Method and device for dynamic protection of the stability boundary of a lifting operation of a rear-discharge semi-trailer

Abstract

This invention relates to the field of joint control technology, and in particular to a method and device for dynamic protection of the stability boundary during the lifting operation of a rear-unloading semi-trailer. The method includes: calculating the stability boundary at the current lifting angle and obtaining an initial monitoring dataset; smoothing the angular velocity data, extracting disturbance feature parameters, generating an angular velocity fluctuation sequence, and constructing an attitude disturbance vector by combining the distance change between state points; simulating random disturbances based on the attitude disturbance vector to generate multiple sets of future short-term attitude change trajectories, generating a path probability distribution and an average probability value; and generating intervention commands based on the probability distribution shape, average probability value, and current operating parameters to adjust the lifting operation status. This invention solves the problem of difficulty in timely identifying and intervening in dynamic instability risks during the lifting process of a rear-unloading semi-trailer, and achieves dynamic assessment of the stability boundary of the lifting operation, early prediction of overturning risks, and proactive safety control of the operation status.

Description

Technical Field

[0001] This invention relates to the field of joint control technology, and in particular to a method and device for dynamic protection of stability boundaries in the lifting operation of a rear-unloading semi-trailer. Background Technology

[0002] With the increasing demand for transporting bulk materials such as coal, sand, gravel, mineral powder, and construction waste, rear-dump semi-trailers, especially electric rear-dump semi-trailers, are widely used due to their high unloading efficiency and strong adaptability. In actual operation, vehicles typically require a hydraulic lifting mechanism to raise the cargo box, allowing the cargo to be unloaded by gravity. However, during the lifting process, the vehicle's center of gravity continuously shifts upwards as the cargo box tilt angle increases, and it is susceptible to factors such as uneven ground, load shifting, cargo adhesion, unloading impact, and lateral disturbances, leading to rapid changes in vehicle posture and, in severe cases, rollover accidents. Therefore, how to monitor and actively protect the vehicle's stability in real time during lifting operations has always been a key technical issue in this field.

[0003] In existing technologies, the control methods for the lifting safety of rear-unloading semi-trailers mostly employ empirical threshold judgment, mechanical limit protection, or alarm methods based on a single tilt angle parameter. For example, when the lifting angle of the trailer reaches a preset value, or when the yaw or tilt of the vehicle body reaches a fixed threshold, an early warning is issued or the lifting is stopped. While such solutions can provide basic protection to a certain extent, they are usually based on static safety boundaries and fail to fully consider the dynamic effects of continuous changes in vehicle posture, load transfer, and external disturbances during the lifting process. Therefore, they are difficult to accurately reflect the true stability state under different working conditions. Furthermore, existing solutions typically focus on instantaneous judgment of the current state and lack the ability to predict future short-term posture evolution trends. For example, when the vehicle is close to the instability boundary but has not yet reached the fixed alarm threshold, traditional methods often cannot identify potential overturning risks in advance, nor can they generate targeted intervention measures in a timely manner according to the degree of risk, resulting in delayed early warnings and affecting operational safety and control effectiveness.

[0004] Therefore, in view of the problem that the stability boundary changes dynamically with the working conditions during the lifting operation of the rear-unloading semi-trailer, and that the existing technology is unable to achieve early risk prediction and graded intervention, there is an urgent need to propose a method that can combine real-time attitude data, dynamic boundary changes and future short-term disturbance trends to dynamically assess and actively protect the stability of the lifting operation. Summary of the Invention

[0005] This invention provides a method and device for dynamic protection of stability boundaries in lifting operations of rear-unloading semi-trailers, which improves the safety of lifting operations by calculating stability boundaries, extracting attitude disturbances, predicting future trajectories, and controlling risks.

[0006] In a first aspect, the present invention provides a dynamic protection method for the stability boundary of a rear-unloading semi-trailer lifting operation, the method comprising: Step S1: Collect raw angular velocity and acceleration data during the lifting operation using sensors, calculate the current lifting angle, fuse the static boundary threshold and dynamic margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary under the current lifting angle, and obtain the initial monitoring dataset. Step S2: Perform signal smoothing processing on the angular velocity data in the initial monitoring dataset, extract interference feature parameters, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. Step S3: Based on the attitude disturbance vector, perform random disturbance simulation to generate multiple sets of future short-term attitude change trajectories. Evaluate the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory sets, and generate the path probability distribution and average probability value. Step S4: Determine the risk level based on the shape of the probability distribution and the average probability value; generate an intervention command in conjunction with the current operating parameters to adjust the lifting operation status.

[0007] Secondly, the present invention also provides a dynamic protection device for the stability boundary of a rear-unloading semi-trailer lifting operation, used to implement the above-mentioned method, the device comprising: The monitoring data acquisition unit is used to collect raw angular velocity and acceleration data during the lifting operation through sensors, calculate the current lifting angle, fuse the static boundary threshold and margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary under the current lifting angle, and obtain the initial monitoring dataset. The interference feature extraction unit is used to perform signal smoothing processing on the angular velocity data in the initial monitoring dataset, extract interference feature parameters, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. The overturning probability assessment unit is used to perform random disturbance simulation based on the attitude disturbance vector, generate multiple sets of future short-term attitude change trajectories, assess the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory set, and generate path probability distribution and average probability value. The intervention and adjustment unit is used to determine the risk level based on the shape of the probability distribution and the average probability value; and to generate intervention instructions in combination with the current working condition parameters to adjust the lifting operation status.

[0008] Thirdly, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0009] The beneficial effects of this invention are as follows: This invention collects raw angular velocity and acceleration data during the lifting process and integrates static and dynamic boundary thresholds to calculate the dynamic stability boundary at the current lifting angle in real time. This not only reflects the basic stability margin of the vehicle at different lifting stages but also incorporates actual working conditions such as load transfer, uneven ground, hydraulic fluctuations, and external disturbances into the boundary correction process, thus providing initial monitoring data that more closely reflects the actual operating conditions for subsequent risk analysis. Secondly, by smoothing the angular velocity data, effective attitude change information and high-frequency disturbance information are separated, disturbance feature parameters are extracted, and an attitude disturbance vector is constructed by combining the change in distance between state points. This allows the system to characterize the true evolution trend of the vehicle's lifting attitude and quantify key factors affecting instability such as cargo release, vibration impact, and boundary approach speed. Furthermore, random disturbances are applied based on the aforementioned attitude disturbance vector. The simulation generates multiple sets of future short-term attitude change trajectories, and combines dynamic stability boundaries to perform cross-boundary probability assessments on each trajectory, obtaining path probability distributions and average probability values. This expands the traditional method of judging only the current moment to a probabilistic prediction of future short-term overturning risks, improving the foresight of risk identification. Finally, the risk level is determined based on the probability distribution pattern and average probability value, and intervention commands are generated by combining real-time lifting angle, lifting speed, load weight, and distance from the state point, enabling proactive control of operational states such as reducing lifting speed, pausing lifting, or maintaining alert. Through the cooperation of the above steps, this invention can achieve dynamic assessment of the stability boundary of electric rear-dump semi-trailer lifting operations, early prediction of overturning risks, and graded intervention for dangerous working conditions, thereby improving the safety, timeliness of control, and accuracy of risk protection in lifting operations. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of the dynamic protection method for the stability boundary of the rear unloading semi-trailer lifting operation in the embodiment. Figure 2 This is a flowchart of the method for generating a set of future short-term attitude change trajectories in the embodiment; Figure 3 The curve showing the average probability value changing over time in the example; Figure 4 This is a structural diagram of the dynamic protection device for the stability boundary of the rear unloading semi-trailer lifting operation in the embodiment. Detailed Implementation

