Railway comprehensive detection speed control method and device based on distance and antiskid compensation

Abstract

The invention discloses a railway comprehensive detection vehicle speed control method based on distance and anti-slip compensation, which comprises the following steps: starting a railway comprehensive detection vehicle, and loading operation parameters; the method comprises the steps of periodically collecting operation data and geometric parameters; obtaining a load torque estimated value, an adhesion coefficient, a creep rate and an optimal creep rate through a state observation algorithm based on the operation data; obtaining original mileage data through periodic calculation based on the operation data and the geometric parameters; the mileage data is updated through a wheel diameter compensation strategy and a slip compensation strategy based on the operation data and the geometric parameters, the current wheel diameter value and the calibrated mileage data are updated through an absolute calibration strategy, and a mileage compensation coefficient is obtained through curve data query based on the mileage position of the railway comprehensive detection vehicle to obtain final mileage data; and vehicle speed control parameters are adjusted based on the final mileage data, the adhesion coefficient and the optimal creep rate, a motor control instruction is output through an anti-skid control algorithm based on the working condition state, and wheel rail skid is prevented.

Description

Technical Field

[0001] This invention belongs to the field of underground railway transportation and safety inspection technology in coal mines, and particularly relates to a control method and device for a comprehensive railway inspection vehicle based on distance and anti-slip compensation. Background Technology

[0002] With the trend of intelligent and unmanned development in underground coal mines, mine track inspection vehicles are key equipment for efficient and safe auxiliary transportation and condition inspection. However, due to the harsh and complex working conditions underground, existing inspection vehicles still face significant technical bottlenecks in the core aspect of speed control.

[0003] Firstly, in terms of accurate mileage measurement and positioning, traditional methods employ a single wheel-axle pulse counting method. The core accuracy of this method heavily relies on the accuracy of the wheel rolling radius and the pure rolling state of wheel-rail contact. The slippery rails, coal sludge adhesion, switches, and frequent starts, stops, and brakes in the underground environment easily cause wheel slippage or spinning, while wear from long-term wheel operation also leads to dynamic changes in the actual rolling radius. These factors directly introduce unpredictable pulse counting errors, which accumulate over time and distance, making it difficult to meet the stringent requirements of long-distance, high-precision detection tasks. Although some technologies attempt to use interval calibration methods, the lack of underground GPS signals, the susceptibility of physical markers to damage and contamination, and the excessive spacing between markers result in few calibration opportunities and low reliability, making it impossible to achieve continuous, reliable, high-precision positioning and error closed-loop correction.

[0004] Secondly, regarding the smooth control of operating speed, underground tracks present complex nonlinear conditions such as gradient changes, curves, track joints, and random abrupt changes in wheel-rail adhesion conditions caused by water spray and coal dust. Existing speed control strategies mostly employ traditional PID algorithms with fixed parameters. These algorithms cannot adapt to changing road surface adhesion coefficients and load disturbances, easily leading to significant speed fluctuations and even drive wheel slippage during sudden drops in adhesion. This not only affects the stability and safety of the testing equipment itself but also makes it difficult for the onboard precision testing instruments to obtain high-quality testing data due to unstable speed. A more prominent problem is that in the current technology system, the speed control subsystem and the mileage measurement subsystem typically operate independently and in isolation, failing to establish an effective collaborative mechanism. On the one hand, the rich dynamic information generated during speed control cannot be effectively used to assist and optimize the identification and compensation of mileage errors; on the other hand, high-precision mileage and location information cannot proactively serve the pre-adaptive control of speed. This information silo phenomenon prevents both core functions from achieving optimal performance, hindering the overall improvement of the intelligence level of underground testing vehicles. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a railway comprehensive inspection vehicle speed control method and device based on distance and anti-slip compensation. It acquires the internal operating state through a state observer, ensures external positioning accuracy through a mileage compensation chain, and finally outputs safe and optimized execution commands through an adaptive anti-slip controller, achieving precise positioning of the railway comprehensive inspection vehicle and smooth drive with pre-slip prevention. This application constructs three PID controllers: a gain adaptive PID, an actual operating speed balance point correction PID, and a speed estimation derivative, respectively achieving rapid correction with large gain for large errors, fine smoothing with small gain for small errors, improving traction efficiency and anti-slip performance during acceleration, and providing a smooth speed signal. Through wheel diameter compensation strategy, slippage compensation strategy, absolute calibration strategy, and curve data query correction, this application eliminates distance calculation errors caused by wheel wear, corrects mileage data deviations caused by slippage or lock-up, and prevents long-term accumulated errors, ensuring the absolute accuracy of the railway comprehensive inspection vehicle. It also corrects mileage errors generated by the vehicle on curved tracks such as curves and slopes, ultimately obtaining high-precision, reliable, and multi-verified final mileage data, providing a solid data foundation for subsequent precise control and task execution. By improving the three-stage recursive observer to quantify the adhesion coefficient of the track's slipperiness and the creep rate of the difference between wheel speed and vehicle speed, the optimal creep rate is obtained through the least squares method, finding a golden balance point for the railway comprehensive inspection vehicle to drive without slipping. This application also introduces a balance point correction algorithm based on the optimal creep rate and a vector control algorithm based on load torque feedforward compensation to achieve both fast and stable start-up and acceleration, eliminating idling and predictively increasing torque output during sudden load changes such as uphill climbs. Ultimately, this enables the railway comprehensive inspection vehicle to perceive road conditions and actively avoid entering dangerous slipping states, ensuring smooth driving. A smooth control process is achieved by solving for the optimal motor torque command using a model predictive control algorithm. The gain in the three-stage recursive observer is adaptively adjusted according to the error magnitude, and the model parameters for calculating the optimal creep rate using the recursive least squares method with an adaptive forgetting factor satisfy the railway comprehensive inspection vehicle's selection of the optimal operating mode under various working conditions.

[0006] The first aspect of the present invention provides a railway integrated inspection speed control method based on distance and anti-skid compensation, comprising: Start the railway comprehensive inspection vehicle and load operating parameters; By periodically collecting operational data and geometric parameters; Based on the aforementioned operational data, the load torque estimate, adhesion coefficient, creep rate, and optimal creep rate are obtained through a state observation algorithm. Based on the operating data and the geometric parameters, the original mileage data is obtained through periodic calculation; based on the operating data and the geometric parameters, the mileage data is updated through wheel diameter compensation strategy and slippage compensation strategy; the current wheel diameter value is updated and the mileage data is calibrated through absolute calibration strategy; based on the mileage position of the railway comprehensive inspection vehicle, the mileage compensation coefficient is obtained through curve data query to obtain the final mileage data. Based on the final mileage data, the adhesion coefficient, and the optimal creep rate, the vehicle speed control parameters are adjusted. Based on the operating conditions, motor control commands are output through an anti-slip control algorithm to prevent wheel-rail slippage. The anti-slip control algorithm includes: in the acceleration phase, an equilibrium point correction algorithm based on the optimal creep rate is used to output motor torque; in the speed control phase, a vector control algorithm based on load torque feedforward compensation is used.

[0007] Preferably, the step of obtaining the load torque estimate further includes: Based on the motor's three-phase current, three-phase voltage, and speed, the q-axis current and rotor electrical angle are obtained through Clark transformation, Park transformation, and low-pass filtering. The actual motor speed at the current moment is calculated based on the pulse count of the operating data. Based on the actual motor speeds at the current moment and the previous two moments, the estimated motor speed at the previous moment is used to obtain the speed estimation error, differential feedforward gain, and error differential term through adaptive differential gain adjustment of motor speed; Based on the q-axis current, the actual motor speed at the current moment, the estimated motor speed at the previous moment, the estimated load torque at the previous moment, the speed estimation error, and the error differential term, the estimated motor speed and the estimated load torque at the current moment are obtained through the recursive formula of the improved three-stage recursive observer.

