Balancing method for pedal brake of counterbalanced fork truck

Abstract

The application discloses a counterbalance type forklift pedal brake balancing method, comprising the following steps: S1, multi-dimensional dynamic data real-time acquisition and preprocessing; S2, intelligent matching and dynamic switching of brake mode; S3, dynamic optimization calculation of pedal force feedback characteristic based on intelligent algorithm; S4, real-time adjustment of pedal force and execution mechanism response; S5, effect evaluation and adaptive iterative optimization. The application effectively improves the adaptability of the brake system to complex working conditions, dynamically matches the pedal feedback characteristic with the operation intention and the vehicle state. The system can automatically identify the emergency braking demand and optimize the response speed, reducing the physical consumption of the operator. Through continuous learning of operation habits and environmental characteristics, the human-machine cooperation efficiency is significantly improved. While ensuring brake safety, the energy recovery efficiency and the balance optimization of the operation comfort are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of industrial vehicles, in particular to a counterbalance forklift pedal braking balancing method. BACKGROUND

[0002] In the field of industrial vehicle braking technology, the traditional braking system has significant defects in working condition adaptability and response delay. The existing technology mainly relies on fixed parameter mechanical hydraulic structure, which cannot be self-adapted according to dynamic factors such as real-time changing load distribution and road slope. When the forklift is in complex working conditions such as heavy load, slippery road or slope, the operator often needs to exert additional pedal force to compensate for the insufficient system response, which not only increases the operation fatigue, but also may cause safety hazards such as extended braking distance and decreased directional stability. More seriously, the traditional system lacks intelligent recognition ability of the operator's braking intention, and it is difficult to achieve precise force feedback adjustment in emergency braking or fine control scenarios.

[0003] With the popularity of electric forklifts, the short board of traditional braking system in energy recovery efficiency is increasingly prominent. The fixed parameter braking characteristics cannot be optimized with the battery management system, resulting in a large amount of braking energy dissipated in the form of heat, which not only reduces the endurance, but also aggravates the risk of thermal decay of the braking system. In addition, the single pedal force feedback characteristics are difficult to adapt to the operating habits of different operators, and are prone to muscle fatigue in long-term operation, affecting the safety and efficiency of operation.

[0004] Although some intelligent braking schemes have tried to introduce sensor networks, their data processing capabilities are limited, mode switching has obvious delay, and lacks closed-loop adjustment mechanism. Especially in the mode switching process, the sudden change of pedal force is easy to cause the operator to misjudge, and the separation of "mechanical feeling" and "intelligence" seriously restricts the improvement of human-machine cooperation performance. The existing system also generally lacks self-learning ability and cannot optimize control parameters according to long-term running data, resulting in the performance of the system being stagnant at the initial debugging level. SUMMARY

[0005] Therefore, the present application provides a counterbalance forklift pedal braking balancing method, which has the advantages of improving complex working condition adaptability, achieving precise force feedback adjustment, improving human-machine cooperation safety and optimizing energy recovery efficiency.

[0006] To achieve the above purpose, the present application provides the following technical solutions:

[0007] A counterbalanced forklift pedal brake balancing method, comprising the following steps: S1, multi-dimensional dynamic data real-time acquisition and preprocessing: collecting operator pedal operation characteristics, forklift running state and environmental data, and preprocessing the collected raw data by using a central processing unit; S2, intelligent matching and dynamic switching of brake mode: based on the real-time data after preprocessing, combining a preset mode library and intelligent decision logic, precise matching and smooth switching of the brake mode are realized; S3, dynamic optimization calculation of pedal force feedback characteristic based on intelligent algorithm: the intelligent algorithm fusing multi-dimensional data is run by using the central processing unit, and the optimal pedal force feedback characteristic parameters adapting to the current working condition and operation intention are calculated in real time; S4, actuator response and real-time adjustment of pedal force: based on the optimal feedback characteristic parameters output by S3, the actuator of the intelligent brake pedal force adjustment system is used to realize precise and dynamic adjustment of the pedal force; S5, effect evaluation and adaptive iterative optimization: the adjustment effect is evaluated in multiple dimensions, the algorithm and mode library are iteratively updated combined with long-term running data, and the human-machine cooperation performance is continuously improved.

[0008] Preferably, the operator pedal operation characteristic collection includes: the stepping force is collected in real time by using a pedal force sensor installed at the brake pedal, the pedal displacement amount and change rate are recorded by a displacement sensor, the stepping frequency is captured by a high-frequency sampling module, and the foot palm contact position is obtained by a pressure distribution sensor at the pedal to assist in judging the operation intention; the forklift running state and environmental data collection includes: key parameters are collected by a vehicle state sensor group, including real-time load amount and load distribution obtained by a load sensor, driving speed and acceleration collected by a speed sensor, road slope detected by an inclination sensor, steering state recorded by a steering angle sensor, and remaining power and motor output power uploaded synchronously by a battery management system; in addition, an environmental sensor identifies the road condition to provide environmental parameters for subsequent mode matching; the data preprocessing includes: the central processing unit filters, normalizes and extracts features from the collected raw data to form a structured real-time data matrix, providing high-quality input for subsequent steps.

[0009] Preferably, the pressure distribution sensor adopts a flexible thin film design, covers the entire pedal contact surface, and outputs a pressure distribution heat map of the foot palm contact area in real time through a pressure sensing point array.

[0010] Preferably, the filtering in the data preprocessing adopts a Kalman filtering algorithm for dynamic noise reduction, the state equation of each sensor is established to predict the next time data in real time and compare it with the actual collected value, and the error is controlled within 5%; secondly, the min-max normalization method is used to map different dimensional parameters to the 0-1 interval to eliminate the influence of dimensional differences on subsequent algorithms.

[0011] Preferably, step S2 specifically comprises: mode library construction and parameter presetting: presetting a basic mode library containing standard mode, overload mode, ramp mode, energy-saving mode and comfort mode, each mode corresponding to differentiated braking characteristic parameters, and storing typical pedal force feedback curves in each mode as initial reference; dual-path mode selection mechanism: manual selection: the operator directly selects the target mode through the forklift human-machine interface, and the system responds in real time and loads the corresponding parameters, which is suitable for scenarios where the operator clearly needs the working condition; intelligent recommendation: the central processing unit performs multi-dimensional evaluation based on real-time data; combined with the operator's historical selection preferences, the recommended mode and the recommended reason are displayed on the pop-up window of the human-machine interface, and the operator can confirm or select other modes with one key; mode smooth switching control: when the mode is switched, the system avoids the operator's misoperation caused by the sudden change of pedal force feedback by gradually adjusting the actuator, and records the vehicle state at the switching time to ensure the continuity of the braking effect during the switching process.

[0012] Preferably, the forklift human-machine interface adopts a "physical button + touch screen" dual-entry design: a mode switching lever is arranged on the left side of the steering wheel, and mode icons are displayed on the home page of the touch screen, and the operation response delay is less than 50 ms, meeting the needs of the operator's active decision.

[0013] Preferably, step S3 specifically comprises: algorithm model construction: a hybrid algorithm architecture of "rule base + machine learning" is adopted, wherein the rule base is embedded with the physical law of forklift braking, and the machine learning model is trained through historical operation data and has self-learning ability, the input layer is the multi-dimensional feature parameters preprocessed in step S1, and the output layer is multiple feedback characteristic parameters such as damping coefficient, peak rebound force and feedback curve slope; dynamic optimization calculation logic: real-time intention recognition: the operator's intention is judged through the change rate of pedal speed and force, and the algorithm adjusts the feedback response speed accordingly; working condition adaptive correction: dynamically correcting the reference feedback curve in combination with the forklift state; energy recovery coordination: in energy-saving mode, the algorithm synchronously correlates the braking energy recovery system parameters, and when calculating the pedal force feedback, the damping characteristics with the optimal recovery efficiency are preferentially matched to balance the braking effect and energy recovery; output parameter verification: the feedback characteristic parameters calculated need to pass through a safety verification module to ensure that the parameters are within the bearing range of the forklift mechanical structure, and to avoid pedal failure caused by excessive adjustment.