[0012] This invention provides a method and apparatus for dynamic protection of stability boundaries during the lifting operation of a rear-unloading semi-trailer. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0013] For ease of understanding, the specific process of the embodiments of the present invention will be described below, such as... Figure 1 As shown in the embodiment of the present invention, the dynamic protection method for the stability boundary of the lifting operation of the rear-unloading semi-trailer includes: Step S1: Collect raw angular velocity and acceleration data during the lifting operation using sensors, calculate the current lifting angle, fuse the static boundary threshold and dynamic margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary at the current lifting angle, and obtain the initial monitoring dataset; specifically including: Raw angular velocity and acceleration data are collected during the lifting operation using gyroscopes and accelerometers. Based on this data, the current lifting angle is calculated. A preset static boundary threshold is queried based on the current lifting angle to determine the static stability boundary at that angle. A dynamic margin coefficient is determined based on the fluctuation amplitude of real-time angular velocity and acceleration data, and this coefficient is used to adjust the static stability boundary to obtain the dynamic stability boundary. The current attitude state point is determined in real time, and the distance between the attitude state point and the dynamic stability boundary, as well as the boundary approach rate, are calculated. The raw angular velocity data, raw acceleration data, distance between the attitude state point, and boundary approach rate are integrated to obtain the initial monitoring dataset.

[0014] Specifically, during the lifting process of the cargo box of the rear-dump semi-trailer, as the lifting angle increases, the center of gravity of the vehicle continues to shift backward and rise. At the same time, it is affected by factors such as cargo imbalance, unloading adhesion, crosswind disturbance, uneven supporting ground, and hydraulic lifting fluctuations. As a result, the margin between the current posture of the vehicle and the stability boundary changes dynamically. Therefore, it is necessary to calculate and dynamically correct the stability boundary in real time during the lifting process to provide basic data for subsequent risk assessment and active protection.

[0015] During implementation, raw angular velocity and raw acceleration data are collected during the lifting operation using gyroscopes and accelerometers. The gyroscopes are installed on the vehicle body, frame, or hinged parts related to the lifting motion to measure the angular velocity of the vehicle body rotating around the lifting hinge point; the accelerometers measure the linear acceleration of the vehicle body and frame in each axis during the lifting process. The raw angular velocity and raw acceleration data are collected synchronously according to a preset sampling period and timestamped to form a continuous time-series data stream. To mitigate the impact of initial installation deviations, the accelerometer can be zero-calibrated in the initial horizontal position of the vehicle, and static drift compensation can be applied to the gyroscope output. An onboard coordinate system is established: the X-axis represents the longitudinal direction of the vehicle body in a horizontal state; the Y-axis points laterally and is perpendicular to the X-axis in a horizontal state; and the Z-axis points vertically upward and is perpendicular to the horizontal position of the vehicle body. Based on the original angular velocity and acceleration data and this coordinate system, the current lifting angle is calculated. Specifically, in the initial horizontal position of the vehicle, the components of the accelerometer along each axis are mapped to the coordinate system. During the lifting process, the angular velocity output by the gyroscope is integrated over time to obtain the angle change, where the current lifting angle is... The calculation formula is: in, Let be the angular velocity at time t. The angle value at time t is the angle value of the previous time step. The sampling period is defined as follows: The integral result is corrected using the gravity direction detected by the accelerometer to obtain the current lift angle relative to the horizontal plane. The lift angle at the current time t is calculated using the following formula: in, These are the components of the three-axis acceleration along the three coordinate axes. In the low-frequency range, the accelerometer's calculation result is trusted to suppress drift, while in the high-frequency range, the gyroscope's integral result is trusted for rapid response to changes. The final output is the current lift angle. Since the accelerometer is susceptible to vibration interference under dynamic conditions, this angle is stable in the long term but has high short-term noise. The gyroscope's integral result has high short-term accuracy but drifts in the long term. Therefore, complementary filtering or Kalman filtering is used to fuse these two angles. First, the gyroscope-integrated angle is used as the state prediction value, and the prior estimation error is updated based on the system noise covariance matrix. Second, the accelerometer-calculated angle is used as the observed value, and the Kalman gain is calculated based on the observed noise covariance matrix. Finally, this gain is used to weight and correct the state prediction value, obtaining the posterior estimate as the current optimal lift angle output. Through this recursive process, Kalman filtering can dynamically balance the dynamic response advantage of the gyroscope and the anti-drift characteristics of the accelerometer, effectively reducing the cumulative integration error while suppressing vibration noise, thus achieving accurate estimation of the lift angle.

[0016] After obtaining the current lifting angle, the preset static boundary threshold is queried based on the current lifting angle to determine the static stability boundary at the current lifting angle. The aforementioned static boundary threshold is a pre-established and stored set of boundary parameters, which can be obtained by comprehensively considering the vehicle structural parameters, namely wheelbase, track width, suspension support relationship, rated load, cargo box geometry, lifting mechanism parameters, and prototype vehicle calibration or mechanical experiment simulation results. The aforementioned lifting mechanism parameters include at least the lifting hinge point position between the cargo box and the frame, the hydraulic cylinder installation position, the hinge point positions at both ends of the hydraulic cylinder, the effective stroke of the hydraulic cylinder, the maximum lifting angle of the cargo box, and the forces related to the lifting. Lever arm parameters; the aforementioned static boundary thresholds can be stored in the controller in the form of a lookup table. The precise boundary thresholds in the table are obtained through experimental simulation. After obtaining the current lifting angle, the controller calculates the static boundary thresholds corresponding to the lifting angle by looking up the table or interpolation. The static boundary thresholds can be composed of the allowable stability moment arm, allowable overturning critical moment, or equivalent limit attitude parameters corresponding to different lifting angles for a specific vehicle model. The aforementioned static stability boundaries are used to characterize the allowable limit attitude range, limit load range, or limit stability moment range when the vehicle maintains a stable state under conditions without significant dynamic disturbances.

[0017] Furthermore, based on the fluctuation amplitude of real-time angular velocity and real-time acceleration data, a dynamic margin coefficient is determined, and the static stability boundary is adjusted using this dynamic margin coefficient to obtain the dynamic stability boundary. Specifically, within a preset time window, fluctuation characteristic quantities are calculated for the real-time angular velocity and real-time acceleration data, including at least one or more of standard deviation, root mean square value, and peak-to-peak value. The dynamic margin coefficient is then determined based on these fluctuation characteristic quantities, and converted to a positive value not exceeding a preset upper limit (e.g., 1) according to a preset mapping rule (e.g., by looking up a table). The product of the dynamic margin coefficient and the static boundary threshold is taken as the dynamic stability boundary. When the fluctuations in angular velocity and acceleration increase, the dynamic margin coefficient... The dynamic margin coefficient is reduced to shrink the dynamic stability boundary toward the unstable region. When the fluctuations in angular velocity and acceleration decrease, the dynamic margin coefficient is increased or restored to the reference value, such as 0.9, so that the dynamic stability boundary approaches the aforementioned static stability boundary. Thus, the stability boundary can be transformed from a static stability boundary that only reflects the theoretical limit into a dynamic stability boundary that can reflect the actual working conditions such as load offset, hydraulic fluctuation, uneven ground, and external disturbances during the lifting process. The aforementioned dynamic margin coefficient can be obtained by a preset mapping relationship, which can be a lookup table relationship. For example, the standard deviation of angular velocity and the root mean square value of acceleration can be normalized and weighted to obtain a comprehensive disturbance index, and then the dynamic margin coefficient can be obtained by looking up a table based on the comprehensive disturbance index.

