Joint control methods, devices, and media for preventing track derailment in electric tracked vehicles

Abstract

This invention relates to a method, device, and medium for joint control of track derailment prevention in electric tracked vehicles, and pertains to the field of tracked vehicle control technology. The method includes the following steps: S1. Acquiring multi-source sensor data from the electric tracked vehicle, including at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data based on quantum inertial sensors. S2. Performing spatiotemporal alignment and consistency correction on the multi-source sensor data, and fusing the corrected data using an adaptive filtering algorithm to obtain the vehicle's real-time fused state quantity. S3. Calculating a comprehensive risk index based on the real-time fused state quantity through multi-dimensional feature coupling analysis, and determining the current track derailment risk level. S4. Matching a preset hierarchical control strategy according to the determined track derailment risk level, generating joint control commands, and sending them to the vehicle's actuators, including a motor drive system and a mechanical tensioning system.

Description

Technical Field

[0001] This invention relates to the field of tracked vehicle control technology, and more specifically, to a combined control method, device, and medium for preventing track derailment in electric tracked vehicles. Background Technology

[0002] With the widespread application of pure electric high-speed tracked vehicles in complex terrains and high-speed maneuvering conditions, effectively preventing track derailment has become a key requirement for ensuring vehicle safety and continuous operation. Especially in high-speed driving, severe vibration, and variable environments such as mud and slopes, tracks are prone to loosening, slipping, and lateral deviation. Traditional methods relying on manual inspection and periodic tensioning are no longer sufficient to meet the requirements for real-time and precise control.

[0003] Several technical solutions have been proposed to address the aforementioned problems, primarily including: a mechanical anti-slip design that enhances lateral blocking capability by improving the internal tooth structure of the tracks; a static tensioning method that uses manually adjustable bolts and telescopic sleeves to compensate for track slack; an automatic adjustment system that uses a balance elbow angle sensor to calculate track length changes and control the tensioning cylinder; and a hydraulic synchronous control method that utilizes an electro-hydraulic proportional valve to achieve smooth speed reduction during the tensioning process. These solutions offer partial solutions from the perspectives of structure, manual adjustment, automatic feedback, and hydraulic control.

[0004] However, existing methods still have significant limitations. They mostly rely on displacement or angle sensors based on the vehicle's relative coordinate system, which are prone to cumulative drift under high speed and continuous vibration, leading to gradual distortion of the measurement reference. At the same time, each sensor usually operates independently, lacking deep fusion of multi-source information, and cannot comprehensively assess the dynamic correlation between slip ratio, terrain adhesion coefficient, and vehicle attitude. This results in a lag in the system's identification of derailment risk, making it difficult to implement precise intervention before the risk occurs. Summary of the Invention

[0005] The present invention provides a method, device, and medium for joint control of track derailment prevention in electric tracked vehicles, in order to improve at least one of the above-mentioned technical problems.

[0006] In a first aspect, the present invention provides a combined control method for preventing track derailment in an electric tracked vehicle, comprising steps S1 to S4.

[0007] S1. Acquire multi-source sensing data of the electric tracked vehicle. The multi-source sensing data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data based on quantum inertial sensors.

[0008] S2. Perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and use an adaptive filtering algorithm to fuse the corrected data to obtain the real-time fused state of the vehicle.

[0009] S3. Based on the real-time fusion state quantity, calculate the comprehensive risk index through multi-dimensional feature coupling analysis, and determine the current de-banding risk level.

[0010] S4. Based on the determined risk level of belt derailment, match the preset hierarchical control strategy, generate a joint control command and send it to the vehicle's actuator, which includes a motor drive system and a mechanical tensioning system.

[0011] As a further aspect of the present invention, S2 includes S21 to S23.

[0012] S21. Project all sensor measurement data onto the vehicle coordinate system with the quantum inertial sensor as the core, and synchronize the time.

[0013] S22. Monitor the measurement values ​​of different sensors for the same physical quantity, and trigger the calibration process when the deviation exceeds the consistency threshold.

[0014] S23. Monitor the rate of change of each measurement value in real time. When the change value exceeds the single-cycle change limit or falls outside the confidence interval for several consecutive cycles, mark the measurement value as abnormal and remove it from the current fusion cycle.

[0015] As a further aspect of the present invention, S2 also includes S24 to S26.

[0016] S24. Construct a state-dependent adaptive extended Kalman filter and define the system state vector and observation vector.

[0017] S25. Based on the real-time slip rate and terrain adhesion coefficient, dynamically construct the measurement noise covariance matrix.

[0018] S26. The Kalman gain is updated by weighting using the measurement noise covariance matrix to calculate the optimal estimated fusion state quantity.

[0019] As a further aspect of the present invention, the system state vector of S24 and observation vector The expression is as follows.

[0020] .

[0021] .

[0022] In the formula, This is the north-facing position. The location is facing east. Northbound speed. The speed is eastward. For roll angle, For pitch angle, This is the heading angle. It is an accelerometer. To achieve zero bias with the gyroscope. This indicates transpose. This represents the speed of the left drive wheel. This represents the speed of the right drive wheel. Linear acceleration measured by the IMU. The angular velocity measured by the IMU. For track tension. This refers to the amount of downward sag.

[0023] As a further aspect of the present invention, the measurement noise covariance matrix of S25 is constructed as a dynamically weighted matrix. .

[0024] .

[0025] In the formula, It is a diagonal matrix. The standard deviation of the measurement noise for the left wheel speed. The standard deviation of the measurement noise for the right wheel speed. The standard deviation of the acceleration measurement noise. The standard deviation of the measurement noise for angular velocity. The standard deviation of the tension measurement noise. This represents the standard deviation of the measurement noise of the sag sensor.

[0026] .

[0027] In the formula, This is the glide confidence function. This refers to the real-time slip ratio. The terrain adhesion coefficient is estimated using LiDAR / visual methods. It is a function with a natural base. for The weight. for The weight.

[0028] When the vehicle is in a condition with a high slip ratio or a low coefficient of adhesion, Approaching 0, the noise weight of wheel speed measurement approaches infinity.

[0029] As a further aspect of the present invention, S26 specifically includes the following steps.

[0030] The Kalman gain is updated using the measured noise covariance matrix with weights, and then the updated Kalman gain is used... The observation vectors from multiple sensors With the system's state prediction value By fusing the data, we can obtain the optimal state estimate for the current moment. This represents the number of iteration steps. For a moment The observation vector. For a moment The estimated state vector.

[0031] Monitoring new information sequences .

[0032] like And duration If the event is determined to be an instantaneous mechanical impact, no status update will be performed to prevent false L1 level warnings. To observe the residual for the tension sensor. This is the preset residual threshold.