[0014] Preferably, the historical operation data includes multiple samples, each sample containing "input features + output parameters + effect labels", and the model generalization ability is ensured through 5-fold cross-validation, and the model is designed to be lightweight when deployed to ensure that the single calculation time on the central processing unit is less than 10 ms.

[0015] Preferably, in the working condition adaptive correction, the algorithm dynamically adjusts the reference curve for different working conditions: when the road is wet and slippery, the damping force peak is reduced by 10%, and the sensitivity of the rebound force is increased; when the load is unbalanced, the left and right wheel braking force distribution algorithm is used to make the left side of the pedal damping force 10% higher than the right side; when working at low speed, the damping force is weakened, and the fine tuning accuracy is improved; in energy recovery coordination, the algorithm in energy saving mode communicates with the battery management system in real time: when the BMS feedback SOC < 30%, the braking time is preferentially extended, and the energy recovery amount is increased by 15%; when SOC ≥ 90%, the standard damping force is restored, and the operator is prompted to reduce the braking frequency through the rebound force curve.

[0016] Preferably, step S4 specifically includes: actuator cooperative control: electrically controlled adjustable damper: according to the damping force coefficient parameter, the damping oil flow is adjusted through the proportional electromagnetic valve to change the resistance in the pedal stepping process in real time; servo motor and transmission mechanism: receiving rebound force parameter, driving pedal return through screw nut structure, outputting rebound force matching displacement and speed; auxiliary buffer device: setting electrically controlled buffer at the pedal stroke endpoint, when the pedal approaches the maximum displacement, avoiding rigid impact by reducing the damping change rate, improving the operation comfort; closed loop feedback regulation: displacement sensor and pedal force sensor monitor the actual pedal state in real time, feedback the data to the central processing unit, dynamically correct the actuator output through the PID control algorithm, ensure that the deviation between the feedback characteristics and the calculated value is controlled within ± 3%; extreme working condition emergency regulation: when the sensor detects abnormal state, the system immediately triggers the emergency mode, provides basic rebound force through the standby spring mechanism, and alarms on the human-machine interface at the same time, while retaining the mechanical braking function to ensure operation safety.

[0017] Preferably, in the real-time cycle of the closed loop feedback regulation mechanism "calculation-execution-correction", the displacement sensor and the pedal force sensor send actual values to the central processing unit every 10ms, the error value is output by the deviation calculation module, and the PID controller dynamically adjusts the output signal according to the error: the proportional term quickly responds to the current error, the integral term eliminates the cumulative error, and the differential term suppresses the overshoot; the parameters of closed loop control can be dynamically adjusted according to the mode: increase the proportional term in heavy load mode to improve the response speed, and increase the integral term in comfort mode to reduce the fluctuation.

[0018] Preferably, step S5 specifically comprises: controlling the target deviation within ±5% through brake distance deviation and speed adjustment error evaluation, thereby improving operation accuracy; then analyzing the uniformity of the force on the soles through the pressure distribution sensor installed at the pedal, combining with the physiological sensor worn by the operator to monitor the operation fatigue, and the comfort score is greater than or equal to 85, thereby improving the operation comfort; in the energy-saving mode, the brake energy recovery rate is counted, and the safety indicators such as the number of emergency braking and the distance of coasting are recorded to ensure zero safety accidents; then data storage and analysis are carried out, in this process, the central processing unit stores the complete data of each braking to form a historical database, and the edge computing module is used to regularly clean and mine the data to identify typical operation habits and high-frequency working conditions, and finally the model and mode are iterated, the machine learning model is incrementally trained based on historical data every month, and the algorithm weight is optimized; the mode library parameters are updated according to the high-frequency working condition characteristics; the subjective comfort parameters of the feedback curve are adjusted combined with the operator feedback to realize the continuous optimization of the system performance.

[0019] Preferably, in the process of realizing the continuous evolution of the system through the iteration of the model and the mode, the machine learning model is incrementally trained once a month: based on the newly added multiple groups of data, the transfer learning strategy is adopted, the training time is less than 2 hours to avoid affecting the normal operation, and the model after training needs to be tested offline to ensure that the accuracy is improved by more than 3% before being deployed to the forklift; the mode library parameters are updated once a quarter: for high-frequency working conditions, transition sub-modes are added between existing modes; the subjective parameters are adjusted according to the operator feedback; the iteration effect is verified through A / B testing: randomly select 10 forklifts to install the new version of the model, and compare them with 10 forklifts that have not been updated for 30 days, if the comfort score of the new version is improved by more than 5 points and the brake distance deviation is reduced by more than 2%, the new version is promoted.

[0020] The application further provides a central processing unit configured to execute the program of the balancing method of the pedal brake of the counterbalanced forklift as described in any one of the above embodiments.

[0021] The application further provides a computer readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the steps of the balancing method of the pedal brake of the counterbalanced forklift as described in any one of the above embodiments.

[0022] The application further provides a counterbalanced forklift adopting the central processing unit as described in the above embodiments.

[0023] The application further provides a computer program product comprising a computer program / instruction, which, when executed by a processor, implements the steps of the balancing method of the pedal brake of the counterbalanced forklift as described in any one of the above embodiments.

[0024] The beneficial effects of the present application are: compared with the prior art, the application effectively improves the adaptability of the brake system to complex working conditions, dynamically matches the pedal feedback characteristics with the operation intention and the vehicle state. The system can automatically identify the emergency braking demand and optimize the response speed, reducing the physical exertion of the operator. By continuously learning the operation habits and environmental characteristics, the human-machine cooperation efficiency is significantly improved. While ensuring braking safety, the balance optimization of energy recovery efficiency and handling comfort is realized.

[0025] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 is a flowchart of the counterbalanced forklift pedal brake balancing method of the present application. DETAILED DESCRIPTION

[0027] Embodiments of the present application are described in detail below with reference to examples illustrated in the accompanying drawings, in which the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0028] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0029] Reference is made below Figure 1 The counterbalanced forklift pedal brake balancing method in the embodiments of the present application is described.

[0030] A counterbalanced forklift pedal brake balancing method is disclosed in the embodiments of the present application, comprising the following steps:

[0031] S1, multi-dimensional dynamic data real-time acquisition and preprocessing: collecting operator pedal operation characteristics, forklift running state and environmental data, and preprocessing the collected raw data by using a central processing unit;

[0032] S2, intelligent matching and dynamic switching of brake mode: based on the preprocessed real-time data, combining the preset mode library and intelligent decision logic, precise matching and smooth switching of the brake mode are realized;

[0033] S3, dynamic optimization calculation of pedal force feedback characteristics based on intelligent algorithm: using a central processing unit to run an intelligent algorithm that integrates multi-dimensional data, real-time calculation of optimal pedal force feedback characteristic parameters that adapt to current working conditions and operation intentions;

[0034] S4, real-time adjustment of actuator response and pedal force: based on the optimal feedback characteristic parameters output by S3, the actuator of the intelligent brake pedal force adjustment system is adjusted to achieve precise and dynamic adjustment of the pedal force;

[0035] S5, effect evaluation and adaptive iterative optimization: through multi-dimensional evaluation of the adjustment effect, combined with long-term running data, the algorithm and mode library are iteratively updated to continuously improve human-machine collaboration performance.

[0036] Among them, multi-dimensional dynamic data real-time acquisition and preprocessing refers to obtaining three-dimensional data sets of operation behavior, vehicle state and environmental parameters through multiple types of sensors, such as using force sensors and displacement sensors to capture pedal operation characteristics. Brake mode intelligent matching and dynamic switching involves establishing a baseline parameter library containing multiple preset working conditions, such as automatically activating the slope mode when the detected slope exceeds the threshold. Dynamic optimization calculation based on intelligent algorithm uses a hybrid architecture that combines rule constraints and machine learning, such as combining brake physical equations with neural network predictions to generate control parameters. Actuator response and real-time adjustment of pedal force include the coordinated control of electrically controlled dampers and servo motors, such as adjusting oil flow through proportional electromagnetic valves to change pedal resistance. Effect evaluation and adaptive iterative optimization build an evaluation system that includes performance indicators and operation feedback, such as adjusting algorithm weights based on brake distance deviation.

