A hoist anti-settling adaptive control method based on load fluctuation perception

Abstract

The application relates to a load fluctuation perception-based anti-settling adaptive control method for an elevator, which comprises the following steps: S1, real-time perception of an elevator operation state, and acquisition of multi-source operation data; S2, based on the dynamic load data and longitudinal vibration acceleration data, a preset load fluctuation prediction model is used to identify and predict a load fluctuation mode and a fluctuation trend; S3, data is input into a multi-modal adaptive controller; S4, the multi-modal adaptive controller dynamically calculates and outputs an adaptive brake compensation signal according to input information; S5, a hierarchical brake actuator of the elevator is controlled to act according to the brake compensation signal; and S6, through a preset feedback optimization loop, parameters of the load fluctuation prediction model and the multi-modal adaptive controller are optimized online based on actual settling effects after braking. The application significantly improves the stop position accuracy and operation stability of the elevator under dynamic working conditions.

Description

Technical Field

[0001] This application relates to the field of hoist control, and in particular to an adaptive control method for preventing settlement of hoists based on load fluctuation sensing. Background Technology

[0002] As a key vertical conveying device in systems such as automated storage and retrieval systems (AS / RS), the positioning accuracy and operational stability of hoists are of paramount importance. Currently, anti-settlement control of hoists mostly adopts control strategies based on fixed thresholds, such as preset load limits, triggering braking when the load exceeds the threshold. These methods can prevent settlement caused by static overload to a certain extent. In addition, there are some optimized control schemes in the existing technology, such as the adaptive parameter adjustment hoist control method proposed in patent application CN119389893A, which dynamically optimizes the inverter operating parameters through an intelligent model, aiming to improve operating efficiency and reduce inverter losses.

[0003] However, in actual working conditions, the load fluctuations caused by the shuttle's dynamic entry and exit from the hoist platform are the primary factor leading to platform settlement and affecting positioning accuracy. Neither traditional fixed-threshold anti-settlement control nor efficiency-optimized control methods, such as CN119389893A, address the technical problem of accurately maintaining the platform's posture under dynamic load fluctuations. Specifically, existing solutions lack real-time perception and prediction capabilities for load fluctuation patterns. Their control response acts on the drive system (inverter / motor) rather than directly targeting the braking system to suppress platform settlement. Therefore, they cannot solve the problems of settlement lag or over-braking caused by dynamic impacts.

[0004] Therefore, this invention proposes an adaptive control method for preventing settlement of hoists based on load fluctuation sensing. Summary of the Invention

[0005] To address the aforementioned issues, this application provides an adaptive control method for preventing settlement of a hoist based on load fluctuation sensing, employing the following technical solution:

[0006] An adaptive control method for preventing settlement of a hoist based on load fluctuation sensing includes the following steps:

[0007] S1. Real-time sensing of the hoist's operating status and acquisition of multi-source operating data; the multi-source operating data includes at least dynamic load data collected by a pressure sensor array arranged on the hoist platform, real-time platform height data collected by a displacement sensor, and longitudinal vibration acceleration data collected by a vibration sensor.

[0008] S2. Based on the dynamic load data and longitudinal vibration acceleration data, identify and predict the load fluctuation pattern and fluctuation trend through a pre-set load fluctuation prediction model; the load fluctuation prediction model integrates a vibration prediction mechanism to predict the load fluctuation peak in advance; the load fluctuation prediction model is an adaptive BP neural network model, which is trained through historical working condition data.

[0009] S3. Input the identified load fluctuation pattern, predicted fluctuation trend, real-time platform height data, and longitudinal vibration acceleration data into the multimodal adaptive controller.

[0010] S4. The multimodal adaptive controller dynamically calculates and outputs an adaptive braking compensation signal based on the input information; wherein, the multimodal adaptive controller adopts a fuzzy PID control algorithm and can switch between multiple built-in sets of control parameters according to the load fluctuation mode;

[0011] S5. Based on the braking compensation signal, control the staged braking actuator of the hoist to operate, so as to achieve active control of the platform height;

[0012] S6. Based on the actual settlement effect after braking, the parameters of the load fluctuation prediction model and the multimodal adaptive controller are optimized online through a pre-set feedback optimization loop.