[0033] As a further aspect of the present invention, S3 includes the following steps.

[0034] Select track tension Track sagging slip ratio Roll angle and terrain adhesion coefficient As a key feature quantity.

[0035] Monitor the numerical values ​​of the aforementioned key characteristic quantities and their first-order rate of change within a preset time window.

[0036] If the track tension is found to be lower than the design preload value, the track sag exceeds the pitch ratio threshold, or the slip rate exceeds the slip threshold, the risk level of track derailment will be increased according to the degree of deviation.

[0037] If the roll angle is detected to be approaching the vehicle's static stability limit and the rate of change of lateral acceleration is suddenly increased, it is judged as a risk of roll instability, and the risk level of derailment is increased.

[0038] As a further aspect of the present invention, the risk levels of tape detachment include L1 to L5.

[0039] Level L1 is a warning alert: the criteria for determination are as follows: The control actions are: audio-visual cues and increased sampling rate.

[0040] Level L2 is an auxiliary compensation: the determination criteria are as follows: Threshold. Lateral angle 3°~4°. Control action: tension compensation. Activate the motor torque smoothing algorithm, implement differential torque compensation, and eliminate stress concentration caused by sudden changes in driving force.

[0041] Level L3 is a strategic intervention: the determination criteria are as follows: Threshold. The terrain adhesion coefficient is lower than the preset value. The control action is as follows: execute torque vector control, actively distribute the driving force of the left and right motors, and reduce the lateral shear force of the tracks. At the same time, speed limiting and path replanning are triggered.

[0042] Level 4 is a system takeover: the criteria for determination are: The lateral angle shows an upward trend of 6°~7°. The control action is: the system takes over power and decelerates.

[0043] Level L5 is for safety protection: the criteria for determination are: Threshold. Lateral angle > 8°. Control actions are: emergency braking, power failure, forced path transfer, and data retention.

[0044] In the formula, For track tension, For designing preload value, For slip ratio, This refers to the amount of track sagging.

[0045] As a further aspect of the present invention, S1 specifically includes the following steps.

[0046] Using a quantum inertial sensor installed at the vehicle's center of mass, the vehicle's three-axis attitude angles, angular velocities, and linear accelerations in the global coordinate system are collected as the main reference benchmark for data fusion.

[0047] The track tension sensor and track sag sensor are used to collect the real-time track tension and the vertical distance from the vehicle body reference point to the track surface, respectively.

[0048] The rotational speeds of the left and right drive wheels are collected using wheel speed encoders, and the slip ratio is calculated in conjunction with the vehicle speed.

[0049] Using lidar or cameras to collect terrain point cloud and road surface texture information in front of the vehicle, the road surface type can be identified and the terrain adhesion coefficient can be estimated.

[0050] As a further aspect of the present invention, the combined control method for preventing desquamation also includes a closed-loop optimization step.

[0051] After executing the control command, key feedback data of the vehicle is collected in real time, including changes in tension force and vehicle attitude response after execution.

[0052] Calculate the deviation between the feedback data and the expected response.

[0053] When the deviation exceeds the set threshold, the threshold parameters for risk assessment and the weight parameters in the fusion algorithm are corrected using an online adaptive algorithm.

[0054] Secondly, the present invention provides a joint control device for preventing track derailment in an electric tracked vehicle, which includes a data acquisition module, a fusion module, a risk module, and a control module.

[0055] The acquisition module is used to acquire multi-source sensor data of the electric tracked vehicle. The multi-source sensor data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data based on quantum inertial sensors.

[0056] The fusion module is used to perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and to fuse the corrected data using an adaptive filtering algorithm to obtain the real-time fused state of the vehicle.

[0057] The risk module is used to calculate a comprehensive risk index based on the real-time fused state quantity through multi-dimensional feature coupling analysis, and to determine the current de-banding risk level.

[0058] The control module is used to match a preset hierarchical control strategy according to the determined risk level of belt derailment, generate joint control commands and send them to the actuators of the vehicle, the actuators including a motor drive system and a mechanical tensioning system.

[0059] Thirdly, the present invention provides an electric tracked vehicle, which includes a processor, a memory, and a computer program stored in the memory. The computer program can be executed by the processor to implement a joint control method for preventing track derailment of an electric tracked vehicle as described in any paragraph of the first aspect.

[0060] Fourthly, the present invention provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a combined control method for preventing track derailment of an electric tracked vehicle as described in any paragraph of the first aspect.

[0061] By adopting the above technical solution, the present invention can achieve the following technical effects:

[0062] This invention fundamentally solves the problem of cumulative drift of traditional sensors under high-speed and vibration conditions by introducing a quantum inertial sensor as a high-precision absolute spatiotemporal reference, ensuring the long-term stability and reliability of the sensing source and enabling the system to accurately capture minute changes in the track status.

[0063] Furthermore, this invention employs a multi-source information deep fusion algorithm based on working condition adaptation. By dynamically adjusting the weights of each sensor, unreliable data sources are automatically shielded under extreme working conditions, greatly enhancing the robustness and fault tolerance of the system under complex road conditions.

[0064] Most importantly, this invention represents a leap from passive response to proactive prevention. The system can perform short-term trend prediction based on fused high-confidence state variables, identify risks in advance, and implement tiered, refined interventions, thereby transforming post-event remediation into pre-event prevention and significantly improving the safety of high-speed driving.

[0065] Meanwhile, this invention deeply integrates the advantages of a pure electric platform, innovatively adopting a collaborative control architecture of "electronic control fine-tuning priority and mechanical tensioning assistance," utilizing motor torque vector compensation to achieve millisecond-level rapid response, effectively overcoming the lag problem of traditional hydraulic systems. Furthermore, the system possesses closed-loop self-optimization capabilities, enabling it to correct parameters online based on execution feedback, adaptively compensating for vehicle wear and environmental changes, ensuring optimal monitoring and control performance throughout its entire lifecycle. Attached Figure Description

[0066] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the specific embodiments of the present invention will be briefly introduced below. It should be understood that the following drawings only show some specific embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0067] Figure 1 It is a sensor layout for tracked systems designed for derailment monitoring.

[0068] Figure 2 This is a flowchart of a combined control method for preventing band degeneration.

[0069] Figure 3 This is a flowchart of the risk assessment logic. Detailed Implementation

[0070] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention.

[0071] Example 1, please refer to Figures 1 to 2 The first embodiment of the present invention provides a joint control method for preventing track derailment of an electric tracked vehicle, which includes steps S1 to S4.

[0072] S1. Acquire multi-source sensing data of the electric tracked vehicle. The multi-source sensing data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data based on quantum inertial sensors.