[0008] Preferably, the step of obtaining the load torque estimate by improving the recursive formula of the three-stage recursive observer further includes: The motor torque compensation term is obtained based on the direct-axis current, q-axis current, number of motor pole pairs, differential feedforward gain, and error differential term. The calculation expression is as follows: In the formula, The electromagnetic torque of the motor. For differential feedforward gain, This is the differential term of the error; A nonlinear error feedback function and a motor speed prediction model are constructed to obtain the predicted motor speed. The expression for the nonlinear error feedback function is as follows: In the formula, To estimate the rotational speed error, a preset error threshold is set. K is a shape parameter. p1 K p2 These are the first proportional gain parameter, the second proportional gain parameter, and K. d1 K d2 These are the first differential gain parameter and the second differential gain parameter, respectively. The formula for calculating the predicted motor speed is: In the formula, This is the estimated motor speed at time k-1. For the collection period, For rotational inertia, The damping coefficient; Based on the predicted motor speed and the actual motor speed, the torque imbalance value is obtained through torque imbalance detection. The calculation expression is as follows: In the formula, This refers to the actual speed of the motor; The estimated load torque value is obtained based on the adaptive forgetting factor, torque imbalance value, and motor torque compensation term. The calculation expression is as follows: In the formula, An adaptive forgetting factor. The mutation correction term is calculated as follows: Kcorr is the dynamic parameter of the observer.

[0009] Preferably, the step of obtaining the adhesion coefficient further includes: Based on the estimated load torque at the current moment, the operating parameters are converted through torque balance relationship to obtain wheel-rail adhesion. Based on the wheel-rail adhesion and the vertical load on the wheel, the adhesion coefficient is constructed using the Coulomb friction model, and the calculation expression is as follows: In the formula, n is the mechanical transmission ratio, N is the vertical load on the wheel, and D is the effective diameter of the wheel.

[0010] Preferably, the step of obtaining the creep rate further includes: Based on the pulse count from the operational data, the effective wheel diameter from the geometric parameters, and the acquisition period, the linear velocity of the wheel is obtained, and the calculation expression is: In the formula, The acquisition period is defined as PPR, which represents the number of pulses per revolution of the operational data. These are the pulse count data for the current and previous moments in the operating parameters, respectively, and the effective diameter of the wheel; The creep rate is obtained using the Carter formula based on the linear velocity of the wheel and the actual running speed in the running data. The calculation expression is as follows: In the formula, This refers to the actual operating speed.

[0011] Preferably, the step of obtaining the optimal creep rate further includes: Based on adhesion coefficient-creep rate data, a function relating adhesion coefficient and creep rate is constructed using the recursive least squares method. The specific process is as follows: The function relating adhesion coefficient and creep rate is constructed, and its expression is: In the formula, , respectively, are the adhesion coefficient and creep rate in the adhesion coefficient-creep rate data, and a, b, and c are the model parameters; The adaptive forgetting factor is calculated based on the function of the adhesion coefficient and the creep rate, and the function of the adhesion coefficient and the creep rate is updated accordingly. The initial value of the optimal creep rate is obtained by calculating the peak creep rate and dynamic filtering constraints based on the model parameters. The optimal creep rate is obtained based on the initial value of the optimal creep rate, the operating condition, and the creep rate at the current moment. Specifically, if the operating condition is a stable state, the optimal creep rate is obtained through a smoothing tracking algorithm, calculated as follows: In the formula, These are the optimal creep rate at the current moment, the optimal creep rate at the previous moment, the initial value of the optimal creep rate, and the tracking coefficient. If the working condition is a sudden change, the model parameters are further determined to obtain the optimal creep rate.

[0012] Preferably, the step of further determining the model parameters to obtain the optimal creep rate further includes: if a > 0, If a < 0, further differentiate the function of adhesion coefficient and creep rate and calculate the current derivative value du / dλ = 2a*λ(k) + b; if du / dλ > 0, then... ,otherwise ,in This is the step size obtained based on the operating condition.

[0013] Preferably, the step of updating mileage data through wheel diameter compensation strategy and slippage compensation strategy further includes: Updated mileage data is obtained based on the pulse count, wheel diameter measurement, and number of pulses per revolution of the aforementioned operating data; A mileage compensation factor is constructed based on the operating conditions, and the mileage data is updated by weighting the mileage compensation factor and the mileage data.

[0014] Preferably, the step of adjusting the vehicle speed control parameters based on the final mileage data, the adhesion coefficient, and the optimal creep rate further includes: Based on the final mileage data, the estimated speed is obtained through a first-order low-pass differential filter, and the calculation expression is as follows: In the formula, This represents the final mileage data for the current and previous moments. The actual running speed at the previous moment. For the collection period, These are the filter coefficients; Based on the actual operating speed and estimated load torque, online optimization is performed using a model predictive control algorithm to solve for the reference torque. The construction of the model predictive control algorithm includes: constructing discretized vehicle dynamics equations that satisfy system dynamic constraints, wherein the expressions for the system dynamic constraints are: In the formula, To predict speed, M represents the mass of the railway inspection vehicle, D represents the effective diameter of the wheel, and T represents the speed. m (k) represents the current motor torque. Here is the estimated load torque at the current moment, and n is the mechanical transmission ratio. The optimization objective is to track the running resistance of the railway comprehensive inspection vehicle, and to perform error tracking of the running speed and motor torque within the sampling period. Constraints include ensuring the motor torque is within a preset threshold range. , The predicted speed sequence value is obtained based on the motor torque through system dynamic constraints; where , respectively, creep rate and optimal creep rate, For safety reasons, Where is the adhesion coefficient, and N is the vertical load on the wheel. To predict adhesion; The reference torque is corrected based on the acceleration state of the railway comprehensive inspection vehicle: when in the acceleration phase, a balance point correction algorithm based on the optimal creep rate is used for correction; when in the speed control phase, a feedforward compensation correction based on the load torque is used; the corrected torque is then limited and smoothed before outputting control commands.

[0015] A second aspect of the present invention provides a railway integrated detection speed control device based on distance and anti-skid compensation, comprising: Initialization module: used to start the railway comprehensive inspection vehicle and load operating parameters; Data acquisition module: used to periodically collect operational data and geometric parameters; State observation module: used to obtain the load torque estimate, adhesion coefficient, creep rate and optimal creep rate based on the operating data through the state observation algorithm; Data preprocessing module: used to obtain raw mileage data through periodic calculations based on the running data and the geometric parameters; Data compensation and calibration module: used to update mileage data based on the running data and geometric parameters through wheel diameter compensation strategy and slippage compensation strategy; used to update the current wheel diameter value and calibrate the mileage data through absolute calibration strategy; used to obtain the final mileage data by querying curve data based on the mileage position of the railway comprehensive inspection vehicle; Speed ​​control and anti-slip module: used to adjust speed control parameters based on the final mileage data, the adhesion coefficient and the optimal creep rate, and output motor control commands based on the working condition through the anti-slip control algorithm to prevent wheel-rail slippage.