[0018] After obtaining the dynamic stability boundary, the current attitude state point is determined in real time, and the distance between the attitude state point and the dynamic stability boundary, as well as the boundary approach rate, are calculated. Specifically, the attitude state point is used to characterize the actual state of the vehicle at the current moment, and is preferably composed of the current lifting angle and the current equivalent load torque. The current equivalent load torque can be calculated based on the load weight, the position of the load center of gravity, the geometric parameters of the vehicle body, and the relationship of gravity. The calculation formula is as follows: in, For the current equivalent constant torque, This represents the current lifting angle of the carriage relative to the horizontal plane. It reflects the height distribution of the load's center of gravity relative to the car floor. To reflect the distribution of load along the length of the carriage, The total mass of the goods currently inside the carriage is the aforementioned load weight. This is the acceleration due to gravity.

[0019] Projecting the aforementioned attitude state points into the same state space as the dynamic stability boundary, the shortest distance to the dynamic stability boundary or the projection distance along the preset overturning sensitive direction is calculated. The resulting quantity is the state point distance, which characterizes the stability margin of the current state relative to the dynamic stability boundary. The smaller the value, the closer the current state is to the instability boundary. Subsequently, the state point distances obtained in multiple consecutive sampling periods are used to form a distance sequence. The ratio of the difference between the corresponding state point distances in the current period and the previous period to the corresponding time interval is calculated using the finite difference method as the rate of change of distance over time to obtain the boundary approach rate. When the boundary approach rate is negative, it indicates that the current attitude state point is approaching the dynamic stability boundary. When the absolute value of the boundary approach rate increases, it indicates that the approach speed is accelerating. Quickly; finally, the above-mentioned raw angular velocity data, raw acceleration data, state point distance, and boundary approach rate are integrated to obtain the above-mentioned initial monitoring dataset. Preferably, each sampling time is used as a unit to establish a record item containing raw angular velocity data, raw acceleration data, state point distance, boundary approach rate, and timestamp, and arranged in chronological order to form the above-mentioned initial monitoring dataset. The above-mentioned initial monitoring dataset not only retains the original sensing information during the lifting operation, but also introduces state quantities and trend quantities directly related to the stability boundary, so as to simultaneously reflect the current lifting state, the current stability margin, and the risk evolution speed, providing a unified data foundation for subsequent angular velocity signal smoothing processing, interference feature parameter extraction, attitude interference vector construction, and future short-term attitude change trajectory prediction.

[0020] Step S2: Perform signal smoothing processing on the raw angular velocity data in the initial monitoring dataset, extract interference feature parameters based on residual analysis, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. In step S2, the angular velocity fluctuation sequence and the corresponding disturbance characteristic parameters are generated, including: For the raw angular velocity data in the initial monitoring dataset, a sliding window mean filtering method is used to smooth the signal to suppress high-frequency random interference, resulting in a smoothed angular velocity data sequence. Based on the difference between the smoothed angular velocity data sequence and the raw angular velocity data sequence, an interference residual sequence is obtained. The interference amplitude distribution and frequency distribution of the interference residual sequence are analyzed to extract interference characteristic parameters. The smoothed angular velocity data sequence is then used as the angular velocity fluctuation sequence.

[0021] Specifically, since rear-discharge semi-trailers are easily affected by hydraulic pulsation, sudden release of cargo after sticking, frame vibration and lateral disturbance during the lifting and unloading process, the original angular velocity signal usually contains both low-frequency change components reflecting the actual lifting posture evolution and high-frequency interference components reflecting instantaneous impact and mechanical vibration. Therefore, it is necessary to separate the two first to facilitate subsequent analysis.

[0022] During implementation, a raw angular velocity sequence composed of raw angular velocity data arranged in chronological order is extracted from the initial monitoring dataset. A sliding window of fixed time length is set according to a preset sampling frequency. The raw angular velocity values ​​within the current sampling time and the consecutive windows before it are averaged. The average value is used as the smoothed angular velocity value corresponding to the current time. The sliding window is progressively adjusted point by point as the sampling time progresses until the entire raw angular velocity data sequence is traversed. This yields a smoothed angular velocity data sequence corresponding to each moment of the raw angular velocity data. This suppresses rapid fluctuations caused by sensor noise, transient shocks in the hydraulic system, and local mechanical vibrations, while preserving low-frequency trend angular velocity changes related to lifting actions, load transfer, and vehicle attitude evolution.

[0023] After obtaining the smoothed angular velocity data sequence, the original angular velocity value and the smoothed angular velocity value at the same sampling time are subtracted to obtain the interference residual sequence. This interference residual sequence represents the interference component separated from the original angular velocity signal. Amplitude statistics are performed on the interference residual sequence within a preset analysis period, calculating its maximum, minimum, and root mean square (RMS) values. The maximum and minimum values ​​represent the upper and lower boundaries of interference fluctuations within the analysis period, and the RMS value represents the average energy level of the interference. Simultaneously, a Fast Fourier Transform (FFT) is performed on the interference residual sequence to convert the time-domain residual signal into a frequency-domain spectrum. A preset noise floor threshold is used as the criterion to identify significant frequency components with amplitudes exceeding a preset threshold, and the dominant frequency value and corresponding amplitude of each significant frequency component are recorded. The maximum, minimum, and RMS values, along with the dominant frequency value and corresponding amplitude, are correlated and integrated to form the interference characteristic parameters. These interference characteristic parameters are used to quantitatively characterize the intensity level of interference during the current lifting operation phase. The data consists of the main frequency components to help determine whether the current disturbance has sudden impact characteristics or specific periodic vibration characteristics. The smoothed angular velocity data sequence is directly used as the angular velocity fluctuation sequence, which characterizes the effective angular velocity change process after noise suppression. It mainly reflects the real dynamic trend generated during the lifting of the semi-trailer cargo box due to hydraulic extension and retraction, cargo transfer, and gradual changes in vehicle attitude. The angular velocity fluctuation sequence represents the effective fluctuation component, while the aforementioned disturbance characteristic parameters represent the disturbance component, thus avoiding the loss of disturbance information by only retaining the smoothed result. Through this technical solution, not only is an angular velocity fluctuation sequence suitable for subsequent attitude evolution analysis obtained, but also disturbance characteristic parameters that reflect the disturbance intensity and periodic disturbance characteristics are simultaneously obtained. This allows for the subsequent construction of the attitude disturbance vector, enabling the use of the angular velocity fluctuation sequence to characterize the actual change trend of the lifting attitude, and the combination of disturbance characteristic parameters to quantify the additional impact of the disturbance on the stability boundary approximation process, thereby improving the accuracy of subsequent trajectory simulation and overturning probability assessment.

[0024] Further, in step S2, the attitude disturbance vector is constructed, including: Based on the angular velocity fluctuation sequence, the amplitude and frequency features of the angular velocity changes are extracted to characterize the fluctuation intensity and dominant frequency of the effective attitude evolution; based on the interference feature parameters, the interference intensity features are extracted to characterize the energy level and dominant frequency structure of the suppressed interference; based on the real-time tracked state point distance sequence, the rate of change of the state point distance is calculated to characterize the trend and speed of the current attitude approaching the stability boundary; after normalizing the amplitude features, the frequency features, the interference intensity features, and the rate of change of the state point distance, they are integrated to construct the attitude interference vector.

[0025] Specifically, although the aforementioned angular velocity fluctuation sequence can characterize the effective attitude change trend after noise suppression, it is still necessary to further extract its amplitude and frequency characteristics, and combine them with the interference characteristic parameters and the dynamic changes in the distance between state points to uniformly quantify the current disturbance state. Only in this way can the subsequent trajectory prediction reflect both the attitude change itself and the evolution direction of the instability risk. Specifically, continuous subsequences are extracted from the aforementioned angular velocity fluctuation sequence according to a preset time window, and amplitude and frequency characteristics are extracted respectively. The amplitude characteristics can be obtained by calculating the difference between the peak and valley values ​​within the subsequence, which is used to quantify the strength of the effective fluctuation of the lifting attitude of the truck body in the current time period. The frequency characteristics can be obtained by performing a fast Fourier transform on the subsequence and identifying its dominant frequency value, which is used to quantify the dominant oscillation rate of the effective attitude change. For rear-unloading semi-trailers, the amplitude characteristics are small and the frequency characteristics are low-frequency when unloading is smooth. When cargo sticks together and suddenly slips or the vehicle body is impacted, the amplitude characteristics will increase significantly, and the frequency characteristics may become prominent in the mid-to-high frequency range.