[0037] Specifically, the system first collects the operator's pedal force, displacement trajectory and environmental parameters through a distributed sensor network, and forms a standardized data matrix after filtering. The central processing unit matches the most suitable braking mode according to the real-time data stream, and automatically switches to the load compensation mode when detecting abnormal load distribution. The hybrid algorithm model simultaneously analyzes the operation intention and vehicle dynamics state, dynamically calculates the damping coefficient and rebound force curve parameters. The actuator adjusts the pedal mechanical properties according to the optimized parameters, and the closed-loop feedback mechanism continuously corrects the actual output value. The data of each braking process is recorded in the historical database, and the operation rules are identified through edge computing and the control model is updated.

[0038] Compared with the prior art, the traditional scheme can only realize single brake curve control, while the present method realizes dynamic parameter adjustment through multi-dimensional data fusion. The manual mode switching of the prior art has a response delay, and the intelligent recommendation mechanism of the present scheme can predict the working condition change in advance. The traditional open-loop control system is prone to cumulative errors, and the closed-loop feedback mechanism of the present scheme can correct the execution deviation in real time. The existing fixed algorithm is difficult to adapt to changes over time, and the iterative optimization mechanism of the present scheme enables the system to continuously evolve.

[0039] Through the above technical solutions, the application effectively improves the adaptability of the brake system to complex working conditions, dynamically matches the pedal feedback characteristics with the operation intention and the vehicle state. The system can automatically identify the emergency braking demand and optimize the response speed, reducing the physical exertion of the operator. By continuously learning the operation habits and environmental characteristics, the human-machine cooperation efficiency is significantly improved. While ensuring braking safety, the balance optimization of energy recovery efficiency and handling comfort is realized.

[0040] In some embodiments, the operator pedal operation feature acquisition includes: collecting the stepping force in real time by using the pedal force sensor installed at the brake pedal, recording the pedal displacement and the change rate by the displacement sensor, capturing the stepping frequency by the high-frequency sampling module, and assisting in judging the operation intention by the pressure distribution sensor at the pedal to obtain the foot palm contact position; the forklift running state and environmental data acquisition includes: collecting key parameters by the vehicle state sensor group, including the load sensor to obtain the real-time load and load distribution, the speed sensor to collect the driving speed and acceleration, the inclination sensor to detect the road slope, the steering angle sensor to record the steering state, and the battery management system to synchronously upload the remaining power and motor output power; in addition, the environmental sensor identifies the road conditions to provide environmental parameters for subsequent mode matching; the data preprocessing includes: the central processing unit filters, normalizes and extracts features from the collected raw data to form a structured real-time data matrix, providing high-quality input for subsequent steps.

[0041] Among them, the pedal force sensor refers to a device for measuring the force value applied by the operator on the brake pedal in real time, which converts mechanical force into electrical signal to provide quantitative basis for judging operation strength. The displacement sensor refers to a device for detecting the displacement and speed of the pedal, which captures the operation dynamic characteristics through the displacement change rate. The high-frequency sampling module refers to a circuit unit that collects signals at a fixed frequency to ensure that transient operation characteristics are recorded completely. The pressure distribution sensor refers to a multi-dimensional pressure detection device covering the contact surface of the pedal, which captures the pressure distribution difference of the foot palm contact area to assist in identifying the operation intention. The load sensor refers to a force detection device installed on the fork carriage, which is used to monitor the weight and center of gravity of the goods in real time. The environmental sensor refers to a detection device for identifying external road conditions, which analyzes the road texture and friction coefficient to provide environmental parameters for mode matching. Normalization refers to a data processing method that converts parameters of different dimensions to a unified interval, eliminating the influence of parameter magnitude difference on subsequent calculation. Feature extraction refers to the calculation process of extracting effective information from raw data to form a structured data matrix representing operation characteristics and working condition state.

[0042] Specifically, the force, displacement and frequency information of the operator's stepping action are synchronously collected by multiple types of sensors, combined with the pressure distribution characteristics of the foot contact position, to construct a multi-dimensional operation feature dataset. The forklift running state parameters are obtained in real time by a vehicle sensor group, including load distribution, driving speed, road slope and steering angle, while integrating battery power and environmental parameters to form a comprehensive working condition dataset. The central processing unit filters and denoises the original data, eliminating high-frequency interference components in the sensor signal, and then maps different dimension parameters to a unified numerical interval through a normalization algorithm, and finally extracts key feature quantities to form a standardized data matrix. The data matrix serves as the input for subsequent pattern matching and optimization calculations, ensuring that the brake control system can accurately identify the operation intention and perceive the real-time working conditions.

[0043] Compared with the prior art, the traditional forklift brake system only relies on a single pedal force or displacement signal for control, which cannot distinguish between full-foot emergency braking and toe fine-tuning operations, and lacks dynamic perception of load distribution and road conditions. The present scheme builds a comprehensive data acquisition system covering operation features, vehicle status and environmental parameters by integrating multi-dimensional sensor data and intelligent preprocessing mechanisms, effectively solving the intention misjudgment and working condition perception deviation problems caused by the lack of data dimensions in traditional systems.

[0044] Through the above technical solutions, the present application realizes fine feature extraction of the operator's stepping action and comprehensive perception of complex working conditions, providing high-precision data support for brake pattern matching. The multi-dimensional data fusion processing mechanism effectively improves the accuracy of operation intention recognition and the reliability of working condition state judgment, solving the control deviation problem caused by single data acquisition in traditional systems, and laying a precise data foundation for dynamic brake optimization.

[0045] In some embodiments, the pressure distribution sensor adopts a flexible thin film design, covering the entire pedal contact surface, and outputs a pressure distribution heat map of the foot contact area in real time through a pressure sensing point array. The filter in data preprocessing uses Kalman filter algorithm for dynamic noise reduction, by establishing the state equation of each sensor, real-time prediction of next time data and comparison with actual collection value, the error is controlled within 5%, secondly, the min-max normalization method is used to map different dimension parameters to 0-1 interval, eliminate the influence of dimension difference on subsequent algorithm.

[0046] The flexible film type design refers to a sensor structure made of a deformable polymer material, which can completely cover the contact area by adhering to the curved surface of the pedal. The pressure sensing dot matrix refers to a detection grid composed of micro pressure sensing units arranged at equal intervals, which can be implemented by a 20x20 array, and each sensing unit independently detects the local pressure value. The Kalman filter algorithm dynamic noise reduction refers to predicting the data trend by establishing a time series model of the sensor output, which can be implemented by a second-order autoregressive model, and the predicted value and the measured value are weighted and fused to eliminate random noise. The min-max normalization method refers to linearly converting the original data to a specific numerical interval, which can be implemented by setting the maximum and minimum values of each parameter, and eliminating the interference of different physical dimensions on the algorithm input.

[0047] Specifically, the flexible film sensor captures the pressure distribution characteristics of each region of the foot by densely distributed sensing units, forming a pressure heat map with spatial resolution. The Kalman filter algorithm establishes an independent state equation for each sensing unit, predicts the pressure value at the next time and dynamically calibrates with the actual measured value, effectively suppressing the signal fluctuations caused by environmental vibration. Normalization processing converts the original data of different dimensions such as pedal force, displacement speed, slope angle, etc. into dimensionless values, so that the subsequent machine learning algorithm can process multi-dimensional feature parameters evenly. These three technical means form a technical closed loop of data acquisition and processing, providing high-precision, low-noise input data for operation intention recognition.

[0048] Compared with the prior art, the traditional scheme uses a single pressure sensor to detect only the overall pedal force, which cannot distinguish the force characteristics of different regions of the foot. The conventional mean filter method will cause phase delay when processing dynamic signals, affecting real-time performance. The present scheme combines full-contact surface pressure detection with intelligent noise reduction algorithm, controls the data error within 5% under the premise of maintaining real-time performance, and eliminates the dimensional barriers when fusing multi-dimensional parameters through standardization processing.