[0013] Preferably, in step S1, the multi-source operating data also includes the current data, operating speed and acceleration data of the hoist main motor, and the attitude data of the shuttle entering and exiting the hoist captured by the vision sensor; and the collected multi-source operating data is fused in real time using Kalman filtering and attention mechanism, and after removing abnormal signals, it is transmitted to the load fluctuation prediction model.

[0014] Preferably, the multiple sets of control parameters are multiple sets of PID parameters; the fuzzy PID control algorithm takes the platform height deviation and the platform height deviation change rate as input, constructs a seven-level fuzzy subset including negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and establishes a dynamic adjustment rule base for three parameters Kp, Ki, and Kd to achieve phased control: increasing the Kp value in the early stage of fluctuation to improve the response speed, decreasing the Kp value in the middle of fluctuation to prevent overshoot, and increasing the Ki value in the later stage of fluctuation to eliminate steady-state error.

[0015] Preferably, in step S4, the fuzzy PID control algorithm adopts a feedforward-feedback dual closed-loop architecture: the feedforward loop outputs a feedforward quantity for pre-braking based on the fluctuation trend output by the load fluctuation prediction model; the feedback loop performs PID calculations based on the real-time height data of the platform to provide a feedback quantity for real-time correction of braking deviation; the feedforward quantity and the feedback quantity are superimposed to form the braking compensation signal.

[0016] Preferably, when the longitudinal vibration acceleration data exceeds 1 m / s², the fuzzy PID control algorithm dynamically increases the Kd value and controls the graded braking actuator to reduce the pressure loading rate, so as to suppress secondary settlement caused by longitudinal vibration.

[0017] Preferably, in step S3, a digital twin model of the hoist is also introduced; the digital twin model simultaneously receives the multi-source operating data and the fluctuation trend output by the load fluctuation prediction model as input, performs real-time simulation through the built-in platform dynamics model, and outputs the platform pose prediction at future time.

[0018] The multimodal adaptive controller also uses the pose prediction and the fluctuation trend output by the load fluctuation prediction model as the basis for calculating the feedforward amount; when it is determined based on the platform pose prediction that the settlement will exceed the allowable range, a braking compensation signal containing the feedforward amount is generated and output before the platform settlement occurs.

[0019] Preferably, the graded braking actuator is used to perform braking actions of different intensities according to the braking compensation signal, including micro-modulation for maintaining accuracy, holding braking for stable load bearing, and emergency braking for extreme situations.

[0020] Preferably, a safety redundancy mechanism is also included: when the longitudinal vibration acceleration exceeds a preset safety threshold, the equipment temperature is abnormal, or the control command overflows, an emergency pre-braking procedure is triggered; and hard limits are set for the key control parameters in the fuzzy PID control algorithm; at the same time, fixed threshold control is reserved as a backup scheme and automatically switched to be activated when the main adaptive control logic fails.

[0021] Preferably, the load fluctuation mode includes at least: the impact fluctuation mode caused by the shuttle entering the platform, the front-end load mode caused by the shuttle being partially outside the platform, and the unloading fluctuation mode caused by the shuttle leaving the platform.

[0022] Preferably, the step of switching between multiple built-in control parameter sets according to the load fluctuation mode specifically involves: calling the corresponding PID parameter combination in the preset mode-parameter mapping table according to the identified load fluctuation mode; wherein, for the impact fluctuation mode, a parameter combination with a large proportional gain and a large derivative time is called to quickly suppress platform rise; for the front-end load-bearing mode, a parameter combination with a specially customized power distribution strategy is called to compensate for the forward tilting torque; and for the unloading fluctuation mode, a parameter combination with a large integral time and a large derivative time is called to suppress platform sinking.