[0016] By employing the above technical solutions, this invention has the following advantages and positive effects compared with existing technologies: Through wheel diameter compensation strategies, slippage compensation strategies, absolute calibration strategies, and curve data query correction, this invention eliminates distance calculation errors caused by wheel wear, corrects mileage data deviations caused by slippage or lock-up, and prevents long-term cumulative errors, ensuring the absolute accuracy of the railway comprehensive inspection vehicle. It also corrects mileage errors generated by the vehicle on curved tracks such as curves and slopes, ultimately obtaining high-precision, reliable, and multi-verified final mileage data, providing a solid data foundation for subsequent precise control and task execution. By improving the three-stage recursive observer to quantify the adhesion coefficient of the track's slipperiness and the creep rate of the difference between wheel speed and vehicle speed, the optimal creep rate is obtained through the least squares method, finding a golden balance point for the railway comprehensive inspection vehicle to drive without slipping. This application also introduces a balance point correction algorithm based on the optimal creep rate and a vector control algorithm based on load torque feedforward compensation to achieve both fast and stable start-up and acceleration, eliminating idling and predictively increasing torque output during sudden load changes such as uphill climbs. Ultimately, this enables the railway comprehensive inspection vehicle to perceive road conditions and actively avoid entering dangerous slipping states, ensuring smooth driving. A smooth control process is achieved by solving for the optimal motor torque command using a model predictive control algorithm. The gain in the three-stage recursive observer is adaptively adjusted according to the error magnitude, and the model parameters for calculating the optimal creep rate using the recursive least squares method with an adaptive forgetting factor satisfy the railway comprehensive inspection vehicle's selection of the optimal operating mode under various working conditions. Attached Figure Description

[0017] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is the main flowchart of the railway integrated inspection speed control method based on distance and anti-skid compensation in this invention. Detailed Implementation

[0018] Detailed Implementation The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description and claims. It should be noted that the drawings are in a very simplified form and use non-precise ratios, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0020] First Embodiment See Figure 1 The first aspect of the present invention provides a railway integrated inspection speed control method based on distance and anti-skid compensation, comprising: S100: Start the railway comprehensive inspection vehicle and load operating parameters; S200: Collects operational data and geometric parameters periodically; S300: Based on operational data, load torque estimates, adhesion coefficient, creep rate, and optimal creep rate are obtained through a state observation algorithm. S400: Based on operating data and geometric parameters, the original mileage data is obtained through periodic calculations; based on operating data and geometric parameters, the mileage data is updated through wheel diameter compensation strategy and slippage compensation strategy; the current wheel diameter value and calibrated mileage data are updated through absolute calibration strategy; based on the mileage position of the railway comprehensive inspection vehicle, the mileage compensation coefficient is obtained through curve data query to obtain the final mileage data. S500: Adjusts vehicle speed control parameters based on final mileage data, adhesion coefficient, and optimal creep rate. Outputs motor control commands based on operating conditions through an anti-slip control algorithm to prevent wheel-rail slippage. The anti-slip control algorithm includes: using a balance point correction algorithm based on optimal creep rate to output motor torque during acceleration, and using a vector control algorithm based on load torque feedforward compensation during speed control.

[0021] The PID in this application is an adaptive regulator, where P is the proportional regulator, which corrects the difference between the guessed value and the actual value; I is the integral regulator, which generates a correction value to compensate for the difference if there is a persistent deviation between the guessed value and the actual value; and D is the regulator that corrects the difference that is increasing rapidly.

[0022] The railway comprehensive inspection vehicle of this application is equipped with an incremental photoelectric encoder, a Hall current sensor, a bus voltage sensor, a laser rangefinder, an observer, and a controller. The incremental photoelectric encoder is used to collect pulse counts and pulse frequency, which are used to measure the motor rotor angle and motor speed, respectively. The number of pulses per revolution is obtained through the specifications of the incremental photoelectric encoder. The Hall current sensor collects the three-phase current of the motor. The DC bus voltage is used, and the three-phase voltage of the motor is reconstructed by combining the switching state of the inverter. The laser rangefinder periodically updates the effective wheel diameter. The observer obtains the motor's moment of inertia. The controller processes the collected data to obtain a mileage position-compensation coefficient fitting curve. The mileage position-curve fitting curve can be modified through a human-machine interface to generate an updated mileage position-curve fitting curve. The controller analyzes the motor's speed decay curve under different operating conditions and obtains the damping coefficient of the railway comprehensive inspection vehicle by fitting the decay interval.

[0023] The S100 actual operating speed step mainly realizes the loading of operating parameters and driving the railway comprehensive inspection vehicle. The operating parameters include wheel diameter (in meters), mechanical transmission ratio (in dimensionless variable), moment of inertia (in kg·m²), damping coefficient (in N·m·s / rad), torque constant (in N·m / A), and curve data of mileage position / mileage compensation coefficient (in meters / dimensionless parameter). S200: Periodically collects operating data and geometric parameters; the operating data includes at least pulse count (in units), pulse frequency (in units kHz), actual motor speed (in units rad), three-phase motor current (in units A), number of pulses per revolution (in units pulse / revolution), three-phase motor voltage (in units V), and actual operating speed (m / s); the geometric parameters include at least the actual diameter of the wheels of the railway comprehensive inspection vehicle. S300: Wherein S310: The step of obtaining the load torque estimate further includes: S311: The input motor three-phase current is used to obtain the α-axis current component and the β-axis current component through the equal amplitude Clark transform formula. The rotor electrical angle is obtained based on the pulse count, the number of pulses per revolution, and the frequency multiplication factor. The calculation expression is: N represents the pulse count, M represents the frequency multiplication factor, and PPR represents the number of pulses per revolution. Based on the current components along the α-axis and β-axis, and the rotor electrical angle, the direct-axis and q-axis currents in the rotating coordinate system are obtained through Park transformation. A digital notch filter and a first-order low-pass smoothing algorithm are used to suppress PWM switching harmonics during the q-axis current processing. The actual motor speed at the current moment is obtained through sliding window processing. Specifically, the length of the window is calculated as follows: , Let be the actual motor speed at the previous moment. The expression for calculating the motor speed at the current moment is: .

[0024] S312: Based on the actual motor speeds at the current moment and the previous two moments, the estimated motor speed from the previous moment is used to obtain the speed estimation error, differential feedforward gain, and error derivative term through adaptive differential gain adjustment of motor speed. Specifically, if the absolute value of the speed estimation error is less than a preset error threshold, then... , In the formula These are, respectively, the differential gain, the minimum value of the differential error, the maximum value of the differential error, the speed estimation error, and the preset error threshold. This is the differential feedforward gain. If the absolute value of the speed estimation error is greater than or equal to the preset error threshold, then... , In the formula These are the first attenuation coefficient and the second attenuation coefficient, respectively. The error differential term is calculated based on the differential feedforward gain, differential gain, speed estimation error, and difference value. The calculation expression is: In the formula The central difference value is calculated using the following expression: .

[0025] S313: Based on the q-axis current, the current actual motor speed, the previous estimated motor speed, the previous estimated load torque, the speed estimation error, and the error differential term, the estimated motor speed and the estimated load torque at the current moment are obtained through the recursive formula of an improved three-stage recursive observer. The steps further include: S313-1: The motor torque compensation term is obtained based on the direct-axis current, q-axis current, number of motor pole pairs, differential feedforward gain, and error differential term. The calculation expression is as follows: In the formula, The electromagnetic torque of the motor. For differential feedforward gain, This is the differential term of the error; S313-2: Construct a nonlinear error feedback function and a motor speed prediction model to obtain the predicted motor speed. The expression for the nonlinear error feedback function is: In the formula, To estimate the rotational speed error, a preset error threshold is set. K is a shape parameter. p1 K p2 These are the first proportional gain parameter, the second proportional gain parameter, and K. d1 K d2 These are the first differential gain parameter and the second differential gain parameter, respectively. S313-3: The formula for calculating the predicted motor speed is: In the formula, This is the estimated motor speed at time k-1. For the collection period, For rotational inertia, The damping coefficient; S313-4: Based on the predicted motor speed and the actual motor speed, the torque imbalance value is obtained through torque imbalance detection. The calculation expression is as follows: In the formula, This refers to the actual speed of the motor; S313-5: The estimated load torque value is obtained based on the adaptive forgetting factor, torque imbalance value, and motor torque compensation term. The calculation expression is as follows: In the formula, An adaptive forgetting factor. For mutation correction terms, the calculation expression for mutation correction terms is: Kcorr represents the observer's dynamic parameters. The differential gain and attenuation coefficient are dynamically adjusted based on the error magnitude to achieve smoother steady-state operation and faster dynamic response. The nonlinear error feedback function in step S313-2 provides superior dynamic performance compared to linear feedback. Torque imbalance detection and abrupt change correction terms in S313-4 and 5 enable rapid response to load abrupt changes, such as adhesion abrupt changes. This process not only estimates the load but also measures the wheel-rail tangential force in real time. This is the foundation for achieving non-contact adhesion measurement.