[0049] Through the above technical solutions, the present application realizes accurate perception of the operator's foot contact characteristics, effectively distinguishes the operation differences between full-foot force and toe fine-tuning, and reduces the intention recognition delay to within 50 milliseconds. Dynamic noise reduction processing suppresses high-frequency interference in the sensor signal, ensuring data reliability. The normalization method solves the compatibility problem of multi-source heterogeneous data, provides standardized input features for intelligent algorithms, and significantly improves the accuracy and response speed of brake feedback adjustment.

[0050] In some embodiments, step S2 specifically comprises: mode library construction and parameter presetting: presetting a basic mode library containing standard mode, overload mode, ramp mode, energy-saving mode, and comfort mode, each mode corresponding to differentiated braking characteristic parameters, and storing typical pedal force feedback curves in each mode as initial references; dual-path mode selection mechanism: manual selection: the operator directly selects a target mode through the forklift human-machine interface, and the system responds in real time and loads corresponding parameters, which is suitable for scenarios where the operator clearly needs the working condition; intelligent recommendation: the central processing unit performs multi-dimensional evaluation based on real-time data; in combination with the operator's historical selection preferences, the recommended mode and the recommended reasons are displayed on the human-machine interface pop-up window, and the operator can confirm or select other modes with one key; mode smooth switching control: when the mode is switched, the system avoids operator misoperation caused by sudden changes in pedal force feedback by gradually adjusting the actuator, and records the vehicle state at the switching time to ensure the continuity of the braking effect during the switching process. The forklift human-machine interface adopts a "physical button + touch screen" dual-entry design: a mode switching lever is arranged on the left side of the steering wheel, and mode icons are displayed on the home page of the touch screen, and the operation response delay is less than 50 ms, meeting the needs of the operator's active decision-making.

[0051] The mode library construction and parameter presetting refers to pre-setting a set of braking parameters corresponding to different working conditions, which can specifically be associated and stored by using a parameterized modeling method to link braking characteristic parameters and pedal force feedback curves, and a mapping relationship between modes and parameters is established to provide reference adjustment targets for different working conditions. The dual-path mode selection mechanism refers to an interactive way of combining operator's active selection and system's intelligent recommendation, which can specifically be realized by a human-computer interaction module to realize bidirectional transmission of operation instructions, while retaining manual operation authority and introducing intelligent decision support. The mode smooth switching control refers to a technical means to avoid sudden changes in braking parameters, which can specifically use a linear interpolation algorithm to transitionally adjust the output of the actuator to ensure the continuity of the pedal feedback through gradual changes in parameters over time. The dual-entry design of the forklift human-machine interface refers to the cooperative operation interface of physical buttons and touch screens, which can specifically use an embedded control system to realize synchronous processing of signals of the two input methods to improve operation convenience through hardware layout optimization.

[0052] Specifically, the mode library construction stage stores braking parameters under different working conditions by classification to provide reference for subsequent mode matching. During the operation of the dual-path selection mechanism, the operator can manually switch modes according to experience, or the system can trigger intelligent recommendation based on real-time collected load and slope data. When the mode switching instruction is generated, the actuator gradually adjusts parameters such as damping force and rebound force according to the current vehicle state parameters using a time control algorithm. The human-machine interface provides a quick switching channel through physical buttons, and the touch screen displays the current mode and recommendation information in real time. The data of the two input methods are analyzed by the central processing unit and then updated synchronously.

[0053] Compared with the prior art, the traditional forklift brake mode switching only supports a single manual selection mode, cannot provide intelligent recommendations according to real-time working conditions, and the parameter adjustment has a step change when the mode is switched, which can easily cause sudden pedal feedback. The application fuses artificial decision and intelligent analysis through a double-path selection mechanism, adopts a parameter gradual change algorithm to eliminate mechanical impact during mode switching, optimizes the layout of the human-machine interface, and solves the technical defects of the traditional scheme, such as harsh operation and single decision.

[0054] Through the above technical solutions, the application balances the operation comfort and system intelligence during brake mode switching, effectively avoids the discomfort caused by parameter mutation, ensures the coherence and safety of the transition between different modes, and takes into account the subjective selection needs of the operator and the working condition adaptability of the system through the double-path selection mechanism.

[0055] In some embodiments, step S3 specifically includes: algorithm model construction: a hybrid algorithm architecture of "rule base + machine learning" is adopted, wherein the rule base embeds the physical law of forklift braking, and the machine learning model is trained through historical operation data and has autonomous learning ability, the input layer is the multi-dimensional feature parameters preprocessed in step S1, and the output layer is multiple feedback characteristic parameters such as damping coefficient, rebound peak, feedback curve slope, etc.; dynamic optimization calculation logic: real-time intention recognition: the operator's intention is judged through the change rate of the stepping speed and force, and the algorithm adjusts the feedback response speed accordingly; working condition adaptive correction: the reference feedback curve is dynamically corrected in combination with the forklift state; energy recovery coordination: in the energy-saving mode, the algorithm synchronously correlates the brake energy recovery system parameters, when calculating the pedal force feedback, the damping characteristics with the optimal recovery efficiency are preferentially matched to balance the braking effect and energy recovery; output parameter verification: the feedback characteristic parameters obtained by calculation need to pass through a safety verification module to ensure that the parameters are within the bearing range of the forklift mechanical structure, and to avoid pedal failure caused by excessive adjustment.

[0056] The rule base refers to a database storing the basic physical laws of the forklift braking system, which is implemented by an embedded storage module and contains core formulas such as the relationship between braking distance and speed square and the relationship between load and braking torque, to ensure that the algorithm decision conforms to the mechanical motion law. The machine learning model refers to a computing module with autonomous learning ability, which can be implemented by an improved BP neural network structure, and the weight matrix is trained by inputting multi-dimensional feature parameters to realize dynamic fitting of nonlinear relationship. Real-time intention recognition refers to a computing module that analyzes the change rate of stepping speed and force, which specifically extracts the time derivative of operation characteristics by using a difference algorithm to distinguish the behavior difference between emergency braking and regular operation. The working condition adaptive correction refers to a parameter adjustment mechanism based on the real-time running state of the forklift, which specifically realizes the mapping of parameters such as load and slope to the correction amount of the feedback curve by establishing a dynamic correction coefficient matrix. Energy recovery coordination refers to a linkage control module of braking characteristic parameters and energy recovery system, which specifically realizes parameter sharing by using a cross-system communication protocol to balance the relationship between damping characteristics and recovery efficiency in energy-saving mode. The output parameter verification refers to a safety threshold constraint module, which has built-in mechanical structure bearing limit parameters and ensures that the output parameters are in the safe interval through a boundary condition judgment algorithm.

[0057] Specifically, in the algorithm model construction phase, the rule base pre-sets the basic physical constraint conditions of the forklift braking system, such as the positive correlation between braking torque and load, to prevent the algorithm output from violating mechanical principles. The machine learning model is trained by historical operation data, such as sample data covering multiple operators and working conditions, to learn the mapping relationship between different stepping behaviors and braking effects. In the dynamic optimization calculation process, the real-time intention recognition module continuously monitors the first derivative of the stepping speed and the rate of force change. When a sudden change in stepping speed is detected beyond a threshold value, it is determined as an emergency braking requirement and triggers a fast response mechanism. The working condition adaptive correction module dynamically adjusts the slope and peak value of the reference feedback curve according to the real-time collected load distribution and slope data, such as automatically increasing the initial damping force coefficient when a slope working condition is detected. The energy recovery coordination module is started in energy-saving mode, which dynamically adjusts the damping characteristics to maximize the energy recovery efficiency at the initial stage of braking by calling the remaining power parameters of the battery management system. The output parameter verification module performs safety verification after each calculation is completed, such as comparing the rebound force peak value with the maximum allowed value of the mechanical structure design, and automatically correcting it to a safe value when the threshold value is exceeded.