[0073] S2. Perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and use an adaptive filtering algorithm to fuse the corrected data to obtain the real-time fused state of the vehicle.

[0074] S3. Based on the real-time fusion state quantity, calculate the comprehensive risk index through multi-dimensional feature coupling analysis, and determine the current de-banding risk level.

[0075] S4. Based on the determined risk level of belt derailment, match the preset hierarchical control strategy, generate a joint control command and send it to the vehicle's actuator, which includes a motor drive system and a mechanical tensioning system.

[0076] Figure 2 The flowchart of the joint control method for preventing desquamation mainly illustrates the working principle and process of the system.

[0077] After the system starts up and runs, it first collects real-time data on vehicle attitude, track tension, speed and environmental information through multi-source sensors.

[0078] The collected raw data then enters the data fusion and processing stage, where advanced algorithms are used to integrate, calibrate, and extract features to form an accurate and unified state estimate of vehicle dynamics and the environment.

[0079] Based on the fused multi-source state data, the system's risk assessment and trend prediction module is activated. By comprehensively judging key characteristics such as track tension, sag, slip ratio, vehicle attitude changes, and ground adhesion conditions, it determines the current derailment risk level in real time and classifies it into five progressive levels, from L1 to L5. Different levels will automatically trigger corresponding control strategies, realizing a step-by-step intervention from warning, compensation, intervention, takeover to emergency protection.

[0080] Ultimately, all decision commands converge into the execution control loop, driving the vehicle's actuators (such as drive motors, tensioning mechanisms, and braking systems) to respond. The real-time vehicle status after execution is then re-inputted into the system as feedback information, thus forming a continuously optimizing and self-correcting intelligent closed-loop control process that fundamentally ensures the vehicle's stable and safe operation in complex dynamic environments.

[0081] This invention fundamentally solves the problem of cumulative drift of traditional sensors under high-speed and vibration conditions by introducing a quantum inertial sensor as a high-precision absolute spatiotemporal reference, ensuring the long-term stability and reliability of the sensing source and enabling the system to accurately capture minute changes in the track status.

[0082] Preferably, S1 specifically includes S11 to S14.

[0083] S11. Using a quantum inertial sensor installed at the vehicle's center of mass, the vehicle's three-axis attitude angles, angular velocities, and linear accelerations in the global coordinate system are collected as the main reference benchmark for data fusion.

[0084] S12. Using track tension sensor and track sag sensor, collect the real-time tension of the track and the vertical distance from the vehicle body reference point to the track surface, respectively.

[0085] S13. Use wheel speed encoders to collect the rotational speeds of the left and right drive wheels, and calculate the slip ratio in combination with the vehicle speed.

[0086] S14. Use lidar or cameras to collect terrain point cloud and road surface texture information in front of the vehicle to identify road surface type and estimate terrain adhesion coefficient.

[0087] Sensor mounting structure for high vibration environments, if Figure 1 As shown. Considering the high-frequency vibration characteristics of tracked vehicles, all critical sensors (especially the sag sensor and IMU) are not directly rigidly connected to the vehicle body. A "double-layer floating damping mount" is preferred. The inner layer of the mount uses a high-damping viscoelastic material (such as silicone rubber) to encapsulate the sensor core, filtering out high-frequency mechanical noise (>500Hz). The outer layer of the mount uses a metal spring-damper composite structure connected to the frame, isolating low-frequency large-amplitude swaying caused by road impacts. This physical mounting structure, combined with the back-end "consistency correction algorithm," eliminates the impact of vibration on measurement accuracy from both physical and software perspectives. Figure 1 The sensors used in this embodiment and their installation locations are shown.

[0088] The control method in this embodiment requires the use of multiple sensors. A quantum inertial sensor serves as a high-precision absolute reference sensor, providing global attitude, angular velocity, and acceleration data. Other sensors (such as tension sensors, sag sensors, and wheel speed encoders) act as specialized monitoring units, providing track-related state variables. The system employs a redundant sensing network design and an adaptive fusion algorithm to ensure accuracy and robustness under different operating environments.

[0089] In the system architecture of this invention, the quantum inertial sensor serves as a high-precision absolute reference, providing high-precision, low-drift reference information such as vehicle attitude, angular velocity, and linear acceleration. As a final backup sensor, it continues to provide necessary navigation and attitude data when other sensors fail or malfunction, ensuring the stability and fault tolerance of the system.

[0090] The sensors used in this embodiment, their functions, and the data collected are as follows.

[0091] Quantum inertial sensors and IMUs are mounted at the center (or center of mass) of the vehicle to monitor the vehicle's three-axis attitude, angular velocity, and linear acceleration in real time, serving as the primary reference coordinate system for the fusion algorithm. The collected data represents the vehicle's attitude: pitch angle. Roll angle Angular velocity Linear acceleration .

[0092] The track tension sensor, installed near the lower track, is used to monitor the track tension. The collected data is the real-time tension. .

[0093] Wheel speed encoders, mounted on the main drive wheel and road wheels, monitor the speed difference of the tracks to help calculate the slip ratio and determine if track slippage is occurring. The collected data is the speed of the left drive wheel. Right drive wheel speed Calculate the slip ratio .

[0094] LiDAR sensors and cameras, mounted on the front or roof of the vehicle, are used to perceive ground features and obstacles. The collected data includes terrain point clouds, slope, and surface type (mapped to friction coefficient via semantics / rules). ).

[0095] The track sag sensor is installed on the chassis side of the vehicle body in the middle of the upper track return section, with the probe pointing vertically downwards towards the inner surface of the track. The collected data is obtained by measuring the vertical distance from a vehicle body reference point to the track surface in real time using a non-contact ranging principle (such as laser time-of-flight method or ultrasonic echo method). This is used to calculate the degree of sag and looseness of the tracks.

[0096] During data fusion, high-precision attitude, angular velocity, and acceleration data provided by quantum inertial sensors serve as the system's spatiotemporal reference. Data from other sensors (such as tension and sag) are fused after spatiotemporal alignment and coordinate system transformation. Through an adaptive fusion algorithm, the system dynamically adjusts the data weights of each sensor according to real-time operating conditions, ensuring the accuracy and robustness of data fusion.

[0097] Preferably, S2 includes S21 to S23.

[0098] S21. Project all sensor measurement data onto the vehicle coordinate system with the quantum inertial sensor as the core, and synchronize the time.

[0099] S22. Monitor the measurement values ​​of different sensors for the same physical quantity, and trigger the calibration process when the deviation exceeds the consistency threshold.