[0023] In summary, this application includes at least one of the following beneficial technical effects:

[0024] 1. This application focuses on the accurate maintenance of platform posture under dynamic load fluctuations. It collects key data such as load, height, and vibration through multi-source sensor fusion, and uses a load fluctuation prediction model to achieve real-time identification and peak prediction of load fluctuation patterns. Then, a multi-modal adaptive controller, based on fuzzy PID algorithm and patterned parameter switching, directly outputs braking compensation signals to drive the staged braking actuator. It can not only sense and predict the fluctuation patterns and trends of dynamic loads in advance, but also output accurate braking compensation signals before platform settlement occurs through feedforward-feedback composite control and parameter adaptive switching. This effectively solves the problem of settlement overshoot or false braking caused by response lag in traditional fixed threshold control, and significantly improves the stopping accuracy and running stability of the hoist under dynamic working conditions.

[0025] 2. The feedforward-feedback dual closed-loop architecture constructed in this application forms a complementary and synergistic control through the organic combination of proactive intervention and post-event correction: the feedforward link achieves advanced adjustment based on prediction, fundamentally offsetting major disturbances and solving the inherent lag of traditional feedback control; the feedback link performs precise fine-tuning based on real-time measurement, effectively suppressing model prediction errors and unknown disturbances. The combination of these two aspects ensures that the braking compensation signal possesses both foresight and accuracy, thereby significantly improving the system response speed while completely eliminating steady-state error, achieving millimeter-level high precision and high dynamic stability of the hoisting platform.

[0026] 3. Based on the introduced digital twin model, and through the deep integration of data-driven approaches and physical mechanisms, feedforward control is upgraded from trend prediction to state prediction. Based on multi-source real-time data and a precise platform dynamics model, it can predict the platform's specific pose changes within 0.2 to 0.4 seconds, enabling the controller to output a precisely matched compensation signal before the actual physical settlement occurs. This achieves a paradigm shift from perception-response to prediction-proactive control, fundamentally solving the control lag problem caused by system mechanical inertia and detection delays. It enables precise intervention in the early stages of platform settlement, thereby elevating the accuracy and reliability of anti-settlement control to a new level. Attached Figure Description

[0027] Figure 1 This is a flowchart of a method for adaptive control of a hoist anti-settlement based on load fluctuation sensing in an embodiment of this application;

[0028] Figure 2 This is a flowchart illustrating the specific process of the feedback optimization loop in the embodiments of this application. Detailed Implementation

[0029] The following is in conjunction with the appendix Figure 1 and attached Figure 2This application will be described in further detail.

[0030] This application discloses an adaptive control method for preventing settlement of a hoist based on load fluctuation sensing. (Refer to...) Figure 1 An adaptive control method for preventing settlement of a hoist based on load fluctuation sensing includes the following steps:

[0031] S1. Acquire multi-source operating data: Real-time sensing of the hoist's operating status and acquisition of multi-source operating data; the multi-source operating data includes at least dynamic load data collected by a pressure sensor array arranged on the hoist platform, real-time platform height data collected by a displacement sensor, and longitudinal vibration acceleration data collected by a vibration sensor.

[0032] S2. Pattern Recognition and Trend Prediction: Based on the dynamic load data and longitudinal vibration acceleration data, the load fluctuation pattern and trend are identified and predicted through a pre-set load fluctuation prediction model. The load fluctuation prediction model integrates a vibration prediction mechanism to predict the peak load fluctuation in advance. The load fluctuation prediction model is an adaptive BP neural network model, which is trained through historical working condition data.

[0033] S3. Data Input: The identified load fluctuation pattern, predicted fluctuation trend, real-time platform height data, and longitudinal vibration acceleration data are input to the multimodal adaptive controller.

[0034] S4. Output braking compensation signal: The multimodal adaptive controller dynamically calculates and outputs an adaptive braking compensation signal based on the input information; wherein, the multimodal adaptive controller adopts a fuzzy PID control algorithm and can switch between multiple sets of built-in control parameters according to the load fluctuation mode;

[0035] S5. Active control of platform height: Based on the braking compensation signal, control the action of the staged braking actuator of the hoist to achieve active control of the platform height;