[0058] Compared with the prior art, the traditional forklift brake system adopts a fixed parameter control strategy, which cannot adjust the pedal feedback characteristics according to real-time working conditions, resulting in a deviation between the brake effect and the operation intention. The single rule control method in the prior art is difficult to cope with complex and variable load distribution and road conditions, and lacks an energy recovery coordination mechanism. The scheme fuses physical rules and machine learning models, ensuring the stability of the basic brake performance, and being self-adaptive to different operation habits and working conditions. The introduction of a real-time intention recognition mechanism solves the problem of unresponsive operation caused by the response delay of the traditional system. The working condition self-adaptive correction module breaks through the limitations of the fixed parameter mode, realizing the dynamic matching of the brake characteristics and the vehicle state. The energy recovery coordination mechanism fills the technical gap in energy efficiency optimization of the traditional system, improving energy utilization rate while ensuring brake effect.

[0059] Through the above technical scheme, the application realizes real-time dynamic optimization of pedal force feedback characteristics, effectively solving the technical problem that the traditional system cannot adapt to complex working conditions and operation intention. The operator obtains the expected pedal feedback under different loads and slope conditions, avoiding operation errors caused by system response lag. The coordinated control of brake characteristic parameters and the energy recovery system significantly improves energy utilization efficiency while maintaining brake efficiency. The safety verification mechanism ensures that the algorithm output is always within the safe range of the mechanical structure, preventing system failures caused by parameter over-limit. This method makes the forklift brake system self-adaptive, maintaining stable control performance and energy efficiency in diversified operation scenarios.

[0060] In some embodiments, the historical operation data includes multiple groups of samples, each containing "input features + output parameters + effect labels". Through 5-fold cross-validation, the model generalization ability is ensured, and lightweight design is adopted during model deployment to ensure that the single calculation time on the central processing unit is <10ms. In the working condition adaptive correction, the algorithm dynamically adjusts the reference curve for different working conditions: on wet and slippery roads, the damping force peak is reduced by 10%, and the sensitivity of the rebound force is increased; in the partial load working condition, the left and right wheel braking force distribution algorithm is used to make the pedal left damping force 10% higher than the right side; in low-speed operation, the damping force is weakened, and the fine-tuning accuracy is improved; in energy recovery coordination, the algorithm in energy saving mode communicates with the battery management system in real time: when the BMS feedback SOC <30%, the brake time is extended to increase the energy recovery amount by 15%; when SOC ≥90%, the standard damping force is restored, and the operator is prompted to reduce the brake frequency through the rebound force curve.

[0061] Among them, five-fold cross-validation refers to dividing the data set into five mutually exclusive subsets, training the model using four subsets each time, and verifying the model performance with the remaining subset, taking the average result after five cycles. Specifically, stratified sampling method can be used to ensure consistent data distribution in each fold to prevent model overfitting and improve generalization ability. Lightweight design refers to reducing model parameter quantity through model pruning and quantization compression technology. Specifically, knowledge distillation method can be used to migrate the knowledge of complex model to lightweight model to ensure real-time calculation under limited computing power. The peak value of damping force is reduced by a certain percentage, which means that under wet road conditions, the maximum damping force output is reduced. Specifically, the opening of the proportional electromagnetic valve can be adjusted dynamically to prevent tire lock due to excessive braking force.

[0062] The rebound force sensitivity is increased, which means that the response relationship between the change amount of rebound force per unit time and the change of pedal operation is enhanced. Specifically, by improving the control accuracy of the servo motor speed, the operator can obtain more timely reverse force when quickly releasing the pedal. The left and right wheel braking force distribution algorithm refers to dynamically distributing the braking torque on both sides according to the degree of load transfer. Specifically, a differential damping adjustment mechanism can be used to make the damping forces on both sides of the pedal different, and through tactile feedback, the operator can be prompted to correct the steering operation. The weakening of the damping force refers to reducing the pedal movement resistance under low-speed working conditions. Specifically, by reducing the reference pressure value of the electronically controlled adjustable damper, the operator can easily perform millimeter-level displacement fine adjustment. The energy recovery coordination mechanism refers to dynamically adjusting the braking parameters according to the battery state. Specifically, the real-time data interaction between the battery management system and the central processing unit can be used to extend the braking energy recovery time when the battery is low, and restore the standard parameters to avoid overload when the battery is high.

[0063] Specifically, the historical operation data sample is stored with multiple dimensional feature parameters and corresponding braking effect labels to construct a training data set. The cross-validation method is used to optimize the model parameters to ensure the adaptability of the model under different operators and working conditions. The lightweight model processing reduces the occupation of computing resources through compression algorithm to meet the real-time control demand. In wet road conditions, the algorithm automatically reduces the upper limit of damping force output, while enhancing the response speed of rebound force, so that the operator can still maintain accurate control of the pedal under the risk of slipping. In the load transfer condition, the system adjusts the damping forces on the left and right pedals asymmetrically to produce differential tactile feedback, guiding the operator to actively adjust the steering angle to correct the load distribution. At low speed, the damping force reference value is dynamically reduced, and high-precision displacement sensing is used to realize the fine braking adjustment required for shelf alignment and other fine operations. In the energy recovery coordination process, the battery state is monitored in real time. When the battery is low, the braking parameters are matched with the energy recovery efficiency first, and the energy conversion rate is improved by prolonging the braking time; when the battery is high, the standard parameters are restored, and the operator is prompted to reduce unnecessary braking operations through the change of rebound force curve shape.

[0064] Compared with the prior art, the traditional scheme adopts a fixed parameter braking system, which cannot dynamically adjust the feedback characteristics according to the road conditions, load distribution and battery state, resulting in easy side slip on wet road, steering correction lag in unbalanced load conditions, insufficient low-speed fine adjustment accuracy and low energy recovery efficiency. The scheme realizes real-time optimization of braking parameters through a multi-dimensional data-driven adaptive algorithm, actively adjusts the damping force and rebound force characteristics under complex working conditions, and establishes a collaborative mechanism between the braking system and the energy management system, overcoming the low control performance and energy efficiency problems caused by parameter fixation in traditional technology.

[0065] Through the above technical scheme, the application solves the dynamic adaptation problem of the braking system under complex working conditions, improves the control safety on wet road, enhances the steering correction prompt effect in unbalanced load conditions, optimizes the micro-operation precision of low-speed operation, and realizes the balance between energy recovery efficiency and braking safety. The improvement of model generalization ability ensures that the system adapts to different operation habits, the optimization of real-time calculation performance ensures the timely response of control instructions, and the collaborative adjustment of multi-dimensional parameters significantly improves the human-machine interaction efficiency and energy utilization efficiency.

[0066] In some embodiments, step S4 specifically includes: executive mechanism collaborative control: electrically controlled adjustable damper: adjusting the damping oil flow through proportional electromagnetic valve according to damping force coefficient parameter, changing the resistance in the process of pedal depression in real time; servo motor and transmission mechanism: receiving rebound force parameter, driving pedal return through screw nut structure, outputting rebound force matching displacement and speed; auxiliary buffer device: setting electrically controlled buffer at pedal stroke endpoint, avoiding rigid impact by reducing damping change rate when pedal approaches maximum displacement, improving operation comfort; closed-loop feedback regulation: displacement sensor and pedal force sensor monitor actual pedal state in real time, feeding back data to central processing unit, dynamically correcting executive mechanism output through PID control algorithm, ensuring that the deviation between feedback characteristics and calculated value is controlled within ±3%; extreme condition emergency regulation: when sensor detects abnormal state, system immediately triggers emergency mode, providing basic rebound force through standby spring mechanism, and alarming on human-machine interface at the same time, while retaining mechanical braking function, ensuring operation safety.

[0067] The electrically controlled adjustable damper refers to a device for adjusting fluid flow through a solenoid valve, which can be implemented by a proportional solenoid valve cooperating with a hydraulic circuit to adjust the damping force by changing the oil flow area. The servo motor and transmission mechanism refers to a device that converts electrical signals into mechanical motion, which can be implemented by a high-precision servo motor driving a ball screw structure to convert the rebound force parameter into linear displacement output, ensuring real-time synchronization of the rebound force and the pedal movement state. The auxiliary buffer device refers to a buffer mechanism installed at the end of the pedal stroke, which can be implemented by an electrically controlled variable damping buffer to automatically reduce the damping change gradient when the pedal approaches the limit position, for eliminating the discomfort caused by rigid impact. The closed-loop feedback regulation refers to a control mechanism that corrects the execution output based on sensor data in real time, which can be implemented by a PID controller for dynamic compensation of the output of the actuator, for eliminating system response delay and mechanical error. The extreme condition emergency regulation refers to a backup mechanism enabled in system abnormalities, which can be implemented by a redundant system composed of a mechanical spring and an independent control circuit to maintain the basic braking function when the electronic system fails.