[0100] S23. Monitor the rate of change of each measurement value in real time. When the change value exceeds the single-cycle change limit or falls outside the confidence interval for several consecutive cycles, mark the measurement value as abnormal and remove it from the current fusion cycle.

[0101] The consistency correction process involves the system first synchronizing the time and uniformly projecting the coordinates of each sensor output, establishing a unified measurement reference frame in the vehicle coordinate system. When the deviation of any sensor measurement relative to other similar measurements exceeds a consistency threshold... When the error spreads due to differences in installation location and hysteresis, the consistency correction process is automatically triggered to eliminate the spread of errors caused by differences in installation location and hysteresis, and to prevent local anomalies from affecting the overall judgment.

[0102] The anomaly detection and rejection process is as follows: The system monitors the rate of change of each measurement in adjacent sampling periods in real time. When the change value exceeds the allowable limit, the system will reject the anomaly. , or continuous When a sampling period falls outside the confidence interval defined based on historical statistics, the measurement is marked as abnormal and temporarily removed from the fusion process of the current period to avoid misjudgment caused by transient noise, sensor jitter, or short-term occlusion.

[0103] The confidence level adaptive update operation is as follows: the system updates the confidence level based on the historical fluctuations of the sensor. Real-time noise level The measurement weights are automatically updated based on the current environmental conditions (such as strong light, smoke, dust, and loose road surfaces). Reduce the weight of mechanical sensors under conditions of severe vibration and significant mechanical impact. Reduce the weight of cameras under conditions of low visibility or high dynamic glare. Reduce the weight of magnetic correlation measurements in areas with magnetic interference.

[0104] This mechanism ensures that the system does not rely on a single measurement source, thereby improving overall robustness and fault tolerance at the physical level.

[0105] The symbols used in this embodiment include: This is the consistency deviation threshold, used to determine whether multiple sensors are trustworthy to each other. It is a single-cycle variation limiter used to eliminate instantaneous spike noise interference. It serves as a criterion for the number of consecutive anomalies, used to determine whether an anomaly is persistent rather than sporadic. This represents historical fluctuations and is used to characterize long-term reliability. This represents the instantaneous noise level, used to characterize current short-term reliability. It is a dynamic weighting factor used to control the dominant source of the fusion result.

[0106] Preferably, S2 also includes S24 to S26.

[0107] S24. Construct a state-dependent adaptive extended Kalman filter and define the system state vector and observation vector.

[0108] S25. Based on the real-time slip rate and terrain adhesion coefficient, dynamically construct the measurement noise covariance matrix.

[0109] S26. The Kalman gain is updated by weighting using the measurement noise covariance matrix to calculate the optimal estimated fusion state quantity.

[0110] To address the issue of accuracy divergence in traditional filtering algorithms under conditions of tracked vehicle height slippage and abrupt terrain changes, this system constructs a "State-Related Adaptive Extended Kalman Filter (Adaptive EKF) algorithm." Unlike general algorithms, this invention maps the track tension state and terrain adhesion coefficient into the noise matrix of the filter. The specific implementation is as follows.

[0111] First, the system state vector and observation vector are established. In this embodiment, vehicle kinematic parameters and sensor errors are selected as the state vector, and multi-source sensor data are selected as the observation vector.

[0112] .

[0113] .

[0114] In the formula, This is the state vector. This is the north-facing position. The location is facing east. Northbound speed. The speed is eastward. For roll angle, For pitch angle, This is the heading angle. It is an accelerometer. To achieve zero bias with the gyroscope. This indicates transpose. This is the observation vector. This represents the speed of the left drive wheel. This represents the speed of the right drive wheel. Linear acceleration measured by the IMU. The angular velocity measured by the IMU. For track tension. This refers to the amount of downward sag.

[0115] The key innovation of this embodiment lies in the observation vector. Introduced left / right drive wheel speeds Track tension and sagging amount The preferred coordinate system for position and velocity is the Northeast-Northeast (ENU) geographic coordinate system, but it is not limited to this and can also be converted according to the vehicle coordinate system.

[0116] Next, the terrain-adaptive measurement noise matrix is ​​constructed. This is the core of this embodiment. Traditional EKF assumes that the measurement noise covariance matrix is ​​constant. They cannot adapt to muddy and slippery conditions.

[0117] In this embodiment, the measurement noise covariance matrix is ​​constructed as a dynamically weighted matrix. :

[0118] .

[0119] In the formula, It is a diagonal matrix. The standard deviation of the measurement noise for the left wheel speed. The standard deviation of the measurement noise for the right wheel speed. The standard deviation of the acceleration measurement noise. The standard deviation of the measurement noise for angular velocity. The standard deviation of the tension measurement noise. This represents the standard deviation of the measurement noise of the sag sensor.

[0120] .

[0121] In the formula, This is the glide confidence function. This refers to the real-time slip ratio. The terrain adhesion coefficient is estimated using LiDAR / visual methods. It is a function with a natural base. for The weight. for The weight.

[0122] The system uses a camera to capture images of the road in front of the vehicle, and performs real-time semantic segmentation using a pre-trained deep neural network to identify road surface materials (such as cement, dry soil, loose sand, wet mud, ice, snow, etc.). Then, it uses a pre-set material-adhesion coefficient prior table built into the system to determine the type of road surface (e.g., cement road surface). loose sandy land muddy road ).

[0123] When the vehicle gets stuck in the mud ( (reduction) or violent slippage ( Increase, for example When >20%, the function The variance approaches 0, resulting in a corresponding wheel speed measurement noise variance. It tends toward infinity.

[0124] Technical effect: The filter will automatically "cut off" the trust in the wheel speed sensor and instead rely entirely on the high-precision IMU (quantum inertial sensor) for integral calculation, so as to maintain accurate estimation of attitude and trajectory under slipping conditions and avoid the system misjudging that the vehicle is moving at high speed due to falsely high wheel speed.

[0125] Finally, state update and residual check.

[0126] The Kalman gain is updated using the measured noise covariance matrix with weights, and then the updated Kalman gain is used... The observation vectors from multiple sensors With the system's state prediction value By fusing the data, we can obtain the optimal state estimate for the current moment. This process effectively reduces prediction uncertainty and outputs high-confidence vehicle status information, providing a reliable basis for subsequent risk assessment. This represents the number of iteration steps.

[0127] The system monitors the news sequence To determine if the sensor is malfunctioning. .

[0128] The threshold determination logic is: if (i.e., the observed tension values ​​are significantly different from the model predictions), and the duration The system determines that it is an instantaneous mechanical impact and does not update the status to prevent false L1 level warnings. To observe the residual for the tension sensor. This is the preset residual threshold.

[0129] Preferably, S3 includes S31 to S34.