[0036] S6. Feedback Optimization: Through a pre-set feedback optimization loop, the parameters of the load fluctuation prediction model and the multimodal adaptive controller are optimized online based on the actual settlement effect after braking. Through the above steps, with the core objective of accurately maintaining the platform's posture under dynamic load fluctuations, key data such as load, height, and vibration are collected through multi-source sensor fusion. The load fluctuation prediction model enables real-time identification and peak prediction of load fluctuation patterns. Then, the multimodal adaptive controller, based on a fuzzy PID algorithm and patterned parameter switching, directly outputs braking compensation signals to drive the graded braking actuator. This not only allows for early perception and prediction of dynamic load fluctuation patterns and trends, but also, through feedforward-feedback composite control and adaptive parameter switching, outputs precise braking compensation signals before platform settlement occurs. This effectively solves the problem of settlement overshoot or false braking caused by response lag in traditional fixed threshold control, significantly improving the hoist's stopping accuracy and operational stability under dynamic conditions.

[0037] To achieve highly reliable load fluctuation sensing, in step S1, the system collects dynamic load data through a pressure sensor array deployed on the hoist platform, collects real-time platform height data through a displacement sensor, and collects longitudinal vibration acceleration data through a vibration sensor. In addition, the multi-source operating data also includes the hoist main motor's current data, operating speed and acceleration data, and the attitude data (such as the car body tilt angle and entry angle) captured by a vision sensor when the shuttle enters and exits the hoist.

[0038] Crucially, the system uses Kalman filtering and attention mechanisms to fuse the aforementioned multi-source operational data in real time, removing abnormal signals before transmitting it to the load fluctuation prediction model. The specific process is as follows:

[0039] (1) Data preprocessing and Kalman filtering

[0040] The raw data collected by various sensors are timestamped and normalized to the same time reference and numerical units.

[0041] For time-series dynamic data (including current data, operating speed and acceleration data, and longitudinal vibration acceleration data of the hoist main motor), a multivariable Kalman filter is used for real-time filtering. Specifically, a state-space model based on platform dynamics is established, with motor current and operating acceleration as system inputs and vibration acceleration and operating speed as observations. Through a prediction-update loop, high-frequency measurement noise is effectively filtered out, and the system state that cannot be directly measured (such as the instantaneous disturbance force on the platform) is estimated.

[0042] For the dynamic load data collected by the pressure sensor array, a distributed Kalman filter is used. First, each pressure sensor node is filtered independently, and then the total load value and the load centroid coordinates are fused through a consensus algorithm to obtain an optimized estimate.

[0043] (2) Multi-source data fusion based on attention mechanism

[0044] Various types of data processed by Kalman filtering, as well as shuttle attitude data (including vehicle tilt angle and entry angle) captured and calculated by visual sensors, are input into a lightweight attention fusion network.

[0045] The network dynamically calculates an attention weight for each type of data. The weight calculation formula is as follows:

[0046] ;

[0047] in, For the attention weights of the i-th data class, Let i be the feature vector of the i-th class of data. and Learnable parameters.

[0048] The significance of attention weighting lies in the fact that the system can adaptively focus on the most important information sources according to the current operating conditions. For example, at the moment the shuttle just touches the platform, the weights of vibration data and visual attitude data will be automatically increased to quickly capture the impact; during the stable operation phase, the weights of motor current and pressure data will take precedence to accurately assess the load.

[0049] (3) Abnormal signal rejection and output

[0050] Set a dynamic threshold based on statistics (such as the ±3σ principle) to monitor the integrated data stream after weighted fusion.

[0051] When the residual of a certain type of data (the difference between the observed value and the Kalman filter estimate) continuously exceeds the preset threshold, or when its attention weight fluctuates drastically and abnormally, the signal is determined to be abnormal, and a redundant sensor replacement mechanism or an interpolation compensation algorithm based on historical data is activated.

[0052] Finally, the cleaned and integrated multi-source operating data is transmitted in real time to the load fluctuation prediction model, providing it with high-quality and high-reliability input, fundamentally improving the model's prediction accuracy and the system's control stability.

[0053] The beneficial effects of this step are as follows: high-frequency noise is effectively filtered out and key state variables are estimated through Kalman filtering, and adaptive weighting of different data sources is achieved through attention mechanism, thereby providing clean, reliable and high-value information input for downstream prediction models, ensuring the control accuracy and robustness of the system from the source.