[0026] Wherein S320: the step of obtaining the adhesion coefficient further includes: S321: Based on the current load torque estimate and operating parameters, wheel-rail adhesion is obtained through torque balance conversion. This step converts the load torque of the motor shaft into the tangential force of the wheel-rail contact surface, achieving non-contact adhesion measurement. Load torque includes all resistance torques acting on the wheel (such as torques corresponding to friction, slopes, curves, and air resistance) and the inertial torque required for vehicle acceleration. In the vehicle transmission system, the torque output by the motor is transmitted to the wheels through the transmission device, becoming the driving torque of the wheels. This driving torque is used to overcome the load torque and generates angular acceleration according to Newton's second law. The tangential force between the wheel and the rail, i.e., adhesion, is the force that drives the vehicle forward. When the wheel is not slipping, there is a direct relationship between adhesion and wheel driving torque, where wheel-rail adhesion is generally equal to the load torque estimate multiplied by the efficiency value / 0.5 * effective wheel diameter.

[0027] S322: The adhesion coefficient is constructed based on wheel-rail adhesion and wheel vertical load using the Coulomb friction model. The calculation expression is as follows: In the formula, n is the mechanical transmission ratio, N is the vertical load on the wheel, and D is the effective diameter of the wheel.

[0028] The adhesion coefficient quantifies the degree of adhesion by converting motor torque to wheel torque, then to wheel-rail adhesion, and finally back to the adhesion coefficient. The S313 observer employs adaptive gain, nonlinear feedback, and abrupt change correction. This means it can respond quickly when the load changes drastically due to adhesion abrupt changes. Upon receiving this rapidly changing signal, the S320 can calculate the sudden drop or rise in the adhesion coefficient μ in real time, providing fundamental data and timeliness for control system adjustments. The filtering of PWM harmonics and smoothing of speed in the S313 ensure the dynamics and smoothness of the signal, avoiding control malfunctions caused by signal noise. Because the S313 achieves sensorless estimation of load torque, it enables adhesion measurement. The entire adhesion evaluation process does not rely on any additional force sensors, greatly improving system reliability and maintainability while reducing costs.

[0029] Preferably, the step of obtaining the creep rate further includes: The linear velocity of the wheel is obtained based on the pulse count from the operational data, the effective diameter of the wheel from the geometric parameters, and the acquisition period. The calculation expression is as follows: In the formula, The acquisition period is defined as PPR, which represents the number of pulses per revolution of the data processing unit. These are the pulse count data for the current and previous moments in the operating parameters, and the effective diameter of the wheel; The creep rate is obtained using the Carter formula based on the linear velocity of the wheel and the actual running speed in the running data. The calculation expression is as follows: In the formula, This refers to the actual operating speed.

[0030] Creep rate is a key parameter in the wheel-rail dynamics of a railway comprehensive inspection vehicle, affecting traction, braking efficiency, and operational safety. Real-time creep rate calculation is achieved by directly acquiring pulse counts and the effective wheel diameter combined with actual operating speed. Load torque estimates are obtained through observation at actual operating speed (S310), adhesion coefficients are obtained through state quantification assessment at actual operating speed (S320), and motion detection is combined at S330 to achieve comprehensive, highly reliable, and real-time perception of the wheel-rail adhesion state. These steps quantify the power and slippage degree of the railway comprehensive inspection vehicle, providing a basis for dynamically adjusting the load torque to achieve stable and safe control.

[0031] Wherein S330: Preferably, the step of obtaining the optimal creep rate further includes: S331: Based on adhesion coefficient-creep rate data, a function for constructing the adhesion coefficient and creep rate data is built using the recursive least squares method. The specific process is as follows: Construct the function for the adhesion coefficient and creep rate, with the expression: In the formula, Here, represents the adhesion coefficient and creep rate in the adhesion coefficient-creep rate data, and a, b, and c are model parameters, respectively. The relationship between the adhesion coefficient and creep rate is not linear, but rather follows a hump curve that first rises and then falls. The peak point (the point of maximum adhesion coefficient) is the target pursued by the control system. Compared to approximating with a straight line or piecewise linear function, the quadratic model can capture the nonlinear characteristics of the curve with higher fidelity, especially near the peak point, resulting in a significant improvement in accuracy.

[0032] S332: Calculate the adaptive forgetting factor based on the function of adhesion coefficient and creep rate, and update the function of adhesion coefficient and creep rate; to enable the model to quickly adapt to changes in track surface conditions (such as transitioning from dry to wet sections), a forgetting factor λ is introduced into the recursive least squares method. f The smaller the forgetting factor, the higher the weight the model places on new data, and the faster it tracks changes, but it is also more sensitive to noise. This factor itself can also be adaptively adjusted according to changes in operating conditions.

[0033] S333: Based on model parameters, the initial value of the optimal creep rate is obtained by calculating the peak creep rate and dynamic filtering constraints; the expression for calculating the initial value of the optimal creep rate is: Feasibility verification is performed on the initial value of the optimal creep rate: 1) Determine whether the initial value λpeak(k) of the optimal creep rate is within the creep rate threshold range, such as 2) Determine whether the change range between the initial value of the optimal creep rate and the optimal creep rate at the previous moment is within the range threshold range 3) Calculate the confidence value of the optimal creep rate. If the confidence value is lower than the confidence threshold, the initial value of the optimal creep rate is the optimal creep rate at the previous moment; otherwise, it is kept.

[0034] S334: The optimal creep rate is obtained based on the initial value of the optimal creep rate, the operating condition, and the creep rate at the current moment. Specifically, if the operating condition is a stable state, the optimal creep rate is obtained through a smoothing tracking algorithm, calculated as follows: In the formula, These are the current optimal creep rate, the previous optimal creep rate, the initial value of the optimal creep rate, and the tracking coefficient, respectively. If the operating condition is abrupt, further judgment of model parameters to obtain the optimal creep rate includes: if a > 0, If a < 0, further differentiate the function of adhesion coefficient and creep rate and calculate the current derivative value du / dλ = 2a*λ(k) + b; if du / dλ > 0, then... ,otherwise ,in This is the step size obtained based on the operating condition.

[0035] If the operating condition remains unchanged, it can be understood that the calculated function of adhesion coefficient and creep rate is reliable. Following this logic, the optimal creep rate is tracked to determine if it falls within a preset range. When K... track As the value approaches 1, the optimal creep rate becomes close to the initial value. The system uses the calculated peak value, but it is highly sensitive to noise and fluctuations in the initial value of the optimal creep rate. When K track As the value approaches 0, maintaining the previous value results in a sluggish response. track The selection is dynamically calculated using a confidence model. If the operating condition is abrupt, and the system detects a drastic change in rail surface conditions, such as from dry to wet, the function of adhesion coefficient and creep rate is unreliable. The optimal creep rate is obtained through a hill-climbing method. Further analysis of the function is needed. If a > 0, the data quality used in the adhesion coefficient and creep rate function is poor, such as excessive noise or the absence of valid data in the calculation. Corresponding countermeasures are required. The optimal creep rate is searched using a trial-and-error search strategy. If a < 0, the derivative of the function is further calculated, and the current derivative value du / dλ = 2a*λ(k) + b is calculated. If du / dλ > 0, then... ,otherwise When du / dλ > 0, it indicates that the creep rate has not yet peaked, so the step size needs to be increased to approach the optimal creep rate. When du / dλ > 0, it indicates that the creep rate has already peaked, so the step size needs to be decreased to approach the optimal creep rate. The logic for judging the working condition is as follows: Calculate the adhesion mutation rate, the mean square error between the adhesion coefficient and the creep rate function, and the wheel acceleration using the adhesion coefficient. If the adhesion mutation rate is greater than the upper limit of the adhesion threshold or the wheel acceleration is greater than the acceleration threshold, the working condition is set to a mutation state. If the creep rate difference is greater than the creep rate difference threshold and the adhesion mutation rate is less than the lower limit of the adhesion threshold, the working condition is set to a hold state. Other working conditions are transition states. Obtain the mutation level based on the adhesion mutation rate and the wheel acceleration; obtain the initial step size based on the mutation level; calculate the oscillation suppression factor based on the probability of the directional changes of du / dλ in the last 10 times; obtain the working condition factor based on the mutation level; calculate the gradient factor based on the du / dλ value; and multiply the initial step size, oscillation suppression factor, working condition factor, and gradient factor to obtain the step size. .