[0068] Specifically, the coordinated control of the actuator is realized by the linkage of multiple components. The electrically controlled adjustable damper receives the damping coefficient instruction sent by the central processing unit, and accurately adjusts the oil flow in the hydraulic circuit through the proportional solenoid valve, so that a resistance change curve matching the current working condition is generated during the pedal stepping process. The servo motor drives the screw mechanism according to the rebound force parameter, converting the rotary motion into linear return action of the pedal, and the movement speed and displacement are real-time corresponding to the rebound force setting value. The auxiliary buffer device starts the buffer program at the end of the pedal stroke, dynamically adjusts the damper characteristics of the buffer, and makes the pedal present gradual resistance change near the limit position. The closed-loop feedback regulation system continuously collects pedal actual state data, calculates the actuator output compensation amount through the PID algorithm, and forms a continuous error correction cycle. When the sensor detects that the actuator is stuck or the signal is abnormal, the emergency system immediately switches to the backup spring mechanism, and at the same time triggers the audible and visual alarm and keeps the mechanical braking function available.

[0069] Compared with the prior art, the traditional forklift braking system mostly uses mechanical damper with fixed parameters, which cannot realize dynamic resistance adjustment and lacks real-time feedback compensation mechanism, resulting in limited adjustment accuracy. The actuator in the prior art mostly relies on single driving mode, which is difficult to handle the combined adjustment demand of resistance and rebound force. The present scheme combines multi-actuator coordinated control and closed-loop feedback to build a dynamic adaptive adjustment system, overcoming the defects of response lag and insufficient accuracy of traditional systems. Compared with the open control architecture commonly used in the prior art, the closed-loop regulation mechanism introduced in the present scheme can effectively eliminate the accumulation of execution errors. In addition, the traditional emergency scheme mostly uses complete mechanical backup, and the present scheme realizes seamless connection of electronic control and mechanical braking through intelligent switching mechanism.

[0070] Through the technical scheme, the application solves the problems of response lag and insufficient regulation accuracy of the traditional actuator, and realizes dynamic and accurate control of the pedal force. The cooperative work of the multiple actuator components ensures the synchronous optimization of the resistance and rebound characteristics, and the closed-loop feedback mechanism effectively improves the stability and reliability of system regulation. The double protection design under extreme conditions eliminates the risk of brake failure caused by intelligent system failure, improves the regulation accuracy, and ensures the operation safety. The introduction of the buffer device optimizes the operation experience of the pedal end stroke, and avoids the influence of mechanical impact on the operation comfort.

[0071] In some embodiments, in the real-time cycle of the "calculation-execution-correction" realized by the closed-loop feedback regulation mechanism, the displacement sensor and the pedal force sensor send actual values to the central processing unit every 10 ms, the deviation calculation module outputs error values, and the PID controller dynamically adjusts the output signal according to the error: the proportional term quickly responds to the current error, the integral term eliminates cumulative error, and the differential term suppresses overshoot; the parameters of the closed-loop control can be dynamically adjusted according to the mode: the proportional term is increased in heavy load mode to improve the response speed, and the integral term is increased in comfort mode to reduce fluctuations.

[0072] Among them, the closed-loop feedback regulation mechanism refers to the real-time data loop formed by sensors, controllers and actuators, which is realized by a combination of displacement sensors and pedal force sensors, and actual pedal state parameters are periodically collected and compared with target values to generate error signals, providing a data basis for dynamic correction.

[0073] Among them, the dynamic adjustment of the PID controller parameters refers to automatically adjusting the weight coefficients of the proportional, integral and differential terms according to the current operating mode, which is realized by using the parameter mapping table built-in programmable logic controller, and the parameter switching instruction is triggered by the mode recognition module to match the control characteristics with the working condition requirements.

[0074] Among them, high-frequency data acquisition refers to acquiring sensor data at fixed time intervals, which is realized by using the timing interrupt trigger method, and by setting the data acquisition period and signal processing priority, it ensures that control links with high real-time requirements have priority to obtain computing resources.

[0075] Specifically, the system continuously receives real-time measurements of pedal displacement and force during operation, and calculates the difference between the measurements and the target parameters every fixed period to generate a deviation signal containing error size and change trend. After inputting the deviation signal into the PID controller, the control instruction with the characteristics of advance adjustment is generated through the triple operation of the proportional term, the integral term and the differential term. The proportional term compensates for the current error amplitude quickly, the integral term gradually eliminates the accumulation of historical errors, and the differential term predicts the error trend and suppresses overshoot in advance. When the system identifies that it is currently in heavy load mode, the proportional term coefficient is automatically increased to enhance the immediate response capability; in the comfort mode, the integral term is strengthened to smooth the adjustment process. The actuator dynamically adjusts the electromagnetic valve opening or motor speed according to the control instruction, so that the pedal feedback force is close to the target curve in real time.

[0076] Through the above technical solution, the dynamic optimization of the brake pedal force closed-loop control system is realized, the response speed in heavy load condition and the adjustment accuracy in comfort mode are effectively improved, the overshoot phenomenon in mode switching process is suppressed, and the control stability in complex working conditions is enhanced. The scheme has multi-mode compatibility through the parameter adaptive mechanism, which reduces the hardware modification cost while ensuring the adjustment accuracy.

[0077] In some embodiments, step S5 specifically includes: through brake distance deviation and speed adjustment error evaluation, the target deviation is controlled within ±5%, thereby improving the operation accuracy, then through the pressure distribution sensor installed at the pedal to analyze the uniformity of the palm force, combined with the physiological sensor worn by the operator to monitor the operation fatigue, the comfort score is ≥85 points, thereby improving the operation comfort, in the energy saving mode, the brake energy recovery rate is counted, and the safety indexes such as emergency braking times and coasting distance are recorded to ensure zero safety accidents, then data storage and analysis are performed, in this process, the central processing unit stores the complete data of each brake to form a historical database, through the edge computing module, data cleaning and feature mining are performed regularly to identify typical operation habits and high-frequency working conditions, finally, model and mode iteration are performed, the machine learning model is incrementally trained based on historical data every month, the algorithm weight is optimized; the mode library parameters are updated according to the high-frequency working condition characteristics; the subjective comfort parameters of the feedback curve are adjusted combined with the operator feedback to realize the continuous optimization of system performance.

[0078] The data storage and analysis module refers to the structured storage of complete data of each braking by the central processing unit, and the localized processing by the edge computing module. Specifically, a distributed storage architecture can be used to realize real-time recording of high-frequency data, combined with a data cleaning algorithm to remove noise interference, and a clustering analysis to identify typical operating habits. This module provides a high-quality data basis for system iteration, ensuring the effectiveness of feature mining. The model and pattern iteration mechanism refers to incremental training of machine learning models based on historical data, and updating of pattern library parameters according to high-frequency operating condition features. Specifically, transfer learning technology can be used to retain existing knowledge, optimize model weights through new data, and adjust subjective comfort parameters based on operator feedback. This mechanism enables the system to dynamically adapt to changes in operating conditions and continuously optimize human-machine interaction performance.

[0079] Specifically, the technical solution realizes the continuous evolution of the braking system by building a closed-loop optimization system. First, the evaluation module collects parameters such as braking distance deviation and speed adjustment error in real time, analyzes the uniformity of foot pressure distribution combined with pressure distribution sensors, and monitors operator fatigue through physiological sensors, forming multi-dimensional performance evaluation indicators. Second, the data storage module records the operating characteristics, operating condition parameters, and adjustment results of each braking, uses edge computing for localized cleaning and feature mining, and identifies high-frequency operating conditions and typical operating modes. Finally, the iteration module performs incremental training of machine learning models based on the analysis results, optimizes algorithm weights to improve prediction accuracy, and expands sub-patterns in the pattern library according to high-frequency operating conditions, and adjusts feedback curve parameters based on operator subjective feedback. The three work together to form a closed-loop link of "evaluation-analysis-update", enabling the system to dynamically track changes in operating conditions and operator habits, and continuously improve human-machine collaboration efficiency.