[0026] Simultaneously, corresponding interference characteristic parameters are obtained from the interference residual subsequence corresponding to the time of the aforementioned angular velocity fluctuation subsequence, and interference intensity features are extracted. These interference characteristic parameters include at least the root mean square (RMS) value of the interference residual, the dominant frequency value, and the corresponding amplitude. The RMS value directly characterizes the average energy level of the filtered interference and can serve as a fundamental component of the interference intensity feature. The dominant frequency value and corresponding amplitude reflect the main frequency composition and intensity of the interference, and can be used to further calculate the concentration or dispersion index of the interference energy, for example, by calculating the variance of the energy distribution in each frequency band. Therefore, the aforementioned interference intensity feature is independent of the amplitude and frequency characteristics of the effective attitude and is specifically used to quantify high-frequency or transient disturbances that have been filtered and suppressed but still reflect environmental or mechanical anomalies. Furthermore, based on the boundary proximity trend of the aforementioned state point distance sequence analysis, the distance between state points in multiple consecutive sampling periods is differentially calculated to obtain a distance change rate sequence, which is then smoothed using moving average filtering or linear fitting to suppress noise. The smoothed distance change rate sequence directly... The velocity and direction of the current attitude state point approaching the dynamic stability boundary are characterized, with negative values ​​indicating approach. The mean of this approach within the time window corresponding to the aforementioned angular velocity subsequence is used as the boundary approach trend feature. Finally, the amplitude features, frequency features, disturbance intensity features, and boundary approach trend features from different data sources are normalized to unify their scales and then concatenated in a preset order to construct an attitude disturbance vector. This vector is a multi-dimensional array, with each component directly corresponding to an independent type of attitude risk influencing factor: effective motion intensity, effective motion rate, environmental disturbance intensity, and the rate of change of stability margin. Through the above technical solution, the constructed attitude disturbance vector no longer relies on fuzzy intermediate influence indicators but clearly and completely carries parallel multi-source information. It can be directly used as input for subsequent Monte Carlo simulations to generate a set of future attitude change trajectories that reflect both the main motion trend and the disturbance and risk trends, thereby significantly improving the accuracy and comprehensiveness of stability risk prediction for rear-unloading semi-trailer lifting operations.

[0027] Step S3: Based on the attitude disturbance vector, perform random disturbance simulation to generate multiple sets of future short-term attitude change trajectories. Evaluate the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory sets, and generate the path probability distribution and average probability value. In step S3, a set of future short-term attitude change trajectories is generated, including: Using the attitude disturbance vector as a simulation parameter, the Monte Carlo method is employed to simulate multiple random disturbances to the current attitude state based on the simulation parameter, generating multiple future short-term attitude change trajectories. For each attitude change trajectory, it is projected into a state space defined by a dynamic stability boundary, and the trajectory boundary distance between each state point corresponding to the attitude change trajectory and the dynamic stability boundary is calculated. All generated attitude change trajectories and their corresponding trajectory boundary distance sequences are correlated and integrated to form a set of future short-term attitude change trajectories used for overturning probability assessment.

[0028] Specifically, based on the aforementioned attitude disturbance vector, the possible attitude evolution states in the short term during the lifting and unloading process of the rear-unloading semi-trailer are randomly expanded using multiple paths to generate a set of future short-term attitude change trajectories for assessing the probability of overturning. Since the lifting angle of the trailer continuously increases during the lifting and unloading process, the center of gravity of the cargo dynamically shifts along the longitudinal and vertical directions of the trailer, and is superimposed with uncertain factors such as hydraulic system fluctuations, sudden release of cargo adhesion, crosswind disturbances, elastic deformation of the chassis, and local subsidence of the supporting ground, it is difficult to accurately reflect the short-term instability trend based solely on the single attitude state at the current moment. Therefore, it is necessary to transform the attitude disturbance information at the current moment into multiple possible future state evolution paths.

[0029] During implementation, amplitude features, frequency features, disturbance intensity features, and boundary approach trend features are extracted from the aforementioned attitude disturbance vector and mapped to control parameters in Monte Carlo random disturbance simulation. The amplitude feature characterizes the strength of the current effective attitude fluctuation and is used to limit the basic range of random disturbance changes at each sampling time. The frequency feature characterizes the dominant rate of change of the current attitude fluctuation and is used to determine the time resolution and state update rhythm during trajectory extrapolation. The disturbance intensity feature is derived from the root mean square value, dominant frequency value, and corresponding amplitude extracted from the disturbance residual sequence in the previous steps, and is used to characterize the additional intensity of the current high-frequency or transient disturbance, serving as the basis for adjusting the random diffusion range. The boundary approach trend feature is derived from the rate of change of the distance between state points and is used to characterize the speed and direction of the current attitude state point approaching the dynamic stability boundary, and to correct the tendency of the trajectory to expand into the unstable region. Therefore, the components in the aforementioned attitude disturbance vector are not simply used in parallel during trajectory generation, but rather correspond to different control variables in the random simulation, thus allowing the random simulation to simultaneously retain the effective attitude evolution characteristics, additional disturbance intensity characteristics, and stability margin decay characteristics.

[0030] After completing the simulation parameter mapping, the current attitude state point is used as the initial state. Multiple random disturbance simulations are performed using the Monte Carlo method. Preferably, in each simulation, a random sampling distribution is constructed based on the aforementioned disturbance intensity parameters and random diffusion parameters. The attitude state variables for the next short time are updated hourly according to a preset sampling step size, resulting in a trajectory of future short-term attitude changes. The random sampling distribution preferably uses a Gaussian distribution or a truncated Gaussian distribution, allowing the random disturbance to fluctuate around the dominant adaptive trend under the current working condition, reflecting the natural dispersion caused by factors such as hydraulic pulsation, changes in ground adhesion, and uneven cargo release during the lifting operation of the semi-trailer. When the aforementioned frequency characteristics are high, the state update step size is shortened to improve the characterization accuracy of rapidly fluctuating working conditions. When the aforementioned frequency... When the feature is low, the state update step size is increased to highlight the gradual change trend during the lifting of the carriage. At the same time, when the above boundary is close to the trend feature indicating that the current state point is rapidly approaching the dynamic stability boundary, the correction weight for expanding towards the boundary is increased during the state update process, so that the generated trajectory can better reflect the characteristics of rapid decay of stability margin under high-risk conditions. When the above disturbance intensity feature increases, the random diffusion range is increased accordingly to improve the dispersion between different simulated trajectories, thereby more realistically covering the attitude change paths that may occur under abnormal unloading or sudden disturbance scenarios. By repeatedly executing multiple random disturbance simulations, such as hundreds to thousands of times, multiple future short-term attitude change trajectories are generated to form a candidate path set for the future short-term state evolution under the current conditions.