[0054] In step S2, the fused dynamic load data and longitudinal vibration acceleration data are input into a pre-trained adaptive BP neural network model (i.e., a load fluctuation prediction model). This model integrates a vibration prediction mechanism, capable of predicting load fluctuation peaks 0.3–0.5 seconds in advance. The model can identify at least three typical load fluctuation patterns:

[0055] Shock wave pattern: triggered the instant the shuttle enters the platform;

[0056] Front-end load-bearing mode: caused by the shuttle car being located outside the platform;

[0057] Unloading fluctuation mode: triggered by the shuttle leaving the platform. To achieve precise control, the system has a pre-set mode-parameter mapping table. When the model identifies the current fluctuation mode, the system calls the corresponding optimal PID parameter combination in the mapping table to prepare for subsequent control. Specifically, it switches between multiple built-in control parameter sets (multiple control parameters are multiple PID parameter sets) according to the load fluctuation mode. Specifically: based on the identified load fluctuation mode, it calls the corresponding PID parameter combination in the pre-set mode-parameter mapping table; for the impact fluctuation mode, it calls a parameter combination with a large proportional gain and a large derivative time to quickly suppress platform rise; for the front-end load-bearing mode, it calls a parameter combination with a specially customized power distribution strategy to compensate for the forward tilting torque; for the unloading fluctuation mode, it calls a parameter combination with a large integral time and a large derivative time to suppress platform sinking. This step, through precise identification of load fluctuation patterns and intelligent matching of pattern-parameter mapping tables, enables tailored control strategies, fundamentally solving the problem that traditional single-parameter control cannot accommodate multiple dynamic operating conditions: rapid response to impact fluctuations, precise compensation for off-center load conditions, and smooth suppression of unloading processes. This not only significantly improves the accuracy of settlement control but also effectively avoids over-control and under-control, providing accurate and efficient parameter support for subsequent braking compensation from the source, ensuring the stable operation of the hoist under complex operating conditions and extending the equipment's service life.

[0058] In step S4, the multimodal adaptive controller employs a fuzzy PID control algorithm and constructs a feedforward-feedback dual-loop architecture: the feedforward loop outputs a feedforward quantity for pre-braking based on the fluctuation trend of the load fluctuation prediction model (active intervention); the feedback loop performs PID calculations based on the platform's real-time height data to provide feedback for real-time correction of braking deviations (post-correction); and the feedforward and feedback quantities are superimposed to form the braking compensation signal. The feedforward-feedback dual-loop architecture constructed in this step, through the organic combination of active intervention and post-correction, forms a complementary and synergistic control: the feedforward loop achieves advanced adjustment based on prediction, fundamentally offsetting major disturbances and solving the inherent lag of traditional feedback control; the feedback loop performs precise fine-tuning based on real-time measurement, effectively suppressing model prediction errors and unknown disturbances. The superposition of these two elements ensures that the braking compensation signal possesses both foresight and accuracy, thereby significantly improving the system response speed while completely eliminating steady-state error, achieving millimeter-level high precision and high dynamic stability for the hoisting platform.

[0059] Furthermore, in a preferred embodiment, in step S3, a digital twin model of the hoist is introduced. This digital twin model simultaneously receives the multi-source operating data and the fluctuation trend output by the load fluctuation prediction model as input. It performs real-time simulation using a built-in platform dynamics model, outputting a predicted platform pose for future moments (0.2–0.4 s). The multimodal adaptive controller also uses this pose prediction and the fluctuation trend output by the load fluctuation prediction model as the basis for calculating the feedforward quantity. When the platform pose prediction indicates that the settlement will exceed the allowable range, a braking compensation signal containing this feedforward quantity is generated and output before the platform settlement occurs, achieving true "prevention before the event." Based on the introduced digital twin model, through the deep integration of data-driven and physical mechanisms, feedforward control is upgraded from trend prediction to state prediction: based on multi-source real-time data and a precise platform dynamics model, it can predict the specific pose changes of the platform within the next 0.2–0.4 seconds, enabling the controller to output a precisely matched compensation signal before the actual physical settlement occurs. This achieves a paradigm shift from perception-response to anticipation-proactive control, fundamentally solving the control lag problem caused by system mechanical inertia and detection delay, enabling precise intervention in the early stages of platform settlement, thereby raising the precision and reliability of anti-settlement control to a whole new level.