[0036] The steps of S400 further include: S410: Obtain updated mileage data based on pulse count, wheel diameter measurement, and number of pulses per revolution from the operating data; S420: Constructs a mileage compensation factor based on operating conditions, and updates mileage data through weighted mileage data based on the mileage compensation factor and mileage data.

[0037] S430: Updates the current wheel diameter value and calibration mileage data using an absolute calibration strategy; The theoretical pulse count is calculated based on the initial mileage of the system. The pulse error generated during driving is calculated based on the theoretical pulse count and the pulse count acquired by the system. The current wheel diameter value is updated based on the pulse error. The calculation expression is: Current wheel diameter value = Initial wheel diameter value * (1 + (Pulse count - Theoretical pulse count) / Theoretical pulse count). S440: Based on the mileage position of the railway comprehensive inspection vehicle, the mileage compensation coefficient is obtained through curve data query to obtain the final mileage data.

[0038] The curve data is a curve of mileage position / compensation coefficient. The fitting process collects the start and end mileage of the track relative to the starting point coordinates, the radius of the track curve, the length of the track curve, and the curve type. The curve type includes at least straight-to-curve point, curve-to-round point, round-to-curve point, and curve-to-straight point. The compensation coefficient is: radius of track curve / (radius of track curve + 0.5 * wheelset inner distance). The correlation between the start and end mileage positions relative to the marked position and the compensation coefficient is established.

[0039] S500: Adjusts vehicle speed control parameters based on final mileage data, adhesion coefficient, and optimal creep rate. Outputs motor control commands based on operating conditions through an anti-slip control algorithm to prevent wheel-rail slippage. The anti-slip control algorithm includes: using a balance point correction algorithm based on optimal creep rate to output motor torque during acceleration, and using a vector control algorithm based on load torque feedforward compensation during speed control.

[0040] S510: The estimated speed is obtained based on the final mileage data through a first-order low-pass differential filter. The calculation expression is as follows: In the formula, This represents the final mileage data for the current and previous moments. The actual running speed at the previous moment. For the collection period, These are the filter coefficients; S520: Based on the actual operating speed and estimated load torque, online optimization is performed using a model predictive control algorithm to solve for the reference torque; wherein, the construction of the model predictive control algorithm includes: constructing discretized vehicle dynamics equations and satisfying system dynamic constraints, wherein the expressions for the system dynamic constraints are: In the formula, To predict speed, M represents the mass of the railway inspection vehicle, D represents the effective diameter of the wheel, and T represents the speed. m (k) represents the current motor torque. Here is the estimated load torque at the current moment, and n is the mechanical transmission ratio. The optimization objective is to track the running resistance of the railway comprehensive inspection vehicle, and to perform error tracking of the running speed and motor torque within the sampling period. Constraints include ensuring the motor torque is within a preset threshold range. , The predicted speed sequence value is obtained based on the motor torque through system dynamic constraints; where , respectively, creep rate and optimal creep rate, For safety reasons, Where is the adhesion coefficient, and N is the vertical load on the wheel. To predict adhesion; S530: The reference torque is corrected based on the acceleration state of the railway comprehensive inspection vehicle: when in the acceleration stage, the balance point correction algorithm based on the optimal creep rate is used for correction; when in the speed control stage, the feedforward compensation correction based on the load torque is used; the corrected torque is limited and smoothed before outputting control commands.

[0041] The actual operating speed is processed by a first-order low-pass differential filter to obtain a real-time speed estimate. This estimated speed is compared with the set target speed to calculate the actual operating speed tracking error. The actual operating speed is then optimized by a model predictive control (MPC) optimizer. Based on the vehicle dynamics model, the MPC solves for the optimal actual operating speed reference torque sequence over a future period, under conditions such as adhesion and torque constraints. The system uses multi-source sensors to identify the current actual operating speed status (acceleration phase / speed control phase) in real time. In the acceleration phase, an actual operating speed equilibrium point correction algorithm is used to dynamically adjust the reference torque based on the optimal creep rate. In the speed control phase, an actual operating speed feedforward compensation algorithm is used to perform steady-state compensation based on the load torque estimate. The corrected torque is then subjected to actual operating speed safety limiting and smoothing filtering to ensure the engineering safety of the output command. The final torque command after the above processing is sent to the actual operating speed motor driver.

[0042] Preferably, the step of obtaining the load torque estimate further includes: Based on the motor's three-phase current, three-phase voltage, and speed, the q-axis current and rotor electrical angle are obtained through Clark transformation, Park transformation, and low-pass filtering. The actual motor speed at the current moment is calculated based on the pulse count of the operating data. Based on the actual motor speeds at the current moment and the previous two moments, the estimated motor speed at the previous moment is obtained by adjusting the motor speed adaptive differential gain to obtain the speed estimation error, differential feedforward gain, and error differential term. Based on the q-axis current, the current actual motor speed, the previous estimated motor speed, the previous estimated load torque, the speed estimation error, and the error differential term, the estimated motor speed and the estimated load torque at the current moment are obtained through the recursive formula of the improved three-stage recursive observer.

[0043] The aforementioned method obtains the current load torque estimate through rapid response, error correction, and smooth output. This improved observer overcomes the shortcomings of traditional load torque estimation accuracy. By finely transforming and processing the motor signal and optimizing the recursive structure, the accuracy of the load torque is improved. Compared to the significant errors inherent in traditional methods, this scheme enhances observation accuracy while capturing subtle load changes, providing the control system with accurate and reliable load status information. The response lag problem of traditional observers is fundamentally solved, enabling rapid tracking of both sudden and gradual load changes. Whether the load is suddenly applied or gradually changed, the observer in this application can respond accurately in a very short time, providing timely status feedback for real-time control.

[0044] Preferably, the step of obtaining the load torque estimate by improving the recursive formula of the three-stage recursive observer further includes: The motor torque compensation term is obtained based on the direct-axis current, q-axis current, number of motor pole pairs, differential feedforward gain, and error differential term. The calculation expression is as follows: In the formula, The electromagnetic torque of the motor. For differential feedforward gain, This is the differential term of the error; A nonlinear error feedback function and a motor speed prediction model are constructed to obtain the predicted motor speed. The expression for the nonlinear error feedback function is as follows: In the formula, To estimate the rotational speed error, a preset error threshold is set. K is a shape parameter. p1 K p2 These are the first proportional gain parameter, the second proportional gain parameter, and K. d1 K d2 These are the first differential gain parameter and the second differential gain parameter, respectively. The formula for calculating the predicted motor speed is: In the formula, This is the estimated motor speed at time k-1. For the collection period, For rotational inertia, The damping coefficient; Based on the predicted motor speed and the actual motor speed, the torque imbalance value is obtained through torque imbalance detection. The calculation expression is as follows: In the formula, This refers to the actual speed of the motor; The estimated load torque value is obtained based on the adaptive forgetting factor, torque imbalance value, and motor torque compensation term. The calculation expression is as follows: In the formula, An adaptive forgetting factor. For mutation correction terms, the calculation expression for mutation correction terms is: Kcorr is the dynamic parameter of the observer.