[0031] After obtaining each future short-term attitude change trajectory, the trajectory state points corresponding to each sampling time are projected onto the state space corresponding to the dynamic stability boundary. Based on the same state quantity definition and measurement method as the dynamic stability boundary, the trajectory boundary distance between each trajectory state point and the dynamic stability boundary is calculated. This trajectory boundary distance is preferably defined as the shortest Euclidean distance from the trajectory state point to the dynamic stability boundary, or the projection distance along a preset overturning sensitive direction. When the trajectory state point is located on the safe side of the dynamic stability boundary, the trajectory boundary distance is positive, indicating that there is still a stability margin at the corresponding sampling time. As the trajectory state point approaches the dynamic stability boundary, the trajectory boundary distance gradually decreases. When a state point crosses the dynamic stability boundary, the aforementioned trajectory boundary distance is recorded as a negative value to indicate that a potential overturning risk has occurred at that moment. Furthermore, by continuously recording the trajectory boundary distances corresponding to each attitude change trajectory at multiple sampling moments, a trajectory boundary distance sequence corresponding to each trajectory can be obtained. Based on this, the boundary proximity of the attitude change trajectory relative to the dynamic stability boundary can be obtained, including the minimum boundary distance, the trend of boundary distance changes, and whether there is a boundary crossing moment. This ensures that each trajectory not only reflects a possible future attitude evolution but also carries the temporal and spatial relationship with the dynamic stability boundary, enabling subsequent risk assessments to be made directly based on the correspondence between the path and the boundary distance sequence.

[0032] Finally, all the generated attitude change trajectories and their corresponding trajectory boundary distance sequences are associated and integrated according to trajectory number, sampling time, and state parameters to form a set of future short-term attitude change trajectories. Preferably, the set of future short-term attitude change trajectories is stored in the form of structured time-series data, wherein each trajectory record contains at least a trajectory identifier, the attitude state value at each future sampling time, and the corresponding trajectory boundary distance sequence. For trajectories with negative trajectory boundary distances, boundary crossing markers can be added to prioritize the identification of potentially high-risk paths during subsequent probability assessment. Through the above technical solution, the set of future short-term attitude change trajectories realizes the transformation from the current attitude disturbance vector to the future multi-path state evolution result. It retains both the diversity of future short-term attitude changes of the rear-unloading semi-trailer under the current lifting and unloading conditions and the proximity and crossing risk between each path and the dynamic stability boundary. This provides a direct, complete, and quantifiable data foundation for subsequent trajectory-by-trajectory assessment of the overturning probability of crossing the dynamic stability boundary, generating path probability distributions, and calculating average probability values.

[0033] Further, in step S3, the path probability distribution and average probability value are generated, including: Obtain the sequence of trajectory state points for each attitude change trajectory in the trajectory set over multiple sampling periods; recalculate the real-time trajectory boundary distance between each trajectory state point and the real-time dynamic stability boundary based on the real-time dynamic stability boundary provided by the dynamic boundary monitoring module; evaluate the overturning probability of each attitude change trajectory crossing the real-time dynamic stability boundary over multiple sampling periods based on the real-time trajectory boundary distance sequence corresponding to each attitude change trajectory; summarize and statistically analyze the overturning probabilities of all attitude change trajectories in the trajectory set to generate a path probability distribution; calculate the average probability value based on the path probability distribution.

[0034] Specifically, since the lifting and unloading of the semi-trailer is a dynamic process, its stability boundary continuously evolves with the increase of the lifting angle, the shift of the cargo center of gravity, and changes in ground support conditions. Therefore, directly using the static boundary generated at the simulation time for risk assessment will produce lag errors. It is necessary to introduce real-time boundary monitoring data for dynamic correction. Specifically, the trajectory state point sequence of each attitude change trajectory in the future multiple sampling periods is read from the aforementioned set of future short-term attitude change trajectories. The aforementioned trajectory state point sequence is generated by Monte Carlo random perturbation simulation and represents the attitude state values ​​at each sampling time on different possible evolution paths in the future short time. It does not contain a spatial relationship with the dynamic stability boundary itself. At the same time, the dynamic stability boundary data at the current time is obtained in real time from the dynamic boundary monitoring module. This boundary is represented as a continuous surface or threshold curve in the state space.

[0035] Subsequently, for each attitude change trajectory, the trajectory state points at each sampling time in the trajectory are spatially mapped and distances are recalculated with the aforementioned real-time dynamic stability boundary. Specifically, each trajectory state point is projected into the same state space defined by the aforementioned real-time dynamic stability boundary, and the shortest distance from the state point to the real-time dynamic stability boundary or the distance along the preset instability direction is calculated, thereby obtaining the real-time trajectory boundary distance corresponding to each trajectory state point. Specifically, by traversing all trajectory state points on each trajectory, a sequence of real-time trajectory boundary distances corresponding to that trajectory can be obtained. The above process essentially re-registers the simulated future possible state points with the most realistic stability safety boundary at the current moment, thereby eliminating the risk assessment benchmark error caused by dynamic boundary changes. For rear-unloading semi-trailers, when cargo is suddenly released during unloading or the truck bed is lifted close to its maximum angle, the real-time dynamic... The stability boundary may rapidly shrink towards the danger side. At this time, even for the same trajectory state point, its real-time distance relative to the new boundary will be significantly smaller than its distance relative to the old boundary, thus more accurately reflecting the real risk under the current working condition. After obtaining the real-time trajectory boundary distance sequence corresponding to each attitude change trajectory, the overturning probability of each trajectory crossing the above-mentioned real-time dynamic stability boundary in the next few sampling periods is evaluated based on this sequence. Specifically, the proportion of sampling points in the real-time trajectory boundary distance sequence that cross the boundary at a distance less than a set distance (i.e., close to the state point) in the next few sampling periods is counted out of the total sampling points, and this proportion is used as the overturning probability of the trajectory. In the above way, the overturning probability of each attitude change trajectory is calculated based on the latest spatial relationship between its possible future state points and the current real-time boundary, so that the probability assessment results can respond in real time to the dynamic contraction and expansion of the stability boundary during the lifting operation.

[0036] After completing the rollover probability assessment for all attitude change trajectories, the rollover probability values ​​of all trajectories are aggregated according to trajectory identifiers to form a probability sequence. Statistical analysis is then performed on this probability sequence to generate a path probability distribution. Specifically, this can be achieved by dividing the probability values ​​into intervals and counting the number or frequency of trajectories within each interval, forming a discrete probability distribution in the form of a histogram. This path probability distribution visually demonstrates the dispersion of rollover risk corresponding to all simulated future evolution paths under the current rear-unloading semi-trailer lifting condition. For example, if the distribution exhibits a single-peak shape concentrated in low-probability intervals, it indicates that most possible paths are in a safe state; if the distribution shows... The presence of a bimodal pattern or a tailing towards a high-probability interval indicates that a significant portion of the possible paths will lead to a high-risk state, raising concerns about the overall operational stability. Finally, based on the aforementioned path probability distribution, an average probability value is calculated. Specifically, the arithmetic mean of the overturning probability values ​​of all trajectories in the probability sequence is taken. This average probability value represents the overall average overturning risk level based on all simulated paths at the current moment. This average probability value, as a comprehensive scalar, facilitates direct comparison with a preset risk threshold, thus providing a concise and reliable quantitative input for subsequent steps to determine whether a high-risk state has been entered and to identify the corresponding risk level. Figure 3 The figure shows the average probability value generated under a certain lifting condition as a function of time. The black solid line represents the trend of the average probability value obtained after Monte Carlo simulation and probability assessment, the gray solid line is the corresponding lifting angle change curve, and the three horizontal dashed lines represent the low-risk (0.2), medium-risk (0.4), and high-risk (0.6) thresholds, respectively. As can be seen from the figure, in the initial lifting stage (0-5 seconds), the average probability value remains below 0.1, and the system is in a safe state. As the lifting angle increases to above 30° (8-12 seconds), the average probability value gradually rises and exceeds the low-risk threshold. The simulated sudden release of cargo around 12.5 seconds causes the average probability value to rapidly climb to 0.55. The system briefly touched a high-risk area, at which point it triggered an intervention command at 16.5 seconds to implement speed reduction control (details to be explained later). Subsequently, the probability value dropped back below the medium-risk threshold. This curve visually verifies that the above method can dynamically track the risk evolution trend throughout the lifting process and respond promptly when a high-risk event occurs, effectively suppressing the risk of overturning. The above technical solution, through a series of processes from trajectory state point acquisition, real-time distance recalculation, single trajectory probability assessment to ensemble statistical averaging, achieves deep integration of multi-path future attitude simulation results with real-time boundary monitoring data, ultimately outputting a probabilistic index that accurately reflects the short-term future overturning risk of the rear-unloading semi-trailer lifting operation.

