; when the fault tolerance judgment result is the second fault tolerance judgment result, perform feedback-feedforward composite compensation based on the steering angle state to generate a second fault tolerance compensation angle; when the fault tolerance judgment result is the third fault tolerance judgment result, trigger a deep fault tolerance mode: S1: construct a sliding surface based on the real-time angle deviation value, perform gain analysis on the sliding surface according to the third fault tolerance judgment result, and calculate the equivalent control quantity; S2: set the third fault tolerance compensation angle according to the equivalent control quantity.

[0079] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The angle fault-tolerant control method and specific examples for forklift steering in the foregoing embodiment 1 are also applicable to the angle fault-tolerant control system for forklift steering in this embodiment. Through the foregoing detailed description of the angle fault-tolerant control method for forklift steering, those skilled in the art can clearly understand the angle fault-tolerant control system for forklift steering in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0080] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0081] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. An angle-tolerant control method for forklift steering, characterized in that, The method includes: The desired steering angle sequence of the forklift is set based on the forklift steering path planning module, and the actual steering angle data stream is obtained by real-time sensing of the forklift steering axle based on the angle sensor. Based on the deviation analysis between the expected steering angle sequence and the actual steering angle data stream, the real-time angle deviation value is calculated. Historical trend data of the steering axle is introduced to perform fault tolerance judgment on the real-time angle deviation value, and a fault tolerance judgment result is generated. Based on the fault tolerance judgment result, the fault tolerance compensation angle is calculated to perform steering control analysis of the steering axle, and steering control commands are generated to drive the forklift steering actuator to perform adaptive fault tolerance control of the steering angle.

2. The angle-tolerant control method for forklift steering as described in claim 1, characterized in that, The method involves setting the desired steering angle sequence of the forklift based on the forklift steering path planning module, and obtaining the actual steering angle data stream by real-time sensing of the forklift steering axle based on angle sensors. The forklift steering path planning module retrieves the forklift's navigation map and performs real-time trajectory calculations based on the forklift's operation tasks to generate a global reference path. Based on the global reference path, a steering tracking prediction is performed, and the desired steering angle sequence is set. A magnetic encoder is fixedly connected to the steering knuckle of the steering axle, and an angle sensor is fixed to the axle body opposite the steering knuckle to generate an angle sensing network. The actual steering angle data stream is generated by real-time reading through the angle sensing network at a preset sampling frequency.

3. The angle-tolerant control method for forklift steering as described in claim 1, characterized in that, Based on the deviation analysis between the expected steering angle sequence and the actual steering angle data stream, the real-time angle deviation value is calculated. The method includes: The desired steering angle sequence and the actual steering angle data stream are time-stamped and synchronized to determine the control timing reference data pair; Traverse the control timing reference data pairs, extract the desired steering angle minus the actual steering angle to obtain the original instantaneous angle deviation value, and the control timing of the desired steering angle and the actual steering angle is the same; Set a theoretical response hysteresis threshold range, and determine whether the original instantaneous angle deviation exceeds the theoretical response hysteresis threshold range; If the original instantaneous angle deviation value does not exceed the theoretical response hysteresis threshold range, then the original instantaneous angle deviation value is determined to be reasonable, and the original instantaneous angle deviation value is output as the real-time angle deviation value. If the original instantaneous angle deviation exceeds the theoretical response lag threshold range, the original instantaneous angle deviation value is marked as a suspicious value. Based on the control timing of the expected steering angle and the actual steering angle, the instantaneous angle deviation value at the previous i-th moment is extracted as a valid deviation value and cyclically judged until the valid deviation value does not exceed the theoretical response lag threshold range. Then, the valid deviation value is output as the real-time angle deviation value, where i is a positive integer less than or equal to 2.

4. The angle-tolerant control method for forklift steering as described in claim 3, characterized in that, The method includes performing timestamp synchronization calibration between the desired steering angle sequence and the actual steering angle data stream to determine the control timing reference data pair, comprising: Perform global timing analysis by traversing the desired turning angle sequence and construct a global clock timestamp; The global clock timestamp is matched with the desired turning angle sequence to generate a first timestamp matching result; The actual steering angle data stream is traversed and time-series analysis is performed according to the sampling points to construct a local clock timestamp; The local clock timestamp is matched with the actual steering angle data stream to generate a second timestamp matching result; The first timestamp matching result and the second timestamp matching result are clock synchronized and corrected to determine the control timing reference data pair.

5. The angle-tolerant control method for forklift steering as described in claim 1, characterized in that, The process and methods for constructing historical trend data include: Set a sliding time window, and continuously store the historical angle deviation values ​​according to the sliding time window to generate a historical angle deviation sequence; Numerical difference calculation is performed on the historical angle deviation sequence to generate a deviation change rate sequence; Periodic analysis is performed by combining the deviation change rate sequence with the historical angle deviation sequence to generate the historical change trend data.