[0130] S31. Select track tension Track sagging slip ratio Roll angle and terrain adhesion coefficient As a key feature quantity.

[0131] S32. Monitor the numerical values ​​of the above key characteristic quantities and their first-order rate of change within a preset time window.

[0132] S33. If the track tension is found to be lower than the design preload value, the track sag exceeds the pitch ratio threshold, or the slip rate exceeds the slip threshold, the risk level of track derailment will be increased according to the degree of deviation.

[0133] S34. If the roll angle is detected to be close to the vehicle's static stability limit and the rate of change of lateral acceleration is suddenly increased, it is judged as a risk of roll instability and the risk level of derailment is increased.

[0134] This embodiment is based on a risk assessment and hierarchical control system using multi-dimensional feature fusion. This system will control track tension. Track sagging slip ratio Multi-source attitude data (from quantum inertial sensors) and terrain adhesion coefficient Multidimensional features are fused using an extended Kalman filter (EKF) to generate a highly robust comprehensive risk index. (Among them, the comprehensive risk index can be understood as the risk level of tape shedding).

[0135] Real-time assessment of chain derailment risk no longer relies solely on fixed empirical thresholds for a single feature. Instead, it employs a five-level progressive strategy, from L1 to L5, by comparing the risk level with a dynamically adjusted safety limit after adaptive correction. This hierarchical assessment system allows the system to intervene early in the risk assessment process, prioritizing the use of torque vectoring, a feature unique to pure electric systems, for proactive fine-tuning. This achieves efficient coordinated control with "electronic control first, mechanical tensioning assistance," thereby transforming post-event defense into proactive pre-event suppression.

[0136] Track tension: Real-time monitoring of tension .when ( When the preload value is not specified (as per design preload value), the track is deemed loose, the contact stiffness is insufficient, and the risk is significantly increased.

[0137] Track sagging: Real-time monitoring of sagging .when ( When the track pitch is within a certain range, it is determined that the track is too loose, triggering a risk warning.

[0138] Track slip ratio: The slip ratio is calculated based on the wheel speed and the vehicle body speed. .when At that time, it was determined that the adhesion was severely reduced and there was a risk of slippage and detachment.

[0139] Vehicle attitude: Real-time monitoring of lateral acceleration change rate and roll angle. When the roll angle approaches the vehicle's static stability limit (approximately 6°~8°) or there is a sudden increase in lateral acceleration, it is determined that the risk of roll instability has increased sharply.

[0140] Ground adhesion conditions: Identifying ground type and estimating friction coefficient through sensing data. On surfaces with extremely low adhesion (such as mud) or particularly high viscosity, corresponding environmental risk signs will be triggered.

[0141] In addition to static thresholds, the system also monitors the changing trends of the aforementioned state variables within a short time window. When the first-order rate of change of any parameter exceeds a set slope threshold, the risk level can be upgraded in advance, achieving "trend-based early warning." To avoid false triggering by transient noise, the judgment process must meet the condition of consistent duration or number of consecutive samples.

[0142] Once the risk level is determined, the system adopts a "five-level progressive intelligent protection strategy," implementing control interventions from weak to strong to ensure a safety baseline while avoiding unnecessary impact on task continuity. The corresponding levels and control strategies are as follows.

[0143] Level L1 is a warning alert characterized by slight deviations and a stable trend. The criteria for determination are: The control actions are: audio-visual cues and increased sampling rate.

[0144] Level L2, with auxiliary compensation, is characterized by a controllable, mild anomaly. The criteria for determination are: Threshold. Lateral angle 3°~4°. Control action: Micro-tension compensation. Activate motor torque smoothing algorithm, implement small-amplitude differential torque compensation, and eliminate stress concentration caused by sudden changes in driving force.

[0145] Level L3 is characterized by moderate abnormality and a clear trend, indicating a need for strategic intervention. The criteria for determination are: Threshold. The terrain adhesion coefficient is lower than the preset value. The control action is as follows: execute torque vector control, actively distribute the driving force of the left and right motors, and significantly reduce the lateral shear force of the tracks. At the same time, speed limiting and path replanning are triggered.

[0146] Specifically, the L3 level control action is as follows: based on the difference in slip ratio between the left and right tracks and the vehicle's tilting tendency, the output torque of the left and right motors is actively adjusted. The drive torque on the side with the higher slip ratio is reduced, while the drive torque on the other side is increased, generating a yaw torque to correct the vehicle's attitude and reduce the lateral shear force on the tracks.

[0147] Level L4 is a system takeover status characterized by significant anomalies and a continuously increasing risk. The criteria for determination are: The lateral angle shows an upward trend of 6°~7°. The control action is as follows: the system takes over power, smoothly decelerates, and quickly compensates.

[0148] Level L5 is characterized by an extreme hazardous state and critical instability. The determination criteria are: Threshold. Lateral angle > 8°. Control actions are: emergency braking, power failure, forced path transfer, and data retention.

[0149] In the formula, For track tension, For designing preload value, For slip ratio, This refers to the amount of track sagging.

[0150] The key state parameters used in this embodiment include: pitch angle and roll angle, used to identify roll tendency and center of gravity drift; track tension, used to determine the tendency for track slippage; track sag, used to determine the possibility of the track leaving the sprocket; lateral slip ratio, used to determine drive matching and insufficient adhesion; and ground slope / adhesion, used to determine the passability of the working condition.

[0151] This embodiment incorporates four key quantities—tension, sagging, slippage, and topography—into the same discrimination framework to form an engineering-based risk discrimination rule.

[0152] Specifically, a short-term trend extrapolation and look-ahead control strategy based on adaptive EKF fusion states is adopted. The system uses high-confidence state variables (such as true tension, slip ratio, and terrain coefficient) calculated by adaptive extended Kalman filter (EKF) as benchmark inputs to construct a vehicle dynamics trend model, and performs forward extrapolation of the risk of derailment within a limited time domain (such as the next 2-3 seconds). This allows the system to move beyond the passive mode of 'alarm after exceeding the limit' and achieve proactive prevention of 'intervention before the trend deteriorates'. A hierarchical control strategy is designed and a closed-loop feedback is formed, balancing safety and operational efficiency. All thresholds and weights are calibrable parameters to adapt to different vehicle models and operating conditions.

[0153] Preferably, the joint control method for preventing desquamation further includes closed-loop optimization steps S51 to S53.

[0154] S51. After executing the control command, collect key feedback data of the vehicle in real time. The feedback data includes the change in tension force and the vehicle body attitude response after execution.

[0155] S52. Calculate the deviation between the feedback data and the expected response.

[0156] S53. When the deviation exceeds the set threshold, the threshold parameter for risk assessment and the weight parameter in the fusion algorithm are corrected using an online adaptive algorithm.