[0037] Step S4: If the average probability exceeds a preset threshold, determine the risk level according to the morphological risk level determination model of the probability distribution; based on the risk level and the current working condition parameters, generate an intervention command to adjust the lifting operation status.

[0038] In step S4, determining the risk level includes: Obtain the average probability value and the path probability distribution; based on the path probability distribution, extract and quantify its peak position, distribution width, and skewness features; construct a risk level determination model, wherein the average probability value is used as the basic risk level input, and the peak position, distribution width, and skewness features are used as risk distribution structure correction input; calculate using the risk level determination model and output the corresponding risk level.

[0039] Specifically, since the risk of overturning of a semi-trailer during the lifting and unloading process depends not only on the average level of the overturning probability corresponding to all simulated trajectories, but also on the degree of concentration, dispersion, and asymmetry of the shift of such overturning probabilities towards high-risk areas among various future evolution paths, the above-mentioned average probability value is used as the basic risk level. At the same time, the peak position, distribution width, and skewness characteristics of the path probability distribution are used as the basis for correcting the risk distribution structure. The two are calculated in a unified manner through a risk level judgment model to avoid missing the high-risk path at the tail by judging only based on the average value.

[0040] During implementation, the aforementioned average probability value and path probability distribution are obtained. The average probability value is the overall risk scalar obtained by statistically averaging the overturning probabilities corresponding to all future short-term attitude change trajectories. The path probability distribution is the distribution result formed by statistically analyzing the overturning probabilities corresponding to each attitude change trajectory according to a preset probability interval. Subsequently, feature extraction is performed on the path probability distribution to form structured distribution parameters for determining the risk level. The peak position is used to characterize the probability interval with the highest concentration of overturning probabilities, obtained by identifying the center value of the interval with the highest frequency or probability density in the path probability distribution. This parameter reflects the main risk level corresponding to most possible future evolution paths. The distribution width is used to represent... The dispersion of the overturning probability of each trajectory can be calculated based on the standard deviation, variance, half-peak width, or the span of a preset percentile interval of the probability sequence. The larger the value, the more obvious the risk difference between future paths and the higher the system uncertainty. The above-mentioned skewness feature is used to characterize the asymmetry of the path probability distribution with respect to its central position. It can be calculated by normalizing the third-order central moment. When the skewness is positive and the value is large, it indicates that the distribution has an extended tail towards the high probability side. That is, although the risk of some trajectories may not be the majority, there are a few extremely high-risk paths. Such paths usually correspond to high-risk conditions such as sudden release of cargo, increased off-center loading, rapid instability of the supporting foundation, or sudden increase in crosswinds in the lifting operation of the semi-trailer after unloading. Therefore, they should be given higher weight in the level determination.

[0041] After extracting the aforementioned distribution features, risk level determination is performed. The preferred risk level determination methods are rule-based determination, weighted scoring, or table lookup mapping. The core principle is to use the average probability value as the basic risk level input, and the peak position, distribution width, and skewness characteristics as inputs for risk distribution structure correction. Based on a preset mapping relationship, the corresponding risk level is output. Specifically, the average probability value characterizes the basic intensity of the overall overturning risk under the current working condition; the peak position corrects the degree of danger concentration along future predominantly driven paths; the distribution width corrects the amplification effect of uncertain risks caused by multi-path dispersion; and the skewness characteristics correct the extreme risk amplification effect caused by shifting towards the tail of the high-risk area. This ensures that the risk level formation process reflects a joint determination of average level and structural distribution. The formula is as follows: Preferably, a basic score can be calculated first based on the average probability value, and then a correction score can be added based on whether the peak position enters the high probability zone, whether the distribution width exceeds the preset width threshold, and whether the skewness is greater than the preset skewness threshold. Finally, one of the following—low risk, medium risk, and high risk—is output based on the correspondence between the comprehensive score and the preset level interval: When the average probability value is high and the peak position falls into the high probability zone, it indicates that most future trajectories tend to be in a dangerous state, and the comprehensive score should be raised to a higher level. When the average probability value does not increase significantly, but the distribution width is large and the skewness is positive, it indicates that the system has strong uncertainty and is accompanied by a small number of high-risk tail paths. At this time, the risk level can still be raised through structural correction to overcome the defect that the traditional single threshold logic cannot identify high-risk tail paths.

[0042] Furthermore, to better adapt the aforementioned risk level determination to the lifting operation scenario of rear-unloading semi-trailers, the input parameters used for risk level determination are preferably assigned different weights corresponding to the current lifting angle, load level, and ground support conditions. Specifically, at lower lifting angles, the lifting range of the truck bed's center of gravity is relatively limited, and the average probability value has a stronger dominant effect on the overall risk. At this time, the weight of the basic risk level input can be appropriately increased. However, at higher lifting angles, due to the increased center of gravity, reduced lateral stability margin, and more easily amplified external disturbances, the distribution width and skewness characteristics of the path probability distribution are more critical for identifying extreme risk paths. Therefore, the weight of the distribution structure correction input can be appropriately increased. Thus, the aforementioned risk level determination model can not only output a unified risk level based on the logical relationship between different risk indicators, but also adapt to the risk evolution pattern of rear-unloading semi-trailers at different lifting stages.

[0043] For example, when the peak position of the path probability distribution is concentrated around 0.70, and the average probability value is 0.65, while the distribution width is narrow, it indicates that most simulated paths stably fall into the higher risk range, and the system's future short-term overturning risk has strong consistency. At this time, the comprehensive score will fall into the high-risk range, thus outputting a high-risk level. The above technical solution realizes the standardized conversion from average probability value and path probability distribution to risk level. Among them, the average probability value, peak position, distribution width, and skewness characteristics form a clear data correspondence and algorithm processing relationship. That is, the average probability value determines the basic risk level, while the peak position, distribution width, and skewness characteristics jointly determine the degree of risk structure correction. The risk level output by the risk level determination model can not only characterize the overall danger level of the current lifting operation status of the unloading semi-trailer, but also reflect the concentration, dispersion, and tail anomaly of the risk evolution path in the future short term, thus providing a more comprehensive and accurate decision-making basis for subsequent risk signal triggering and active intervention measures.

[0044] Furthermore, in step S4, based on the current operating parameters, an intervention command is generated to adjust the lifting operation status, including: Based on the risk level, activate the corresponding active amplitude limiting protection strategy; obtain the current operating parameters, which include at least the real-time lifting angle, the current lifting speed setpoint, the lifting load weight, the real-time status point distance, and the attitude disturbance vector; based on the risk level and the current operating parameters, dynamically determine the target intervention mode and corresponding control parameters through a preset intervention decision model, and generate intervention commands to reduce the lifting speed, pause lifting, or maintain alert; send the intervention commands to the lifting operation control execution unit to adjust the lifting operation status.

[0045] Specifically, in this embodiment, step S4, based on the risk level determination already completed in the preceding steps, further transforms the risk identification result into an intervention command that can be directly executed by the rear-unloading semi-trailer lifting control system. This allows for dynamic adjustment of the lifting speed, lifting continuity, and control loop output strength during the lifting and unloading process, thereby suppressing the vehicle's attitude from continuing to develop towards the dangerous side of the dynamic stability boundary. Compared to the processing method that only triggers a single limiting logic when the warning level reaches the highest level, this embodiment does not adopt a static protection mode of single-level triggering and single-level response. Instead, it activates the corresponding level of active limiting protection strategy according to the risk level, and dynamically determines the target intervention mode and control parameters through a preset intervention decision model, combined with real-time lifting condition parameters and attitude disturbance vector characteristics, so that the protection action is adapted to the current risk level, current stability margin, and current disturbance intensity. Thus, step S4 realizes the conversion from "probabilistic risk assessment results" to "real-time execution control commands," forming a complete closed loop for the dynamic stability protection of the rear-unloading semi-trailer lifting operation.