6. The angle-tolerant control method for forklift steering as described in claim 1, characterized in that, The method involves using historical trend data of the steering axle to perform fault tolerance assessment on the real-time angle deviation value, and generating a fault tolerance assessment result. Identify the stable operating conditions of the forklift, and match the historical angle deviation sequence according to the stable operating conditions to construct training samples; Time series analysis was performed using the training samples to construct a dynamic behavior benchmark model; The real-time angle deviation value is synchronized to the dynamic behavior benchmark model for consistency comparison to obtain an anomaly score; Based on the anomaly score, a fault tolerance correlation analysis is performed, multi-level fault tolerance triggering conditions are set, and fault tolerance judgment is performed according to the multi-level fault tolerance triggering conditions and the anomaly score to generate a fault tolerance judgment result.

7. The angle-tolerant control method for forklift steering as described in claim 6, characterized in that, The real-time angle deviation value is synchronized to the dynamic behavior benchmark model for consistency comparison to obtain an anomaly score. The method includes: The prediction deviation value of the dynamic behavior benchmark model is extracted according to the timestamp of the real-time angle deviation value; The residual parameter is calculated by comparing the real-time angle deviation value with the predicted deviation value. Based on the dynamic behavior benchmark model, data confidence analysis is performed, and confidence intervals are set. The residual parameter and the confidence interval are subjected to exceedance calculation, the parameter exceedance value is extracted, and the first anomaly index is set according to the parameter exceedance value. The dynamic behavior benchmark model is used to analyze the changes in the real-time angle deviation value over a fixed period of time to determine the short-term change pattern. Based on the historical trend data, normal change patterns are extracted, and the matching degree between the normal change patterns and the short-term change patterns is calculated to set a second anomaly index. The first anomaly index and the second anomaly index are weighted and calculated to generate the anomaly score.

8. The angle-tolerant control method for forklift steering as described in claim 6, characterized in that, Based on the anomaly score, a fault tolerance correlation analysis is performed, multi-level fault tolerance triggering conditions are set, and fault tolerance judgment is made according to the multi-level fault tolerance triggering conditions combined with the anomaly score to generate a fault tolerance judgment result. The method includes: Based on the anomaly score, the historical trend data is subjected to fluctuation analysis. The data is then adaptively scaled according to the fluctuation results. Multi-level fault tolerance triggering conditions are set. The multi-level fault tolerance triggering conditions include a first dynamic threshold and a second dynamic threshold. The first dynamic threshold is lower than the second dynamic threshold. When the anomaly score is lower than the first dynamic threshold, the steering angle is normal. Backtracking is performed according to the anomaly score to generate the first fault tolerance judgment result. When the anomaly score is between the first dynamic threshold and the second dynamic threshold, the steering angle state is deviated. The anomaly score is continuously monitored to generate a second fault tolerance judgment result. When the anomaly score is higher than the second dynamic threshold, the steering angle state is abnormal. Strong fault-tolerant intervention is performed according to the anomaly score to generate a third fault-tolerant judgment result.

9. The angle-tolerant control method for forklift steering as described in claim 8, characterized in that, The method for calculating the fault tolerance compensation angle based on the fault tolerance judgment result includes: When the fault tolerance judgment result is the first fault tolerance judgment result, feedforward compensation is performed based on the steering angle state to generate the first fault tolerance compensation angle. When the fault tolerance judgment result is the second fault tolerance judgment result, feedback-feedforward composite compensation is performed based on the steering angle state to generate the second fault tolerance compensation angle. When the fault tolerance judgment result is the third fault tolerance judgment result, the deep fault tolerance mode is triggered: S1: Construct a sliding surface based on the real-time angle deviation value, perform gain analysis on the sliding surface according to the third fault tolerance judgment result, and calculate the equivalent control quantity; S2: Set the third fault-tolerant compensation angle according to the equivalent control quantity.

10. An angle-tolerant control system for forklift steering, characterized in that, The steps for implementing the angle-tolerant control method for forklift steering as described in any one of claims 1 to 9 include: The data acquisition component is used to set the desired steering angle sequence of the forklift based on the forklift steering path planning module, and to obtain the actual steering angle data stream by real-time sensing of the forklift steering axle based on the angle sensor. A deviation analysis component is used to perform deviation analysis based on the expected steering angle sequence and the actual steering angle data stream, and to calculate the real-time angle deviation value. The fault-tolerance judgment component is used to introduce historical change trend data of the steering axle to perform fault-tolerance judgment on the real-time angle deviation value and generate a fault-tolerance judgment result. The fault-tolerant control component is used to calculate the fault-tolerant compensation angle based on the fault-tolerant judgment result, perform steering control analysis of the steering axle, and generate steering control commands to drive the forklift steering actuator to perform adaptive fault-tolerant control of the steering angle.

Images