[0157] To ensure the system maintains stable risk assessment accuracy and control response capabilities throughout the entire vehicle lifecycle, this invention constructs an optimization mechanism oriented towards "continuous learning – adaptive compensation – closed-loop evolution." After the control command is executed, the system collects key feedback data (including tension, sag, lateral acceleration, etc.) in real time and compares it with the expected response deviation. When the deviation exceeds a set threshold, the internal discrimination parameters and control amplitude are automatically corrected, achieving long-term adaptation to different vehicle states, wear levels, and environmental conditions, allowing the system performance to gradually approach the optimal value.

[0158] This embodiment employs a "two-stage calibration system": the initial calibration includes static zero-point calibration, sensor bias correction, and the acquisition of dynamic response curves of key state quantities under typical terrain conditions to establish threshold benchmarks and state confidence models. The online adaptive calibration involves the system continuously accumulating historical state distributions during operation, calculating sensor drift trends and environmental change characteristics, and making small adaptive adjustments to the discrimination threshold and state fusion weights to compensate for long-term drift caused by structural fatigue and transmission wear, ensuring consistent risk discrimination capabilities over the long term.

[0159] When the short-term data statistical characteristics deviate from the long-term benchmark by more than a set threshold, the system will trigger a periodic correction process to dynamically reconstruct the experience mapping, realize "in-use evolution", and improve life cycle stability.

[0160] To ensure the sustainable operation of the above functions in real-world complex environments, the present invention has designed the following engineering support measures.

[0161] Redundancy and fault tolerance strategies: Symmetrical redundancy is implemented for key sensors (such as tension and ground sensing), and a confidence cross-validation strategy is introduced. In the event of single-path distortion or failure, the system automatically reduces the weight of the abnormal path or blocks the abnormal path to avoid misjudgment caused by a single point of failure.

[0162] Task priority scheduling: The discrimination and control module is deployed on the onboard real-time computing platform, and assigns the highest priority to safety actions such as tensioning mechanism control and emergency braking. The command issuance link has an independent real-time channel and bandwidth guarantee to avoid control delays caused by communication congestion.

[0163] Data retention and system evolution: The system has a built-in event black box that records triggering conditions, risk levels, action strategies and feedback effects, enabling remote parameter tuning, OTA expansion and algorithm version upgrades, providing verifiable evidence for subsequent model iterations.

[0164] Human-machine collaborative design: When reaching Level 3 or above, the system provides current risks and suggested actions in a multimodal manner using sound, light, and voice, while retaining human intervention rights. In the event of conflict between human and machine commands, safety commands take priority to avoid secondary risks caused by misoperation.

[0165] In addition, to cope with persistently harsh environments, this invention adopts high-protection-level devices (≥IP65) and a wide temperature drift compensation design, and sets an environmental characteristic adaptive weight attenuation strategy at the software level, so that the system can maintain stable discrimination ability under conditions such as sandstorms, rain, snow, and strong light.

[0166] Through a closed-loop design encompassing "real-time feedback correction, online adaptation, parameter evolution, multi-source fault tolerance, and human-machine collaboration," this invention can maintain stable risk perception accuracy and control execution reliability during long-term operation, effectively suppressing the accumulation of track derailment risks and sudden instability.

[0167] This embodiment employs a novel hardware and software integrated hardware and software architecture for a pure electric high-speed tracked vehicle derailment prevention and monitoring system, centered on a quantum sensor and featuring "high-precision benchmark + distributed sensing + multi-level closed-loop control". This architecture uses a quantum inertial sensor at the vehicle's center of mass as the absolute spatiotemporal benchmark for the entire vehicle, and deploys specialized monitoring units for tension, sag, wheel speed, and environmental perception (LiDAR / camera) in a distributed manner, forming a "quantum benchmark + multi-source specialized" layout. Furthermore, a high-frequency vibration damping physical safety layer is incorporated. For high-speed tracked vehicle vibration, a "double-layer suspended vibration damping mounting base" (inner layer of silicone rubber filters high-frequency noise, outer layer of metal spring isolates low-frequency impact) protects the core sensor, ensuring the reliability of physical-level data. The system architecture supports the "torque vector compensation" technology unique to pure electric vehicles, achieving efficient collaborative control with "electronic control fine-tuning priority and mechanical tensioning assistance," forming an electronic control-priority collaborative architecture.

[0168] like Figure 1 As shown, the derailment prevention joint control method for an electric tracked vehicle in this embodiment can be implemented by a tracked vehicle derailment prevention system architecture integrating quantum inertial sensors. The tracked vehicle derailment prevention system architecture includes: a high-precision inertial measurement module (quantum / fiber optic level), a multi-source state acquisition module (tension / sag / wheel speed / environment), a hierarchical control decision module, and a double-layer suspension physical protection structure. It utilizes the "zero drift" characteristic of quantum sensors to achieve an ultra-long-term, high-confidence navigation reference, and employs an architecture design for real-time online correction of other sensor data in complex terrain. The method for fusing quantum sensors with other sensors (spatiotemporal fusion, reference frame) is also included.

[0169] The derailment prevention joint control method uses a quantum inertial sensor as the reference coordinate system and employs a consistency correction algorithm to synchronize the time and spatial projection of each sensor, eliminating installation errors and signal hysteresis. A dynamic weighting matrix based on operating conditions is also used. The core innovation lies in constructing a terrain-adaptive measurement noise matrix and introducing a sliding confidence function. Using slip ratio and terrain adhesion coefficient Dynamically adjust the weights of each sensor. Automatic trust cut-off mechanism: When severe slippage or muddy conditions are detected, the algorithm automatically reduces the weight of the wheel speed encoder, relying entirely on the quantum sensor for integral calculation, preventing falsely inflated wheel speeds from causing misjudgments.

[0170] This embodiment proposes a sensor fusion method based on state-dependent adaptive extended Kalman filtering (Adaptive EKF). This method utilizes absolute attitude parameters provided by quantum sensors to dynamically map the noise matrix along with the vehicle slip ratio and ground friction coefficient, and then calculates terrain slope compensation and dynamic tilt correction for non-contact measurements such as track sag sensors.

[0171] Key technical points: L1~L5 five-level gradient discrimination criterion: tension drooping amount slip ratio The risk index is integrated with attitude and terrain coefficients. The judgment criteria include not only static values ​​but also their first-order rate of change (trend warning). This composite risk judgment criterion, which shifts from "static threshold" to "multi-dimensional trend + pre-deduction," enables more accurate judgment and control.