[0060] The fuzzy PID control algorithm takes the platform height deviation (e) and the rate of change of platform height deviation (ec) as inputs, constructs a seven-level fuzzy subset including negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and establishes a dynamic adjustment rule base of three parameters Kp, Ki, and Kd to achieve phased control: increasing the Kp value in the early stage of fluctuation to improve the response speed, decreasing the Kp value in the middle stage of fluctuation to prevent overshoot, and increasing the Ki value in the later stage of fluctuation to eliminate steady-state error. In this embodiment, the preferred fluctuation stage is defined as follows: in the early stage of fluctuation, the rate of change of platform height deviation (ec) ≤ 0.1% / ms, increasing the Kp value to improve the response speed; in the middle stage of fluctuation, 0.1% / ms < ec ≤ 0.5% / ms, decreasing the Kp value to prevent overshoot; and in the later stage of fluctuation, ec > 0.5% / ms or after the peak of fluctuation, increasing the Ki value to eliminate steady-state error. By deeply integrating fuzzy logic with classical PID control and introducing a quantization stage criterion based on the rate of change of deviation (ec), dynamic and precise matching between controller parameters and the fluctuating process is achieved: proportional gain is strengthened in the early stages of fluctuation for rapid response, proportional gain is suppressed in the middle stages to prevent overshoot, and integral gain is enhanced in the later stages to eliminate steady-state error. This adaptive adjustment, which follows the evolution of fluctuations, fundamentally solves the contradiction between response speed and control accuracy caused by fixed parameters in traditional PID control. This allows the system to achieve both rapid, overshoot-free tracking and high-precision stability when facing rapidly changing dynamic loads, significantly improving the smoothness and reliability of control.

[0061] Furthermore, when the longitudinal vibration acceleration data exceeds 1 m / s², the fuzzy PID control algorithm dynamically increases the Kd value and controls the staged braking actuator to reduce the pressure loading rate, thereby suppressing secondary settlement caused by longitudinal vibration and controlling the settlement of the hoist platform within ±0.5 mm. This application establishes a linkage mechanism between longitudinal vibration acceleration and differential control, achieving proactive identification and precise suppression of secondary settlement: when increased vibration is detected, the system dynamically enhances the differential action to improve system damping, while actively reducing the braking loading rate to avoid excitation. This dual adjustment strategy effectively breaks the vicious cycle from vibration to settlement, transforming the originally uncontrollable dynamic fluctuations into a manageable steady-state process, thus maintaining an extremely high precision of ±0.5 mm even under complex working conditions, significantly improving the system's anti-interference capability and overall stability.

[0062] In step S5 above, the graded braking actuator is used to perform braking actions of different intensities according to the braking compensation signal, including at least fine-tuning for maintaining precision, holding braking for stable load bearing, and emergency braking for extreme situations. More subdivided braking levels can also be set according to actual usage needs. The advantage of this graded braking actuator is that by finely dividing the braking action into three levels—fine-tuning, holding, and emergency—it achieves on-demand distribution of braking force, fundamentally overcoming the drawbacks of the traditional single-mode braking approach: it can perform millimeter-level precise fine-tuning during smooth operation, provide stable holding during load bearing, and respond quickly in abnormal situations. This functional decoupling design allows the system to achieve an optimal balance between precision, stability, and safety, significantly improving control quality, effectively avoiding shocks and oscillations during braking, and extending equipment lifespan.

[0063] Reference Figure 2 The specific process of the feedback optimization loop in step S6 is as follows:

[0064] A1. Calculate settlement deviation: Collect the actual settlement of the platform after braking in real time using a high-precision displacement sensor, and calculate the settlement deviation between the platform and the target height.

[0065] A2. Triggering Optimization Mechanism: When the settlement deviation continues to exceed the set threshold, the optimization mechanism is triggered, which simultaneously performs the following operations:

[0066] A2a. Reverse adjustment of connection weights: Based on the settlement deviation and the corresponding historical multi-source operation data, the connection weights of the load fluctuation prediction model are adjusted in reverse.