[0080] In some embodiments, the edge computing module can use time series analysis algorithms to identify periodic operating characteristics, such as the time interval of consecutive point braking operations. For high-frequency complex operating conditions, a transition sub-pattern can be created between existing patterns, such as for frequently occurring gentle slope light load conditions, a parameter combination between the standard pattern and the ramp pattern is generated through weighted calculation. During incremental training of the model, knowledge distillation technology can be used to compress the model size to ensure real-time operation capability on the vehicle-mounted processing unit.

[0081] Through the above technical solution, the present application effectively solves the problem of human-machine collaboration efficiency decay caused by the inability of traditional systems to adapt to changes in operating conditions. This solution ensures that the braking system maintains optimal performance over the long term through quantitative evaluation and data-driven iteration mechanisms. Specifically, the control accuracy remains consistently within the target range, the operator's fatigue is significantly reduced, and the energy recovery efficiency and safety indicators are systematically improved. The system can adapt to the habits and preferences of different operators and maintain stable braking performance in complex operating conditions, achieving continuous optimization of human-machine collaboration efficiency.

[0082] In some embodiments, in the continuous evolution of the model and mode iterative implementation system, the machine learning model is incrementally trained once a month: based on the newly added multiple sets of data, the transfer learning strategy is adopted, the training time is <2 hours, the normal operation is avoided, the trained model needs to pass the offline test, and the accuracy rate needs to be improved by ≥3% before it is deployed to the forklift; the mode library parameters are updated once a quarter: for high-frequency working conditions, transition sub-modes are added between existing modes; subjective parameters are adjusted according to operator feedback; the iterative effect is verified by A / B testing: randomly select 10 forklifts to install the new version of the model, and compare them with 10 forklifts that have not been updated for 30 days. If the comfort score of the new version is improved by ≥5 points and the braking distance deviation is reduced by ≥2%, it is fully promoted.

[0083] Among them, the incremental training refers to fine-tuning the existing model based on the newly added operation data, which can be implemented by adopting the transfer learning strategy to reserve the historical weights and only update the parameters related to the new working conditions. By reserving most of the historical parameters, the training time is shortened, and at the same time the adaptability of the model to new working conditions is ensured. The transfer learning strategy refers to fixing part of the network layer parameters of the trained model and only adjusting the network layer related to the new data features. Specifically, it can be realized by freezing the base layer and fine-tuning the top layer parameters, effectively balancing the stability and adaptability of the model. The transition sub-mode refers to inserting intermediate transition parameter sets between existing base modes. Specifically, it can use linear interpolation or nonlinear fitting methods to generate intermediate parameter curves, making the mode switching process smoother and more natural. A / B testing verification refers to comparing the updated version with the original version under the same working conditions. Specifically, it can use random grouping and double-blind testing methods to evaluate the iterative effect through objective indicators and ensure the actual performance improvement of the updated version.

[0084] Specifically, when incrementally training the machine learning model from the forklift operation data every month, the transfer learning strategy preserves the original model's ability to recognize general features and only updates the network layer parameters related to specific new working conditions, ensuring the model's generalization ability while shortening the training period. When analyzing high-frequency working condition features every quarter, identify frequently occurring composite working condition combinations and insert transition sub-mode parameter sets between adjacent modes to generate continuous mode switching curves through parameter interpolation. After collecting operator subjective evaluation data, store the comfort preference parameters by age and operation habit, and dynamically match personalized parameter settings when updating the mode library. After completing the model and mode library updates, randomly select some forklifts to install the new version of the system and compare them with the non-updated vehicles in the same working environment. Verify the iterative effect through braking distance deviation and comfort score indicators, and confirm the performance improvement meets the standards before executing full deployment.

[0085] By the technical solution, the application can realize dynamic adaptation of the brake system control model and working condition changes, avoid algorithm performance degradation caused by long-term use, enhance operation continuity under complex working conditions through smooth switching of the transition sub-mode, improve the use experience of different operators in combination with personalized parameter adjustment, ensure the effectiveness and safety of each iteration update by using the A / B test mechanism, and form a virtuous cycle of system self-optimization.

[0086] The application further provides a central processing unit configured to execute a program of the counterbalanced forklift pedal brake balancing method according to any one of the above embodiments.

[0087] The application further provides a computer readable storage medium having a computer program / instruction stored thereon, the computer program / instruction realizing the steps of the counterbalanced forklift pedal brake balancing method according to any one of the above embodiments when executed by a processor.

[0088] The application further provides a counterbalanced forklift adopting the central processing unit according to any one of the above embodiments.

[0089] The application further provides a computer program product comprising a computer program / instruction, the computer program / instruction realizing the steps of the counterbalanced forklift pedal brake balancing method according to any one of the above embodiments when executed by a processor.

[0090] Other configurations and operations of the counterbalanced forklift pedal brake balancing method according to the embodiments of the application are known to those skilled in the art, and will not be described in detail here.

[0091] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples without contradiction.

[0092] Although the embodiments of the present application have been shown and described, those skilled in the art can understand that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and purposes of the present application, and the scope of the present application is defined by the claims and their equivalents.

Claims

1. A counterbalanced forklift truck pedal braking balancing method, characterized by, The method comprises the following steps: S1, real-time acquisition and preprocessing of multi-dimensional dynamic data: collecting operator pedal operation characteristics, forklift running state and environmental data, and preprocessing the collected raw data by using a central processing unit; S2, intelligent matching and dynamic switching of braking mode: based on the preprocessed real-time data, combining a preset mode library and intelligent decision logic, precise matching and smooth switching of the braking mode are realized; S3, dynamic optimization calculation of pedal force feedback characteristics based on intelligent algorithm: the central processing unit is used to run an intelligent algorithm that integrates multi-dimensional data to calculate the optimal pedal force feedback characteristic parameters that adapt to the current working condition and operation intention in real time; S4, execution mechanism response and real-time adjustment of pedal force: based on the optimal feedback characteristic parameters output by S3, the execution mechanism of the intelligent brake pedal force adjustment system is used to realize precise and dynamic adjustment of the pedal force; S5, effect evaluation and adaptive iterative optimization: the adjustment effect is evaluated in multiple dimensions, and the algorithm and mode library are iteratively updated based on long-term running data to continuously improve the human-machine cooperation performance; Step S3 specifically comprises: Algorithm model construction: a hybrid algorithm architecture of "rule base + machine learning" is adopted, wherein the rule base embeds the physical law of forklift braking, and the machine learning model is trained by historical operation data and has self-learning ability. The input layer is the multi-dimensional feature parameters preprocessed in step S1, and the output layer is the damping force coefficient, the rebound force peak value, and the feedback curve slope multiple feedback characteristic parameters; Dynamic optimization calculation logic: real-time intention recognition: the operator's intention is judged by the change rate of the stepping speed and force, and the algorithm adjusts the feedback response speed accordingly; working condition adaptive correction: the reference feedback curve is dynamically corrected in combination with the forklift state; energy recovery coordination: in the energy-saving mode, the algorithm synchronously correlates the parameters of the braking energy recovery system. When calculating the pedal force feedback, the damping characteristics with the optimal recovery efficiency are preferentially matched to balance the braking effect and energy recovery; output parameter verification: the feedback characteristic parameters calculated need to pass through a safety verification module to ensure that the parameters are within the bearing range of the forklift mechanical structure, and to avoid pedal failure caused by excessive adjustment.