[0172] By utilizing the fused high-confidence state variables, proactive inferences can be performed within 2-3 seconds to achieve preemptive intervention before risks escalate, rather than reactive alerts. Coupling criteria: It can identify chain-like instability logic caused by the combined effects of decreased tension, increased sag, and vehicle roll (such as roll angle approaching 6°-8°).

[0173] This embodiment also proposes a multi-dimensional feature fusion-based risk classification criterion for track derailment, which integrates the effective track slack amount corrected by quantum sensors, dynamic tension force, and the rate of change of vehicle motion trend. The closed-loop control strategy based on risk-triggered classification includes, but is not limited to, L2-level micro-tension compensation, L3-level torque vector distribution, and L5-level emergency braking and power failure protection.

[0174] This embodiment overcomes the accuracy bottleneck of the sensing reference and eliminates accumulated drift errors. Traditional monitoring schemes are limited by the performance of MEMS inertial navigation or displacement sensors, and are prone to zero-point drift under high-speed and severe vibration conditions, causing the calculation results to become invalid as the operating time increases. This invention introduces a quantum inertial sensor as an absolute spatiotemporal reference. With its "zero drift and high sensitivity" characteristics, it provides an extremely stable attitude and motion reference for the entire vehicle. This allows the system to accurately calculate minute changes in track sag and tension even under long-term and complex terrain conditions, fundamentally solving the problem of inaccurate sensing sources.

[0175] This embodiment achieves deep fusion of multi-source information based on adaptive working conditions, enhancing system robustness. Unlike the "island-like" operation of sensors in existing technologies, this invention employs an adaptive extended Kalman filter (EKF) algorithm based on dynamic weight adjustment. The system can perceive slip rate and terrain adhesion coefficient in real time and dynamically adjust the "trust level" of each sensor's data in the fusion process. For example, under severe slip conditions, the algorithm automatically reduces the weight of wheel speed data and instead relies on quantum benchmarks for integral calculation. This mechanism effectively avoids misjudgments caused by single sensor failure or extreme environmental interference, significantly improving the system's robustness under complex road conditions such as mud and sand.

[0176] This embodiment shifts from "passive feedback" to "proactive prediction," establishing a multi-level refined prevention and control system. Existing technologies mostly rely on post-event adjustments, meaning mechanical compensation is only implemented after the risk of track derailment occurs. This invention, through trend analysis of fused high-dimensional data, constructs L1-L5 level gradient judgment criteria, enabling the system to predict track movement trends and instability risks 2-3 seconds in advance. This "proactive prediction" allows the system to intervene at the initial stage of risk, transforming "post-event remediation" into "pre-event prevention," significantly improving safety during high-speed driving.

[0177] This embodiment deeply integrates the advantages of a pure electric platform, achieving highly efficient collaborative control at the millisecond level. This invention fully leverages the fast response of the electronic control system in pure electric tracked vehicles, innovatively proposing a combination of "torque vector compensation + mechanical tensioning assistance." The system employs a collaborative architecture. When faced with the risk of momentary instability, it can first suppress track slippage or deviation through millisecond-level motor torque fine-tuning, and then supplement it with a mechanical tensioning mechanism for long-term position compensation. This deep integration of electronic control and mechanics solves the problem of lag in the response of traditional hydraulic systems, achieving a more efficient and energy-saving track derailment prevention effect.

[0178] This embodiment possesses closed-loop evolution capability, ensuring monitoring efficiency throughout the entire lifecycle. The system introduces an adaptive parameter optimization mechanism, capable of online correction of the judgment threshold based on feedback from the actuators. This means the system can sense long-term changes in physical characteristics such as track wear and structural loosening and self-adjust, ensuring optimal anti-track derailment monitoring accuracy throughout the vehicle's lifecycle, significantly reducing subsequent manual maintenance costs.

[0179] Example 2: The present invention provides a joint control device for preventing track derailment in an electric tracked vehicle, which includes a data acquisition module, a data fusion module, a risk module, and a control module.

[0180] The acquisition module is used to acquire multi-source sensor data of the electric tracked vehicle. The multi-source sensor data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data based on quantum inertial sensors.

[0181] The fusion module is used to perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and to fuse the corrected data using an adaptive filtering algorithm to obtain the real-time fused state of the vehicle.

[0182] The risk module is used to calculate a comprehensive risk index based on the real-time fused state quantity through multi-dimensional feature coupling analysis, and to determine the current de-banding risk level.

[0183] The control module is used to match a preset hierarchical control strategy according to the determined risk level of belt derailment, generate joint control commands and send them to the actuators of the vehicle, the actuators including a motor drive system and a mechanical tensioning system.

[0184] Example 3: This invention provides an electric tracked vehicle, which includes a processor, a memory, and a computer program stored in the memory. The computer program can be executed by the processor to implement a combined control method for preventing track derailment in an electric tracked vehicle as described in any paragraph of Example 1.

[0185] Example 4: This invention provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a combined control method for preventing track derailment of an electric tracked vehicle as described in any paragraph of Example 1.

[0186] Obviously, the embodiments described above are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0187] In the several embodiments provided in this invention, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0188] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0189] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a 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, electronic device, 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, random access memory, magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0190] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms unless the context clearly indicates otherwise.

[0191] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0192] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0193] The terms "first" and "second" used in the embodiments are merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first" and "second" can be interchanged in a specific order or sequence where permitted. It should be understood that the objects distinguished by "first" and "second" can be interchanged where appropriate so that the embodiments described herein can be implemented in an order other than those illustrated or described herein.

[0194] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A joint control method for preventing track derailment in an electric tracked vehicle, characterized in that, Include: S1. Acquire multi-source sensor data of the electric tracked vehicle. The multi-source sensor data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data. Among them, the vehicle attitude data is based on a quantum inertial sensor. S2. Perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and use an adaptive filtering algorithm to fuse the corrected data to obtain the real-time fused state of the vehicle. S3. Based on the real-time fusion state quantity, calculate the comprehensive risk index through multi-dimensional feature coupling analysis, and determine the current de-banding risk level; S4. Based on the determined risk level of belt derailment, match the preset hierarchical control strategy, generate a joint control command and send it to the vehicle's actuator, which includes a motor drive system and a mechanical tensioning system; S2 includes: S21. Project all sensor measurement data onto the vehicle coordinate system with the quantum inertial sensor as the core, and synchronize the time. S22. Monitor the measurement values ​​of different sensors for the same physical quantity, and trigger the calibration process when the deviation exceeds the consistency threshold. S23. Monitor the rate of change of each measurement value in real time. When the change value exceeds the single-cycle change limit or falls outside the confidence interval for several consecutive cycles, mark the measurement value as abnormal and remove it from the current fusion cycle. S24. Construct a state-dependent adaptive extended Kalman filter and define the system state vector and observation vector; S25. Based on the real-time slip rate and terrain adhesion coefficient, dynamically construct the measurement noise covariance matrix; S26. The Kalman gain is updated by weighting using the measurement noise covariance matrix to calculate the optimal estimated fusion state quantity.