[0045] The core innovation of this scheme is the replacement of traditional linear error feedback with a nonlinear error feedback function. This function can intelligently switch the strength and characteristics of the feedback gain according to the magnitude of the speed estimation error: when the error is controlled within a small error range, a relatively low gain (K) is used. p1 The actual operating speed (Kd1) focuses on the smoothness of observations and noise suppression, avoiding over-response to minor disturbances and ensuring the stability of steady-state observations. When the error is within a large error range, it automatically switches to high gain (K). p2 K d2 The dual-mode adaptive structure effectively resolves the contradiction between speed and stability in traditional observers, enabling the observer to achieve both wide dynamic range error convergence and excellent steady-state accuracy. A more refined shape parameter further finely controls the smoothness of the switching process, preventing observational jitter caused by sudden gain changes. An explicit motor speed prediction model (including moment of inertia J and damping coefficient B) is constructed, elevating the observation process from purely mathematical fitting to a level driven by physical mechanisms. The predicted motor speed not only depends on the previous estimate but also incorporates the effects of electromagnetic torque, load torque estimation, and its own dynamic characteristics. The mechanism of predicting the load torque estimate using the torque imbalance value creates an inherent error detection and correction mechanism. The torque imbalance value reflects the torque difference between the model prediction and the actual measurement; any unmodeled dynamics, parameter mismatches, or external disturbances are reflected in this value, thus being captured by the system and used for subsequent correction. An innovative mutation correction term T is introduced. corr If ΔT(k) increases sharply due to an external abrupt change, the correction term is activated instantaneously, directly offsetting the observation lag caused by the sudden loading or unloading. As a feedforward compensation, it does not disrupt the stability of the original feedback loop of the observer.

[0046] Preferably, the step of obtaining the adhesion coefficient further includes: Based on the estimated load torque and operating parameters at the current moment, the wheel-rail adhesion is obtained through torque balance relationship conversion; The adhesion coefficient is constructed based on wheel-rail adhesion and wheel vertical load using the Coulomb friction model, and the calculation expression is as follows: In the formula, n is the mechanical transmission ratio, N is the vertical load on the wheel, and D is the effective diameter of the wheel.

[0047] This scheme cleverly transforms the calculation of the tangential force (adhesion) at the wheel-rail interface, which is difficult to measure directly, into the utilization of the load torque that has been accurately observed on the motor side, through the transformation of torque balance relationships. This realizes a paradigm shift from struggling to measure directly at the wheel-rail interface to indirect calculation, fundamentally solving the core engineering problem of obtaining adhesion information with high reliability and low cost in coal mining environments. The core of this method lies in establishing a complete information transmission chain from the motor output at the actual operating speed (electromagnetic torque), the actual operating speed of the transmission system (load torque), the actual operating speed of the wheel-rail interface (adhesion), to the actual operating speed. The torque information on the motor side is transmitted to the wheel-rail interface without distortion and traceably. This allows the rich information originally closed within the drive system to be extracted and used to evaluate the external wheel-rail contact state, greatly expanding the dimension and depth of vehicle state perception. The improved high-precision load torque estimate output by the three-stage recursive observer is the foundation of this method. This ensures that the calculation of the adhesion coefficient is not generated out of thin air, but is rooted in a strengthened observation foundation. The small error in load torque estimation, after being divided by the relatively stable vertical load N, will not be disproportionately amplified in the calculation of the adhesion coefficient. Instead, it achieves a smooth transfer and maintenance of accuracy from the electric drive system to the wheel-rail interface.

[0048] Preferably, the step of obtaining the creep rate further includes: The linear velocity of the wheel is obtained based on the pulse count from the operational data, the effective diameter of the wheel from the geometric parameters, and the acquisition period. The calculation expression is as follows: In the formula, The acquisition period is defined as PPR, which represents the number of pulses per revolution of the data processing unit. These are the pulse count data for the current and previous moments in the operating parameters, and the effective diameter of the wheel; The creep rate is obtained using the Carter formula based on the linear velocity of the wheel and the actual running speed in the running data. The calculation expression is as follows: In the formula, This refers to the actual operating speed.

[0049] This system achieves high-frequency, low-cost real-time calculation of creep rate by utilizing standard signals from existing vehicle pulse encoders and speed sensors, eliminating the need for specialized measurement equipment, thus simplifying the system architecture and reducing hardware costs. By accurately calculating the wheel linear velocity and comparing it with a reference speed, the relative slippage between the wheel and rail is directly and continuously quantified, providing dynamic state feedback for adhesion control. Employing the classic Carter formula ensures rapid completion with minimal computational overhead in embedded control systems, meeting the high-frequency computation requirements of real-time control. It provides high-quality input for upper-level adhesion coefficient identification and optimal creep rate tracking algorithms. Accurate creep rate data is crucial for the recursive least squares method to accurately fit the adhesion-creep curve, improving anti-slip control performance.

[0050] Preferably, the step of obtaining the optimal creep rate further includes: Based on adhesion coefficient-creep rate data, a function relating adhesion coefficient and creep rate is constructed using the recursive least squares method. The specific process is as follows: The function relating adhesion coefficient and creep rate is constructed, and its expression is: In the formula, , respectively, are the adhesion coefficient and creep rate in the adhesion coefficient-creep rate data, and a, b, and c are the model parameters; The adaptive forgetting factor is calculated based on the function of adhesion coefficient and creep rate, and the function of adhesion coefficient and creep rate is updated accordingly; The initial value of the optimal creep rate is obtained by calculating the peak creep rate and dynamic filtering constraints based on the model parameters. The optimal creep rate is obtained based on the initial value of the optimal creep rate, the operating condition, and the creep rate at the current moment. Specifically, if the operating condition is a stable state, the optimal creep rate is obtained through a smoothing tracking algorithm, calculated as follows: In the formula, These are the current optimal creep rate, the previous optimal creep rate, the initial value of the optimal creep rate, and the tracking coefficient. If the working condition is in a sudden change, the model parameters are further judged to obtain the optimal creep rate.

[0051] Online adaptive modeling of adhesion characteristics is achieved by dynamically capturing the nonlinear hump curve characteristics of the adhesion coefficient changing with creep rate under current rail surface conditions through real-time fitting of a quadratic function using the recursive least squares method. An adaptive forgetting factor mechanism is employed to strengthen historical data learning under stable operating conditions to ensure model stability, and to rapidly increase the weight of new data during rail surface abrupt changes to achieve rapid model tracking, resolving the contradiction between model robustness and adaptability. The optimal initial creep rate value is obtained through dual constraints of peak value calculation and dynamic filtering. This involves using the theoretical peak value derived from the model's analytical solution and eliminating abnormal data interference through threshold and confidence level checks, ensuring the reliability of the initial value. A multi-strategy decision-making mechanism based on operating conditions is implemented. Smooth tracking is used for stable optimization under stable conditions, while switching to a logic judgment and search strategy based on model parameters under abrupt changes, ensuring reliable output of the optimal target under various operating conditions. This forms a complete closed loop from data acquisition and model update to decision output, enabling the control system to track and lock the maximum adhesion point of the current rail surface in real time, providing an adaptive and high-precision dynamic optimization target for traction and anti-slip control.

[0052] Preferably, the step of further determining the model parameters to obtain the optimal creep rate further includes: if a > 0, If a < 0, further differentiate the function of adhesion coefficient and creep rate and calculate the current derivative value du / dλ = 2a*λ(k) + b; if du / dλ > 0, then... ,otherwise ,in This is the step size obtained based on the operating condition.

[0053] The sign of the quadratic coefficient 'a' is used to quickly determine whether the identification result violates the physical laws of the adhesion curve. When the model fails (a>0), a conservative trial strategy is immediately switched to prevent the control system from being misled by an erroneous model. Intelligent search decision-making based on gradient sign is implemented. When the model is reliable (a<0), the derivative direction at the current operating point is combined to accurately determine whether the vehicle is to the left or right of the adhesion peak, thus intelligently deciding the search direction to increase or decrease the creep rate. A three-layer progressive decision logic of model diagnosis, strategy selection, and directional search is formed, decomposing the complex nonlinear optimization problem into an executable rule set, adapting to the real-time requirements of embedded systems. The search step size is adaptively adjusted according to the operating conditions, enabling the search process to quickly approach the target under abrupt changes and finely adjust under stable conditions, achieving a balance between convergence speed and steady-state accuracy.