[0067] A2b. Online Rule Correction: Based on the dynamic characteristics of the settlement deviation, the fuzzy rule base and adjustment rules of the proportional gain Kp, integral time Ki, and derivative time Kd of the fuzzy PID control algorithm are corrected online. The advantage of this feedback optimization loop is that by feeding the control effect back to the prediction model and control algorithm in real time, an intelligent system with autonomous learning and continuous evolution capabilities is constructed. It not only corrects the internal weights of the load fluctuation prediction model online to improve its prediction accuracy, but also dynamically optimizes the control rules and parameters of the fuzzy PID based on actual settlement characteristics, thus achieving a leap from one-time debugging to full lifecycle self-optimization. This dual-path collaborative optimization mechanism effectively eliminates the impact of model prediction errors and equipment performance drift on control accuracy, ensuring that the system maintains millimeter-level anti-settlement performance throughout long-term operation, significantly improving the adaptability and robustness of the method.

[0068] To ensure absolute system reliability, a safety redundancy mechanism is also implemented: when the longitudinal vibration acceleration exceeds a preset safety threshold, the equipment temperature is abnormal, or the control command overflows, an emergency pre-braking procedure is triggered; and hard limits are set for key control parameters in the fuzzy PID control algorithm; simultaneously, fixed threshold control is reserved as a backup scheme, which is automatically switched to and activated in the event of a failure in the main adaptive control logic. For example, in this embodiment, when the longitudinal vibration acceleration exceeds 2 m / s², the equipment temperature exceeds 85°C, or the control command exceeds the braking pressure execution range (0.1 MPa to 3 MPa), an emergency pre-braking procedure is triggered; and hard limits of ±50% are set for the key control parameters Kp, Ki, and Kd (based on the default values ​​of the mode-parameter mapping table); simultaneously, fixed threshold control is reserved as a backup scheme, which is automatically switched to and activated in the event of a failure in the main adaptive control logic. By establishing a three-tiered protection system for abnormal states, a fault defense chain is constructed that runs through the entire process of perception, decision-making, and execution: when excessive vibration, abnormal temperature, or command overflow is detected, the system immediately activates an emergency pre-braking procedure to achieve active protection; by setting hard limits for key parameters, extreme divergence of control commands is effectively prevented; and the setting of a fixed threshold backup scheme provides a reliable last line of defense in the event of a main control system failure. This layered and progressive safety design enables the system to maintain basic safety functions even under extreme operating conditions or component failures, fundamentally solving the reliability bottleneck of intelligent control systems in industrial applications and ensuring the long-term safety and stability of the hoist.

[0069] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still combine, add, delete, or otherwise adjust the features of the various embodiments of the present invention according to the circumstances without conflict or creative effort, thereby obtaining different technical solutions that do not fundamentally depart from the concept of the present invention. These technical solutions also fall within the scope of protection of the present invention.

Claims

1. A method for adaptive control of hoist anti-settlement based on load fluctuation sensing, characterized in that, Includes the following steps: S1. Real-time sensing of the hoist's operating status and acquisition of multi-source operating data; the multi-source operating data includes at least dynamic load data collected by a pressure sensor array arranged on the hoist platform, real-time platform height data collected by a displacement sensor, and longitudinal vibration acceleration data collected by a vibration sensor. S2. Based on the dynamic load data and longitudinal vibration acceleration data, identify and predict the load fluctuation pattern and fluctuation trend through a pre-set load fluctuation prediction model; the load fluctuation prediction model integrates a vibration prediction mechanism to predict the load fluctuation peak in advance; the load fluctuation prediction model is an adaptive BP neural network model, which is trained through historical working condition data. S3. Input the identified load fluctuation pattern, predicted fluctuation trend, real-time platform height data, and longitudinal vibration acceleration data into the multimodal adaptive controller. S4. The multimodal adaptive controller dynamically calculates and outputs an adaptive braking compensation signal based on the input information; wherein, the multimodal adaptive controller adopts a fuzzy PID control algorithm and can switch between multiple built-in sets of control parameters according to the load fluctuation mode; S5. Based on the braking compensation signal, control the staged braking actuator of the hoist to operate, so as to achieve active control of the platform height; S6. Based on the actual settlement effect after braking, the parameters of the load fluctuation prediction model and the multimodal adaptive controller are optimized online through a pre-set feedback optimization loop.

2. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that: In step S1, the multi-source operating data also includes the current data, operating speed and acceleration data of the hoist main motor, and the attitude data of the shuttle entering and exiting the hoist captured by the vision sensor; and the collected multi-source operating data is fused in real time using Kalman filtering and attention mechanism, and after removing abnormal signals, it is transmitted to the load fluctuation prediction model.

3. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that: The multiple sets of control parameters are multiple sets of PID parameters; the fuzzy PID control algorithm takes the platform height deviation and the platform height deviation change rate as input, constructs a seven-level fuzzy subset including negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and establishes a dynamic adjustment rule base for three parameters Kp, Ki, and Kd to achieve phased control: increase the Kp value in the early stage of fluctuation to improve the response speed, decrease the Kp value in the middle of fluctuation to prevent overshoot, and increase the Ki value in the later stage of fluctuation to eliminate steady-state error.

4. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 3, characterized in that: In step S4, the fuzzy PID control algorithm adopts a feedforward-feedback dual closed-loop architecture: the feedforward loop outputs a feedforward quantity for pre-braking based on the fluctuation trend output by the load fluctuation prediction model; the feedback loop performs PID calculations based on the real-time height data of the platform to provide a feedback quantity for real-time correction of braking deviation; the feedforward quantity and the feedback quantity are superimposed to form the braking compensation signal.

5. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 4, characterized in that: When the longitudinal vibration acceleration data exceeds 1 m / s², the fuzzy PID control algorithm dynamically increases the Kd value and controls the graded braking actuator to reduce the pressure loading rate in order to suppress secondary settlement caused by longitudinal vibration.

6. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that: In step S3, a digital twin model of the hoist is also introduced; the digital twin model simultaneously receives the multi-source operating data and the fluctuation trend output by the load fluctuation prediction model as input, performs real-time simulation through the built-in platform dynamics model, and outputs the platform pose prediction at future time. The multimodal adaptive controller also uses the pose prediction and the fluctuation trend output by the load fluctuation prediction model as the basis for calculating the feedforward amount; when it is determined based on the platform pose prediction that the settlement will exceed the allowable range, a braking compensation signal containing the feedforward amount is generated and output before the platform settlement occurs.

7. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that: The graded braking actuator is used to perform braking actions of different intensities according to the braking compensation signal, including micro-modulation for maintaining accuracy, holding braking for stable load bearing, and emergency braking for extreme situations.

8. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that, It also includes a safety redundancy mechanism: when the longitudinal vibration acceleration is detected to exceed the preset safety threshold, the equipment temperature is abnormal, or the control command overflows, an emergency pre-braking procedure is triggered; and hard limits are set for the key control parameters in the fuzzy PID control algorithm; at the same time, fixed threshold control is reserved as a backup scheme and automatically switched to be activated when the main adaptive control logic fails.

9. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 1, characterized in that, The load fluctuation modes include at least: the impact fluctuation mode caused by the shuttle entering the platform, the front-end load mode caused by the shuttle being partially outside the platform, and the unloading fluctuation mode caused by the shuttle leaving the platform.

10. The adaptive control method for preventing settlement of a hoist based on load fluctuation sensing according to claim 9, characterized in that, The switching of multiple built-in control parameter sets according to the load fluctuation mode specifically involves: calling the corresponding PID parameter combination in the preset mode-parameter mapping table according to the identified load fluctuation mode; wherein, for the impact fluctuation mode, a parameter combination with a large proportional gain and a large derivative time is called to quickly suppress platform rise; for the front-end load-bearing mode, a parameter combination with a specially customized power distribution strategy is called to compensate for the forward tilting torque; and for the unloading fluctuation mode, a parameter combination with a large integral time and a large derivative time is called to suppress platform sinking.

Images