2. The method for joint control to prevent track derailment in an electric tracked vehicle according to claim 1, characterized in that, S24's system state vector and observation vector The expression is as follows; ； ； In the formula, The location is north-facing; The location is eastward; Northbound speed; The speed is eastward; For roll angle, For pitch angle, For heading angle; For accelerometers; To achieve zero bias with the gyroscope; Indicates transpose; This refers to the speed of the left drive wheel; This refers to the speed of the right drive wheel; Linear acceleration measured by the IMU; Angular velocity measured by the IMU; For track tension; This refers to the amount of downward sag. The measurement noise covariance matrix of S25 is constructed as a dynamically weighted matrix. ; ； In the formula, It is a diagonal matrix; The standard deviation of the measurement noise for the left wheel speed; The standard deviation of the measurement noise for the right wheel speed; The standard deviation of the acceleration measurement noise; The standard deviation of the noise in the measurement of angular velocity; The standard deviation of the tension measurement noise; The standard deviation of the measurement noise of the sag sensor; ； In the formula, It is a glide confidence function; Real-time slip ratio; Terrain adhesion coefficient estimated by LiDAR / visual method; Functions with natural base; for The weights; for The weights; When the vehicle is in a condition with a high slip ratio or a low coefficient of adhesion, Approaching 0, the noise weight of wheel speed measurement approaches infinity; S26 specifically includes: The Kalman gain is updated using the measured noise covariance matrix with weights, and then the updated Kalman gain is used... The observation vectors from multiple sensors With the system's state prediction value By fusing the data, we can obtain the optimal state estimate for the current moment. ; This represents the number of iteration steps. For a moment The observation vector; For a moment State vector estimate; Monitoring new information sequences ; ; like And duration If the incident is determined to be an instantaneous mechanical impact, no status update will be performed to prevent false L1 level warnings. To observe the residuals for the tension sensor; This is the preset residual threshold.

3. The method for joint control to prevent track derailment in an electric tracked vehicle according to claim 1, characterized in that, S3 include: Select track tension Track sagging slip ratio Roll angle and terrain adhesion coefficient As a key feature quantity; Monitor the numerical values ​​of the above key characteristic quantities and their first-order rate of change within a preset time window; If the track tension is found to be lower than the design preload value, the track sag exceeds the pitch ratio threshold, or the slip rate exceeds the slip threshold, the risk level of track derailment will be increased according to the degree of deviation. If the roll angle is detected to be approaching the vehicle's static stability limit and the rate of change of lateral acceleration is suddenly increased, it is judged as a risk of roll instability, and the risk level of derailment is increased.

4. The method for joint control to prevent track derailment in an electric tracked vehicle according to claim 1, characterized in that, The risk levels for tape delamination range from L1 to L5; Level L1 is a warning alert: the criteria for determination are as follows: ; ; The control actions are: audio-visual cues and increased sampling rate; Level L2 is an auxiliary compensation: the determination criteria are as follows: ; ; Threshold; Lateral angle 3°~4°; Control action: tension compensation; Activate motor torque smoothing algorithm, implement differential torque compensation, and eliminate stress concentration caused by sudden changes in driving force; Level L3 is a strategic intervention: the determination criteria are as follows: ; ; Threshold; terrain adhesion coefficient is lower than the preset value; control actions are: execute torque vector control, actively distribute the driving force of the left and right motors, reduce the lateral shear force of the tracks; and trigger speed limit and path replanning at the same time. Level 4 is a system takeover: the criteria for determination are: ; ; The lateral angle shows an upward trend of 6°~7°; the control action is: the system takes over power and decelerates. Level L5 is for safety protection: the criteria for determination are: ; ; Threshold; Lateral angle > 8°; Control actions are: emergency braking, power failure, forced path transfer, and data retention; In the formula, For track tension, For designing preload value, For slip ratio, This refers to the amount of track sagging.

5. The method for joint control to prevent track derailment in an electric tracked vehicle according to claim 1, characterized in that, S1 specifically includes: Using a quantum inertial sensor installed at the vehicle's center of mass, the vehicle's three-axis attitude angles, angular velocities, and linear accelerations in the global coordinate system are collected as the main reference benchmark for data fusion. Using track tension sensors and track sag sensors, the real-time tension of the track and the vertical distance from the vehicle body reference point to the track surface are collected, respectively. The wheel speed encoder is used to collect the rotational speed of the left and right drive wheels, and the slip ratio is calculated in combination with the vehicle speed; Using lidar or cameras to collect terrain point cloud and road surface texture information in front of the vehicle, in order to identify road surface type and estimate terrain adhesion coefficient; The joint control method for preventing desquamation also includes a closed-loop optimization step: After executing the control command, key feedback data of the vehicle is collected in real time, including the change in tension force and the vehicle body attitude response after execution; Calculate the deviation between the feedback data and the expected response; When the deviation exceeds the set threshold, the threshold parameters for risk assessment and the weight parameters in the fusion algorithm are corrected using an online adaptive algorithm.

6. A combined control device for preventing track derailment in an electric tracked vehicle, characterized in that, For implementing the joint control method for preventing track derailment of an electric tracked vehicle as described in any one of claims 1 to 5; The combined control device for preventing desquamation includes: The acquisition module is used to acquire multi-source sensor data of the electric tracked vehicle. The multi-source sensor data includes at least vehicle attitude data, track tension data, track sag data, drive wheel speed data, and environmental perception data. Among them, the vehicle attitude data is based on a quantum inertial sensor. The fusion module is used to perform spatiotemporal alignment and consistency correction on the multi-source sensor data, and to fuse the corrected data using an adaptive filtering algorithm to obtain the real-time fused state of the vehicle. The risk module is used to calculate a comprehensive risk index based on the real-time fused state variables through multi-dimensional feature coupling analysis, and to determine the current de-banding risk level. The control module is used to match a preset hierarchical control strategy according to the determined risk level of belt derailment, generate joint control commands and send them to the actuators of the vehicle, the actuators including a motor drive system and a mechanical tensioning system.

7. An electric tracked vehicle, characterized in that, It includes a processor, a memory, and a computer program stored in the memory; the computer program can be executed by the processor to implement a joint control method for preventing derailment of an electric tracked vehicle as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a combined control method for preventing derailment of an electric tracked vehicle as described in any one of claims 1 to 5.

Images