2. The counterbalance forklift pedal braking balancing method according to claim 1, wherein The operator pedal operation characteristic acquisition includes: real-time acquisition of the stepping force by using a pedal force sensor installed at the brake pedal, recording of the pedal displacement and change rate by using a displacement sensor, capturing of the stepping frequency by using a high-frequency sampling module, and obtaining of the foot palm contact position by using a pressure distribution sensor at the pedal to assist in judging the operation intention; The forklift running state and environmental data acquisition includes: collecting key parameters by using a vehicle state sensor group, including obtaining real-time load and load distribution by using a load sensor, collecting driving speed and acceleration by using a speed sensor, detecting road slope by using an inclination sensor, recording steering state by using a steering angle sensor, and synchronously uploading remaining power and motor output power by using a battery management system; in addition, an environmental sensor identifies the road condition to provide environmental parameters for subsequent mode matching. The data preprocessing includes filtering, normalization and feature extraction of the original data collected by the central processing unit to form a structured real-time data matrix, thereby providing high-quality input for subsequent steps.

3. The counterbalanced fork truck pedal brake balancing method of claim 2 wherein, The pressure distribution sensor adopts a flexible thin film design, covering the entire pedal contact surface, and outputs a real-time pressure distribution heat map of the foot contact area through a pressure sensing point array.

4. The counterbalanced fork truck pedal brake balancing method of claim 2 wherein, The Kalman filter algorithm is used for dynamic noise reduction in data preprocessing, the state equation of each sensor is established to predict the next moment data in real time and compare it with the actual collected value, and the error is controlled within 5%. Secondly, the min-max normalization method is used to map different dimensional parameters to the 0-1 interval, eliminating the influence of dimensional differences on subsequent algorithms.

5. The counterbalanced fork truck pedal brake balancing method of claim 1 wherein, The step S2 specifically comprises: Mode library construction and parameter presetting: presetting a basic mode library containing standard mode, overload mode, ramp mode, energy-saving mode and comfort mode, each mode corresponding to differentiated braking characteristic parameters, and storing the typical pedal force feedback curve under each mode as an initial reference; Dual-path mode selection mechanism: manual selection: the operator directly selects the target mode through the forklift human-machine interface, and the system responds in real time and loads the corresponding parameters, which is suitable for scenarios where the operator has clear working condition requirements; intelligent recommendation: the central processing unit performs multi-dimensional evaluation based on real-time data; combined with the operator's historical selection preferences, the recommended mode and the recommended reason are displayed on the pop-up window of the human-machine interface, and the operator can confirm or select other modes with one key; Mode smooth switching control: when the mode is switched, the system gradually adjusts through the actuator to avoid sudden changes in pedal force feedback that may cause operator errors, and records the vehicle state at the switching time to ensure the continuity of the braking effect during the switching process.

6. The counterbalanced fork truck pedal brake balancing method of claim 5 wherein, The forklift human-machine interface adopts a "physical button + touch screen" dual-entry design: a mode switching lever is arranged on the left side of the steering wheel, and the mode icon is displayed on the home page of the touch screen, and the operation response delay is less than 50ms, meeting the needs of the operator's active decision-making.

7. The counterbalanced fork truck pedal brake balancing method of claim 1 wherein, The historical operation data includes multiple samples, each sample containing "input features + output parameters + effect labels", and the 5-fold cross-validation is used to ensure the model generalization ability. The model is designed to be lightweight to ensure that the single calculation time on the central processing unit is less than 10ms.

8. The counterbalanced fork truck pedal brake balancing method of claim 1 wherein, In the working condition adaptive correction, the algorithm dynamically adjusts the reference curve for different working conditions: on wet and slippery roads, the damping force peak is reduced by 10%, and the sensitivity of the rebound force is increased; in the partial load working condition, the left and right wheel braking force distribution algorithm is used to make the pedal left damping force 10% higher than the right side; in low-speed operation, the damping force is weakened to improve the fine-tuning accuracy. In energy recovery cooperation, the algorithm in the energy-saving mode communicates with the battery management system in real time: when the BMS feedback SOC is less than 30%, the braking time is preferentially extended to increase the energy recovery amount by 15%; when the SOC is greater than or equal to 90%, the standard damping force is restored, and the operator is prompted to reduce the braking frequency through the rebound force curve.

9. The counterbalanced fork truck pedal brake balancing method of claim 1 wherein, The step S4 specifically comprises: Coordinated control of actuators: electronically controlled adjustable damper: according to the damping force coefficient parameter, the damping oil flow is adjusted through the proportional electromagnetic valve to change the resistance in the pedal pedaling process in real time; servo motor and transmission mechanism: receiving the rebound force parameter, driving the pedal back through the screw nut structure, outputting the rebound force matched with displacement and speed; auxiliary buffer device: setting an electrically controlled buffer at the end of the pedal stroke, when the pedal approaches the maximum displacement, avoiding rigid impact by reducing the damping rate, improving the operation comfort; closed-loop feedback regulation: displacement sensor and pedal force sensor monitor the actual pedal state in real time, and feedback the data to the central processing unit, and dynamically correct the actuator output through the PID control algorithm to ensure that the deviation between the feedback characteristics and the calculated value is controlled within ±3%; emergency regulation in extreme conditions: when the sensor detects an abnormal state, the system immediately triggers the emergency mode, provides basic rebound force through the standby spring mechanism, and alarms on the human-machine interface at the same time, while retaining the mechanical braking function to ensure operation safety.

10. The counterbalanced fork truck pedal brake balancing method of claim 9 wherein, In the real-time cycle of the closed-loop feedback regulation mechanism "calculation-execution-correction", the displacement sensor and the pedal force sensor send actual values to the central processing unit every 10ms, the error value is output by the deviation calculation module, and the PID controller dynamically adjusts the output signal according to the error: the proportional term quickly responds to the current error, the integral term eliminates cumulative error, and the differential term suppresses overshoot; the parameters of closed-loop control can be dynamically adjusted according to the mode: increase the proportional term in heavy load mode to improve response speed, and increase the integral term in comfort mode to reduce fluctuations.

11. The counterbalanced fork truck pedal brake balancing method of claim 1 wherein, Step S5 specifically includes: Through the evaluation of braking distance deviation and speed adjustment error, the target deviation is controlled within ±5%, thereby improving the control accuracy, then the pressure distribution sensor installed on the pedal analyzes the uniformity of the foot palm force, combined with the physiological sensor worn by the operator to monitor the operation fatigue, the comfort score is ≥85 points, thereby improving the operation comfort, in the energy saving mode, the braking energy recovery rate is counted, and the emergency braking frequency safety index and the coasting distance safety index are recorded to ensure zero safety accidents, then data storage and analysis are performed, in this process, the central processing unit stores the complete data of each braking to form a historical database, through the edge computing module, data cleaning and feature mining are performed regularly to identify typical operation habits and high-frequency working conditions, finally, model and mode iteration are performed, the machine learning model is incrementally trained based on historical data every month, the algorithm weight is optimized; update the mode library parameters according to the high-frequency working condition characteristics; adjust the subjective comfort parameters of the feedback curve combined with the operator feedback to realize continuous optimization of system performance.

12. The counterbalanced fork truck pedal brake balancing method of claim 11 wherein, In the process of system continuous evolution through model and mode iteration, the machine learning model is incrementally trained once a month: based on the newly added multiple sets of data, the transfer learning strategy is adopted, the training time is <2 hours to avoid affecting normal operation, and the trained model needs to pass offline test to ensure that the accuracy is improved by ≥3% before being deployed to the forklift; The mode library parameters are updated every quarter: for high-frequency working conditions, a transition sub-mode is added between existing modes; subjective parameters are adjusted according to operator feedback; the iterative effect is verified through A / B testing: 10 forklifts are randomly selected to install the new version of the model, and compared with 10 forklifts that have not been updated for 30 days; if the comfort score of the new version is improved by ≥5 points and the braking distance deviation is reduced by ≥2%, the new version is promoted.

13. A central processing unit, characterized by The central processing unit is configured to execute a program for the counterbalanced forklift pedal brake balancing method of any of claims 1-12.

14. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, implement the steps of the counterbalanced forklift pedal brake balancing method of any of claims 1-12.

15. A counterbalanced forklift, characterized in that, The central processing unit of claim 13 is used.

16. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions, when executed by the processor, implement the steps of the counterbalanced forklift pedal brake balancing method of any of claims 1-12.

Images