[0046] Specifically, intervention commands are generated based on the aforementioned risk levels and current operating parameters to proactively adjust the lifting operation status of the rear-unloading semi-trailer. Since the lifting angle, load weight, lifting speed, and vehicle posture of the rear-unloading semi-trailer are constantly changing relative to the dynamic stability boundary during the lifting and unloading process, the risk level alone is insufficient to directly output specific control actions. It is also necessary to link the aforementioned risk level with the current operating parameters so that the generated intervention commands can match the current actual risk level.

[0047] During implementation, the corresponding active limiting protection strategy is activated based on the aforementioned risk level. This active limiting protection strategy constrains the lifting mechanism's operating rate and duration when the risk level reaches a preset level. Lower risk levels correspond to maintaining alertness or mild speed limiting strategies, while higher risk levels correspond to enhanced speed reduction or suspension of lifting strategies. During activation, preset protection parameters matching the rear-discharge semi-trailer model, rated load, hydraulic system parameters, and maximum lifting angle are loaded. A data connection is established between the active limiting protection strategy and the lifting control loop to receive current operating parameters and output intervention commands. Subsequently, the current operating parameters are acquired, including at least the real-time lifting angle, current lifting speed setpoint, lifting load weight, real-time status point distance, and the aforementioned... The attitude disturbance vector feature is defined as follows: the real-time lifting angle characterizes the current lifting degree of the carriage; the current lifting speed setpoint characterizes the target lifting speed currently issued by the control system; the lifting load weight characterizes the current load level of the carriage; the real-time state point distance characterizes the remaining stability margin between the current attitude state point and the dynamic stability boundary; and the attitude disturbance vector feature characterizes the severity of attitude fluctuations during the current lifting operation. Each parameter corresponds to a different constraint dimension in the intervention decision. The real-time lifting angle and lifting load weight are used together to determine the upper limit of the safe speed under the current operating condition; the real-time state point distance characterizes the boundary approximation degree; the attitude disturbance vector feature characterizes the disturbance amplification trend; and the current lifting speed setpoint is used to determine whether the existing control command exceeds the safe control range.

[0048] After obtaining the aforementioned current operating condition parameters, based on the risk level and current operating condition parameters, the target intervention mode and corresponding control parameters are dynamically determined through a preset intervention decision model. Specifically, based on the real-time lifting angle and lifting load weight, a preset load-angle-safe speed mapping table is consulted to obtain the theoretical maximum safe speed under the current operating condition. The difference between the current lifting speed setpoint and the aforementioned theoretical maximum safe speed is calculated, and the ratio of the difference to the theoretical maximum safe speed is used as the speed exceedance ratio. A speed exceedance ratio greater than zero indicates that the current lifting speed setpoint exceeds the theoretical maximum safe speed, equal to zero indicates that the two are equal, and less than zero indicates that the current lifting speed setpoint is less than the theoretical maximum safe speed. The aforementioned speed exceedance ratio and the real-time status point distance are input into the aforementioned intervention decision model, wherein the aforementioned intervention decision model is a preset rule model. When the real-time status point distance is less than a first distance threshold and greater than a second distance threshold, and the speed exceedance ratio is greater than zero, it is determined that the current lifting speed does not match the remaining stability margin, and an intervention is initiated. An intervention command to reduce the lifting speed is generated, with the target speed preferably set as the product of the theoretical maximum safe speed and a preset safety factor. This allows for a further control margin beyond the theoretical safety limit. If the speed exceedance ratio is less than or equal to zero, the system is considered temporarily safe. When the real-time state point distance is less than or equal to a second distance threshold, it indicates that the current attitude is approaching the critical region of the dynamic stability boundary. At this point, regardless of whether the speed exceedance ratio is greater than zero, an intervention command to pause the lifting is generated to prioritize preventing the attitude from evolving further towards the dangerous side. Here, the first distance threshold is greater than the second distance threshold. Furthermore, when the characteristics of the attitude disturbance vector that represent the current disturbance amplitude or disturbance intensity exceed a preset disturbance threshold, the intervention decision model is modified. For example, the intervention intensity is increased. For instance, if the real-time state point distance is less than the first distance threshold but greater than the second distance threshold, even if the speed exceedance ratio is small, the speed reduction can be increased, or the system can switch to the pause lifting mode in advance, thereby improving the system's response sensitivity to sudden disturbance conditions.

[0049] After determining the target intervention mode, a corresponding intervention command is generated. For the intervention mode of reducing the lifting speed, the intervention command includes at least the target speed value and the deceleration control parameter for transitioning from the current speed to the target speed value, so as to control the execution unit to smoothly reduce the lifting speed according to the preset deceleration curve. For the intervention mode of pausing lifting, the intervention command is preferably an immediate stop signal, and may be accompanied by a control indicator to maintain the current lifting state. If the current risk level reaches the preset level but the real-time state point distance is still greater than or equal to the first distance threshold, or greater than the second distance threshold but less than the first distance threshold, and the attitude disturbance vector feature does not exceed the disturbance threshold and the speed exceedance ratio is less than or equal to zero, then a maintain warning can be generated. The intervention command only maintains monitoring and does not change the current lifting control command. Finally, the intervention command is sent to the lifting operation control execution unit, which analyzes the intervention command and executes the corresponding speed adjustment or action pause operation to form the final protection measure and output it to adjust the lifting operation status. The above technical solution realizes the transformation from risk level to active protection action, enabling the rear unloading semi-trailer to adaptively select protective measures such as reducing lifting speed, pausing lifting, or maintaining alert based on the comprehensive relationship of the current lifting angle, load level, speed setting, boundary proximity, and attitude disturbance intensity during the lifting and unloading process. This improves the pertinence and safety of the lifting operation status adjustment.

[0050] This invention also provides a dynamic protection device for the stability boundary of a rear-unloading semi-trailer lifting operation, used to implement the above-mentioned method, such as... Figure 4 As shown, the device includes: The monitoring data acquisition unit is used to collect raw angular velocity and acceleration data during the lifting operation through sensors, calculate the current lifting angle, fuse the static boundary threshold and dynamic margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary under the current lifting angle, and obtain the initial monitoring dataset. The interference feature extraction unit is used to perform signal smoothing processing on the angular velocity data in the initial monitoring dataset, extract interference feature parameters, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. The overturning probability assessment unit is used to perform random disturbance simulation based on the attitude disturbance vector, generate multiple sets of future short-term attitude change trajectories, assess the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory set, and generate path probability distribution and average probability value. The intervention and adjustment unit is used to determine the risk level based on the shape of the probability distribution and the average probability value; and to generate intervention instructions in combination with the current working condition parameters to adjust the lifting operation status.