[0054] Preferably, the step of updating mileage data through wheel diameter compensation strategy and slippage compensation strategy further includes: Updated mileage data is obtained based on pulse counts, wheel diameter measurements, and the number of pulses per revolution from the operational data. A mileage compensation factor is constructed based on the operating conditions, and the mileage data is updated by weighting the mileage compensation factor and mileage data.

[0055] By statically correcting systemic errors caused by wheel wear through wheel diameter compensation, and dynamically suppressing instantaneous slippage interference through a slippage compensation factor adapted to operating conditions, comprehensive error suppression from physical wear to dynamic disturbances is achieved. The compensation factor is intelligently generated based on operating conditions, maintaining high-precision mileage output during normal operation. When slippage / idling is detected, the trust weight of pulse counts is automatically reduced to prevent slippage errors from accumulating into the final mileage. Real-time reliability assessment and weighted fusion of mileage data are achieved. The compensation factor is essentially a quantitative assessment of the reliability of the current pulse count. Through a weighted update mechanism, when pulse data is unreliable, it can smoothly switch to the model-calculated value, ensuring the continuity and robustness of mileage output.

[0056] Preferably, the step of adjusting the vehicle speed control parameters based on the final mileage data, adhesion coefficient, and optimal creep rate further includes: The estimated speed is obtained based on the final mileage data using a first-order low-pass differential filter. The calculation expression is as follows: In the formula, This represents the final mileage data for the current and previous moments. The actual running speed at the previous moment. For the collection period, These are the filter coefficients; Based on the actual operating speed and estimated load torque, online optimization is performed using a model predictive control algorithm to solve for the reference torque. The construction of the model predictive control algorithm includes: constructing discretized vehicle dynamics equations that satisfy system dynamic constraints, wherein the expressions for the system dynamic constraints are: In the formula, To predict speed, M represents the mass of the railway inspection vehicle, D represents the effective diameter of the wheel, and T represents the speed. m (k) represents the current motor torque. Here is the estimated load torque at the current moment, and n is the mechanical transmission ratio. The optimization objective is to track the running resistance of the railway comprehensive inspection vehicle, and to perform error tracking of the running speed and motor torque within the sampling period. Constraints include ensuring the motor torque is within a preset threshold range. , The predicted speed sequence value is obtained based on the motor torque through system dynamic constraints; where , respectively, creep rate and optimal creep rate, For safety reasons, Where is the adhesion coefficient, and N is the vertical load on the wheel. To predict adhesion; The reference torque is corrected based on the acceleration state of the railway comprehensive inspection vehicle: when in the acceleration phase, a balance point correction algorithm based on the optimal creep rate is used for correction; when in the speed control phase, a feedforward compensation correction based on the load torque is used; the corrected torque is then limited and smoothed before outputting control commands.

[0057] A smooth speed estimate is extracted from high-precision mileage data using a first-order low-pass differential filter, providing a reliable feedback benchmark for optimized control. A four-layer collaborative control architecture—state estimation, global optimization, operating condition correction, and safe execution—is constructed. Vehicle dynamics constraints, adhesion safety constraints, and motor physical constraints are uniformly incorporated into the optimization problem for solution. Under the premise of satisfying all safety boundaries, a reference torque that meets the requirements of speed tracking and smoothness is generated, proactively preventing slippage at the optimization level and achieving globally optimal decision-making based on model predictive control. During the acceleration phase, an equilibrium point correction algorithm is used to quickly stabilize the creep rate and prevent idling. During the speed control phase, load feedforward compensation is used to improve steady-state accuracy, achieving precise matching between dynamic and steady-state process control strategies. An operating condition adaptive dual-mode correction mechanism is designed, internalizing the anti-slip requirements into the control law through explicit adhesion and creep rate constraints. This ensures that the traction force does not exceed the current adhesion limit and the creep rate operates within the safe optimal range during optimization, achieving a fundamental shift from post-slippage suppression to pre-slippage prevention. Finally, after limiting and smoothing safety post-processing, the output control commands are ensured to both follow the performance goals of the optimization algorithm and meet the hard safety requirements and actuator dynamic limitations of engineering implementation.

[0058] Second Embodiment A second aspect of the present invention provides a railway integrated detection speed control device based on distance and anti-skid compensation, comprising: Initialization module: used to start the railway comprehensive inspection vehicle and load operating parameters; Data acquisition module: used to periodically collect operational data and geometric parameters; State observation module: used to obtain load torque estimate, adhesion coefficient, creep rate and optimal creep rate based on the state observation algorithm using the running data; Data preprocessing module: used to obtain raw mileage data through periodic calculations based on runtime data and geometric parameters; Data compensation and calibration module: used to update mileage data based on running data and geometric parameters through wheel diameter compensation strategy and slippage compensation strategy; used to update the current wheel diameter value and calibration mileage data through absolute calibration strategy; used to obtain the mileage compensation coefficient and obtain the final mileage data based on the mileage position of the railway comprehensive inspection vehicle through curve data query. Speed ​​control and anti-slip module: It is used to adjust the speed control parameters based on the final mileage data, adhesion coefficient and optimal creep rate, and output motor control commands based on the working condition through the anti-slip control algorithm to prevent wheel-rail slippage.

[0059] By clearly defining six modules—initialization, data acquisition, state observation, data processing, compensation calibration, and control anti-slip—a complete information processing and control closed loop is constructed, from bottom-level perception to top-level decision-making. Key unmeasurable state variables such as load torque, adhesion coefficient, creep rate, and optimal creep rate are acquired in real time through online algorithms, laying a precise perception foundation for intelligent control. Through dynamic compensation (wheel diameter, slippage), absolute calibration, and curve geometric compensation, the original pulse mileage is gradually improved to a final mileage usable for high-precision control and positioning, fundamentally solving the problem of low mileage reliability in the complex environment of coal mines. The speed control and anti-slip module integrates optimization and adaptive control strategies. Based on accurate final mileage and state observation results, it achieves precise speed tracking and proactive prevention of wheel-rail slippage through model predictive control and adaptive correction algorithms. An integrated hardware and software solution is provided, enabling the railway comprehensive inspection vehicle to maintain high-precision operation, efficient inspection, and a high level of safety even in harsh and variable track environments.

[0060] In the description of this application, it should be noted that the terms "inner" and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product is in use. These terms are used solely for the purpose of facilitating and simplifying the description of this application, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] It should also be noted that, unless otherwise explicitly specified and limited, the terms "setup" and "connection" should be interpreted in a broad sense, for example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0062] Those skilled in the art will readily understand that, for the sake of convenience and brevity, the specific identification content executed by the systems and devices described above can be referred to the corresponding processes in the foregoing method embodiments. The embodiments of the present invention have been described in detail above with reference to the accompanying drawings; however, the present invention is not limited to the above embodiments. Even if various modifications are made to the present invention, if these modifications fall within the scope of the claims of the present invention and their equivalents, they shall still fall within the protection scope of the present invention.

Claims

1. A railway integrated inspection speed control method based on distance and anti-skid compensation, characterized in that, include: Start the railway comprehensive inspection vehicle and load operating parameters; By periodically collecting operational data and geometric parameters; Based on the aforementioned operational data, the load torque estimate, adhesion coefficient, creep rate, and optimal creep rate are obtained through a state observation algorithm. Based on the operating data and the geometric parameters, the original mileage data is obtained through periodic calculation; based on the operating data and the geometric parameters, the mileage data is updated through wheel diameter compensation strategy and slippage compensation strategy; the current wheel diameter value is updated and the mileage data is calibrated through absolute calibration strategy; based on the mileage position of the railway comprehensive inspection vehicle, the mileage compensation coefficient is obtained through curve data query to obtain the final mileage data. Based on the final mileage data, the adhesion coefficient, and the optimal creep rate, the vehicle speed control parameters are adjusted. Based on the operating conditions, motor control commands are output through an anti-slip control algorithm to prevent wheel-rail slippage. The anti-slip control algorithm includes: in the acceleration phase, an equilibrium point correction algorithm based on the optimal creep rate is used to output motor torque; in the speed control phase, a vector control algorithm based on load torque feedforward compensation is used.

2. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The step of obtaining the estimated load torque value further includes: Based on the motor's three-phase current, three-phase voltage, and speed, the q-axis current and rotor electrical angle are obtained through Clark transformation, Park transformation, and low-pass filtering. The actual motor speed at the current moment is calculated based on the pulse count of the operating data. Based on the actual motor speeds at the current moment and the previous two moments, the estimated motor speed at the previous moment is used to obtain the speed estimation error, differential feedforward gain, and error differential term through adaptive differential gain adjustment of motor speed; Based on the q-axis current, the actual motor speed at the current moment, the estimated motor speed at the previous moment, the estimated load torque at the previous moment, the speed estimation error, and the error differential term, the estimated motor speed and the estimated load torque at the current moment are obtained through the recursive formula of the improved three-stage recursive observer.

3. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 2, characterized in that, The step of obtaining the estimated load torque value by improving the recursive formula of the three-stage recursive observer further includes: The motor torque compensation term is obtained based on the direct-axis current, q-axis current, number of motor pole pairs, differential feedforward gain, and error differential term. The calculation expression is as follows: In the formula, The electromagnetic torque of the motor. For differential feedforward gain, This is the differential term of the error; A nonlinear error feedback function and a motor speed prediction model are constructed to obtain the predicted motor speed. The expression for the nonlinear error feedback function is as follows: In the formula, To estimate the rotational speed error, a preset error threshold is set. For shape parameters, Kp1, K p2 These are the first proportional gain parameter, the second proportional gain parameter, and K. d1 K d2 These are the first differential gain parameter and the second differential gain parameter, respectively. The formula for calculating the predicted motor speed is: In the formula, This is the estimated motor speed at time k-1. For the collection period, For rotational inertia, The damping coefficient; Based on the predicted motor speed and the actual motor speed, the torque imbalance value is obtained through torque imbalance detection. The calculation expression is as follows: In the formula, This refers to the actual speed of the motor; The estimated load torque value is obtained based on the adaptive forgetting factor, torque imbalance value, and motor torque compensation term. The calculation expression is as follows: In the formula, An adaptive forgetting factor. The mutation correction term is calculated as follows: Kcorr is the dynamic parameter of the observer.

4. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The steps for obtaining the adhesion coefficient further include: Based on the estimated load torque at the current moment, the operating parameters are converted through torque balance relationship to obtain wheel-rail adhesion. Based on the wheel-rail adhesion and the vertical load on the wheel, the adhesion coefficient is constructed using the Coulomb friction model, and the calculation expression is as follows: In the formula, n is the mechanical transmission ratio, N is the vertical load on the wheel, and D is the effective diameter of the wheel.

5. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The step of obtaining the creep rate further includes: Based on the pulse count from the operational data, the effective wheel diameter from the geometric parameters, and the acquisition period, the linear velocity of the wheel is obtained, and the calculation expression is: In the formula, The acquisition period is defined as PPR, which represents the number of pulses per revolution of the operational data. These are the pulse count data for the current and previous moments in the operating parameters, respectively, and the effective diameter of the wheel; The creep rate is obtained using the Carter formula based on the linear velocity of the wheel and the actual running speed in the running data. The calculation expression is as follows: In the formula, This refers to the actual operating speed.

6. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The step of obtaining the optimal creep rate further includes: Based on adhesion coefficient-creep rate data, a function relating adhesion coefficient and creep rate is constructed using the recursive least squares method. The specific process is as follows: The function relating adhesion coefficient and creep rate is constructed, and its expression is: In the formula, , respectively, are the adhesion coefficient and creep rate in the adhesion coefficient-creep rate data, and a, b, and c are the model parameters; The adaptive forgetting factor is calculated based on the function of the adhesion coefficient and the creep rate, and the function of the adhesion coefficient and the creep rate is updated accordingly. The initial value of the optimal creep rate is obtained by calculating the peak creep rate and dynamic filtering constraints based on the model parameters. The optimal creep rate is obtained based on the initial value of the optimal creep rate, the operating condition, and the creep rate at the current moment. Specifically, if the operating condition is a stable state, the optimal creep rate is obtained through a smoothing tracking algorithm, calculated as follows: In the formula, These are the optimal creep rate at the current moment, the optimal creep rate at the previous moment, the initial value of the optimal creep rate, and the tracking coefficient. If the working condition is a sudden change, the model parameters are further determined to obtain the optimal creep rate.

7. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 6, characterized in that, The step of further determining the model parameters to obtain the optimal creep rate further includes: if a > 0, If a < 0, further differentiate the function of adhesion coefficient and creep rate and calculate the current derivative value du / dλ = 2a*λ(k) + b; if du / dλ > 0, then... ,otherwise ,in This is the step size obtained based on the operating condition.

8. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The steps for updating mileage data using wheel diameter compensation and slippage compensation strategies further include: Updated mileage data is obtained based on the pulse count, wheel diameter measurement, and number of pulses per revolution of the aforementioned operating data; A mileage compensation factor is constructed based on the operating conditions, and the mileage data is updated by weighting the mileage compensation factor and the mileage data.

9. The railway integrated inspection speed control method based on distance and anti-skid compensation according to claim 1, characterized in that, The step of adjusting the vehicle speed control parameters based on the final mileage data, the adhesion coefficient, and the optimal creep rate further includes: Based on the final mileage data, the estimated speed is obtained through a first-order low-pass differential filter, and the calculation expression is as follows: In the formula, This represents the final mileage data for the current and previous moments. The actual running speed at the previous moment. For the collection period, These are the filter coefficients; Based on the actual operating speed and estimated load torque, online optimization is performed using a model predictive control algorithm to solve for the reference torque. The construction of the model predictive control algorithm includes: constructing discretized vehicle dynamics equations that satisfy system dynamic constraints, wherein the expressions for the system dynamic constraints are: In the formula, To predict speed, M represents the mass of the railway inspection vehicle, D represents the effective diameter of the wheel, and T represents the speed. m (k) represents the current motor torque. Here is the estimated load torque at the current moment, and n is the mechanical transmission ratio. The optimization objective is to track the running resistance of the railway comprehensive inspection vehicle, and to perform error tracking of the running speed and motor torque within the sampling period. Constraints include ensuring the motor torque is within a preset threshold range. , The predicted speed sequence value is obtained based on the motor torque through system dynamic constraints; where , respectively, creep rate and optimal creep rate, For safety reasons, Where is the adhesion coefficient, and N is the vertical load on the wheel. To predict adhesion; The reference torque is corrected based on the acceleration state of the railway comprehensive inspection vehicle: when in the acceleration phase, a balance point correction algorithm based on the optimal creep rate is used for correction; when in the speed control phase, a feedforward compensation correction based on the load torque is used; the corrected torque is then limited and smoothed before outputting control commands.

10. A railway integrated detection speed control device based on distance and anti-slip compensation, characterized in that, include: Initialization module: used to start the railway comprehensive inspection vehicle and load operating parameters; Data acquisition module: used to periodically collect operational data and geometric parameters; State observation module: used to obtain the load torque estimate, adhesion coefficient, creep rate and optimal creep rate based on the operating data through the state observation algorithm; Data preprocessing module: used to obtain raw mileage data through periodic calculations based on the running data and the geometric parameters; Data compensation and calibration module: used to update mileage data based on the running data and geometric parameters through wheel diameter compensation strategy and slippage compensation strategy; used to update the current wheel diameter value and calibrate the mileage data through absolute calibration strategy; used to obtain the final mileage data by querying curve data based on the mileage position of the railway comprehensive inspection vehicle; Speed ​​control and anti-slip module: used to adjust speed control parameters based on the final mileage data, the adhesion coefficient and the optimal creep rate, and output motor control commands based on the working condition through the anti-slip control algorithm to prevent wheel-rail slippage.

Images