[0051] The present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0052] In summary, this invention collects raw angular velocity and acceleration data during the lifting process and integrates static boundary thresholds and dynamic margin coefficients to calculate the dynamic stability boundary at the current lifting angle in real time. This not only reflects the basic stability margin of the vehicle at different lifting stages but also incorporates actual working conditions such as load transfer, uneven ground, hydraulic fluctuations, and external disturbances into the boundary correction process, thus providing initial monitoring data that more closely reflects the actual operating conditions for subsequent risk analysis. Secondly, by smoothing the angular velocity data, effective attitude change information and high-frequency disturbance information are separated, disturbance feature parameters are extracted, and an attitude disturbance vector is constructed by combining the change in distance between state points. This allows the system to characterize the true evolution trend of the lifting posture of the cargo compartment and quantify key factors affecting instability such as cargo release, vibration impact, and boundary approach speed. Furthermore, based on the above attitude disturbance vector, a follow-up analysis is performed... The invention employs a disturbance simulation to generate multiple sets of future short-term attitude change trajectories. By combining these with dynamic stability boundaries, the cross-boundary probability of each trajectory is assessed, yielding path probability distributions and average probability values. This expands the traditional approach of judging only the current moment to a probabilistic prediction of future short-term overturning risks, improving the foresight of risk identification. Finally, the risk level is determined based on the probability distribution and average probability values. Intervention commands are generated by combining real-time lifting angle, lifting speed, load weight, and distance from the state point, enabling proactive control of operational states such as reducing lifting speed, pausing lifting, or maintaining alert. Through the coordination of these steps, the invention achieves dynamic assessment of the stability boundaries of electric rear-dump semi-trailer lifting operations, early prediction of overturning risks, and graded intervention in hazardous conditions, thereby improving the safety, timeliness of control, and accuracy of risk protection in lifting operations.

[0053] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0054] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0055] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for dynamic protection of stability boundaries during lifting operations of a rear-unloading semi-trailer, characterized in that, The method includes: Step S1: Collect raw angular velocity and acceleration data during the lifting operation using sensors, calculate the current lifting angle, fuse the static boundary threshold and dynamic margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary under the current lifting angle, and obtain the initial monitoring dataset. Step S2: Perform signal smoothing processing on the angular velocity data in the initial monitoring dataset, extract interference feature parameters, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. Step S3: Based on the attitude disturbance vector, perform random disturbance simulation to generate multiple sets of future short-term attitude change trajectories. Evaluate the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory sets, and generate the path probability distribution and average probability value. Step S4: Determine the risk level based on the shape of the probability distribution and the average probability value; generate an intervention command in conjunction with the current operating parameters to adjust the lifting operation status.

2. The method according to claim 1, characterized in that, In step S1, the initial monitoring dataset is obtained, including: Raw angular velocity and acceleration data are collected during the lifting operation using gyroscopes and accelerometers. Based on this data, the current lifting angle is calculated. A preset static boundary threshold is queried based on the current lifting angle to determine the static stability boundary at that angle. A dynamic margin coefficient is determined based on the fluctuation amplitude of real-time angular velocity and acceleration data, and this coefficient is used to adjust the static stability boundary to obtain the dynamic stability boundary. The current attitude state point is determined in real time, and the distance between the attitude state point and the dynamic stability boundary, as well as the boundary approach rate, are calculated. The raw angular velocity data, raw acceleration data, distance between the attitude state point, and boundary approach rate are integrated to obtain the initial monitoring dataset.

3. The method according to claim 1, characterized in that, In step S2, the angular velocity fluctuation sequence and the corresponding disturbance characteristic parameters are generated, including: For the angular velocity data in the initial monitoring dataset, a sliding window mean filtering method is used to smooth the signal to suppress high-frequency random interference, resulting in a smoothed angular velocity data sequence. Based on the difference between the angular velocity data sequence and the original angular velocity data sequence, an interference residual sequence is obtained. The interference amplitude distribution and frequency distribution of the interference residual sequence are analyzed to extract interference characteristic parameters. The smoothed angular velocity data sequence is then used as the angular velocity fluctuation sequence.

4. The method according to claim 1, characterized in that, In step S2, the attitude disturbance vector is constructed, including: Based on the angular velocity fluctuation sequence, the amplitude and frequency features of the angular velocity changes are extracted to characterize the fluctuation intensity and dominant frequency of the effective attitude evolution; based on the interference feature parameters, the interference intensity features are extracted to characterize the energy level and dominant frequency structure of the suppressed interference; based on the real-time tracked state point distance sequence, the rate of change of the state point distance is calculated to characterize the trend and speed of the current attitude approaching the stability boundary; after normalizing the amplitude features, the frequency features, the interference intensity features, and the rate of change of the state point distance, they are integrated to construct the attitude interference vector.

5. The method according to claim 1, characterized in that, In step S3, a set of future short-term attitude change trajectories is generated, including: Amplitude, frequency, disturbance intensity, and boundary proximity trend features are extracted from the attitude disturbance vector as simulation parameters. Using the Monte Carlo method, multiple random disturbance simulations are performed on the current attitude state based on these simulation parameters to generate multiple future short-term attitude change trajectories. For each attitude change trajectory, it is projected into a state space defined by a dynamic stability boundary, and the trajectory boundary distance between each state point corresponding to the attitude change trajectory and the dynamic stability boundary is calculated. All generated attitude change trajectories and their corresponding trajectory boundary distance sequences are correlated and integrated to form a set of future short-term attitude change trajectories used for overturning probability assessment.

6. The method according to claim 5, characterized in that, In step S3, the path probability distribution and average probability value are generated, including: Obtain the sequence of trajectory state points for each attitude change trajectory in the trajectory set over multiple sampling periods; recalculate the real-time trajectory boundary distance between each trajectory state point and the real-time dynamic stability boundary based on the real-time dynamic stability boundary provided by the dynamic boundary monitoring module; evaluate the overturning probability of each attitude change trajectory crossing the real-time dynamic stability boundary over multiple sampling periods based on the real-time trajectory boundary distance sequence corresponding to each attitude change trajectory; summarize and statistically analyze the overturning probabilities of all attitude change trajectories in the trajectory set to generate a path probability distribution; calculate the average probability value based on the path probability distribution.

7. The method according to claim 1, characterized in that, In step S4, the risk level is determined, including: Obtain the average probability value and the path probability distribution; based on the path probability distribution, extract and quantify its peak position, distribution width, and skewness features; construct a risk level determination model, wherein the average probability value is used as the basic risk level input, and the peak position, distribution width, and skewness features are used as risk distribution structure correction input; calculate using the risk level determination model and output the corresponding risk level.

8. The method according to claim 7, characterized in that, In step S4, based on the current operating parameters, an intervention command is generated to adjust the lifting operation status, including: Based on the risk level, activate the corresponding active amplitude limiting protection strategy; obtain the current operating parameters, which include at least the real-time lifting angle, the current lifting speed setpoint, the lifting load weight, the real-time status point distance, and the attitude disturbance vector; based on the risk level and the current operating parameters, dynamically determine the target intervention mode and corresponding control parameters through a preset intervention decision model, and generate intervention commands to reduce the lifting speed, pause lifting, or maintain alert; send the intervention commands to the lifting operation control execution unit to adjust the lifting operation status.

9. A dynamic protection device for the stability boundary of a rear-unloading semi-trailer lifting operation, used to implement the method as described in any one of claims 1-8, characterized in that, The device includes: The monitoring data acquisition unit is used to collect raw angular velocity and acceleration data during the lifting operation through sensors, calculate the current lifting angle, fuse the static boundary threshold and margin coefficient corresponding to the current lifting angle, calculate the dynamic stability boundary under the current lifting angle, and obtain the initial monitoring dataset. The interference feature extraction unit is used to perform signal smoothing processing on the angular velocity data in the initial monitoring dataset, extract interference feature parameters, generate an angular velocity fluctuation sequence, and construct an attitude interference vector by combining the changes in the distance between real-time acquired state points. The overturning probability assessment unit is used to perform random disturbance simulation based on the attitude disturbance vector, generate multiple sets of future short-term attitude change trajectories, assess the overturning probability of each trajectory crossing the dynamic boundary in multiple future sampling periods through the trajectory set, and generate path probability distribution and average probability value. The intervention and adjustment unit is used to determine the risk level based on the shape of the probability distribution and the average probability value; and to generate intervention instructions in combination with the current working condition parameters to adjust the lifting operation status.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the method as described in any one of claims 1-8.

Images