Tunnel transport vehicle multi-motor cooperative control method and system based on virtual principal axis

Abstract

The invention discloses a tunnel transport vehicle multi-motor cooperative control method based on a virtual main axis, and belongs to the technical field of tunnel engineering vehicle control. Comprising the following steps: acquiring multi-source state information in the operation process of a tunnel transport vehicle in real time, and preprocessing an acquired signal; according to the processed vehicle state and gradient information, a dynamic virtual main shaft rotating speed reference value and a torque reference value are generated to serve as a unified benchmark of whole vehicle cooperative control; calculating the rotating speed deviation and the torque deviation of each motor relative to the virtual main shaft, and fusing the rotating speed deviation and the torque deviation into comprehensive deviation; dynamically adjusting the participation weight of each motor according to the comprehensive deviation of each motor; and on the basis of the adjusted weights, double-closed-loop cooperative control of a speed loop and a torque loop is executed on each driving motor. According to the invention, the problems in the prior art that the synchronization precision of multiple motors is low, the torque distribution is non-uniform, and overload or slipping is easy to occur under the working condition of large slope and heavy load climbing of the tunnel are solved, and the traction stability and the operation safety of the tunnel material transport vehicle are obviously improved.

Description

Technical Field

[0001] This invention relates to a multi-motor cooperative control method and system for tunnel transport vehicles based on a virtual spindle, belonging to the field of tunnel engineering vehicle control technology. Background Technology

[0002] With the continuous expansion of underground space development, tunnel engineering places increasingly higher demands on material transport equipment. Especially in the construction of steep tunnels, heavy-duty material transport commonly employs multi-section specialized tunnel transport vehicles. These vehicles typically have multiple drive shafts per section, each driven by an independent drive motor, forming a typical multi-motor distributed drive system. Theoretically, this architecture can provide significant traction power redundancy and a certain degree of drive flexibility. However, in practical engineering applications, particularly under complex conditions such as continuous uphill climbing, sudden gradient changes, and uneven load distribution, existing multi-motor cooperative control methods have revealed some common problems.

[0003] Currently, the commonly used multi-motor cooperative control strategies in engineering mainly include master-slave control, parallel control, and control structures based on deviation coupling. Master-slave control takes one motor as the reference, and the other motors follow its output. The structure is relatively simple, but once the master motor is disturbed, the error will be amplified step by step. Parallel control involves each motor adjusting independently, and there is a lack of effective coordination mechanism between them. Although deviation coupling control introduces error feedback between motors, it has high computational complexity in long train formations with multiple motors and multiple carriages, and it is still difficult to cope with drastic dynamic changes in gradient and load.

[0004] For heavy-duty, multi-axle, long-formation vehicles such as tunnel transport vehicles, existing technological approaches generally face the following technical challenges:

[0005] First, the control architecture lacks a globally unified reference benchmark. Most existing multi-motor systems adopt a distributed independent adjustment method, where each motor controller can only perform closed-loop adjustment based on local information such as speed and current detected by itself. The lack of effective coordination signal channels between motors makes it difficult to achieve overall coordination of the vehicle's drive unit in scenarios with multiple carriages and multiple axles. This is especially true during hill climbing, when the stress state of each axle differs significantly, and the local closed loop cannot reflect the actual power demand of the entire vehicle.

[0006] Secondly, load distribution strategies are ill-suited to the dynamic coupling of slope and load. Traditional methods typically distribute torque based on static or simplified load models, such as distributing it evenly according to rated power or according to a fixed ratio. However, in actual tunnel construction, vehicles face dynamic factors such as continuously changing steep road surfaces, differences in material distribution between truck beds, and shifts in the center of gravity of materials during the uphill climb. These factors cause the real-time load on each drive axle to exhibit nonlinear and spatially distributed changes. Existing methods lack the ability to integrate slope information and load distribution information in real time, making it difficult to respond quickly and differentiatedly to load changes.

[0007] Third, the ability to suppress synchronization error is insufficient. Under the existing architecture, due to the lack of an effective global synchronization benchmark and load estimation mechanism, the output state of each motor is prone to significant differences. Some motors may be in an overload or light load state for a long time, while others may experience torque oscillation due to the accumulation of synchronization error. This unbalanced operating mode will not only further amplify the synchronization error, but also easily cause wheel slippage or single axle overload, affecting the traction stability and operation safety of the whole vehicle.

[0008] In summary, under typical working conditions such as long tunnel slopes and heavy-load climbing, existing multi-motor cooperative control methods mainly have the following problems: lack of a unified reference benchmark that can simultaneously reflect slope changes and load distribution dynamics; speed synchronization and torque distribution are often handled separately, making it difficult to form an effective cooperative adjustment mechanism; the participation weights and compensation strategies of each motor are not linked with real-time working conditions, resulting in low synchronization accuracy and weak load balancing capabilities.

[0009] Therefore, there is an urgent need for a method that can achieve coordinated control of multiple motor drive units under dynamic slope and variable load conditions without increasing the system complexity, thereby improving the traction stability and operational safety of tunnel transport vehicles under complex working conditions. Summary of the Invention

[0010] The technical problem to be solved by the present invention is to provide a multi-motor cooperative control method for tunnel transport vehicles based on a virtual spindle. It solves the problems of low synchronization accuracy, uneven torque distribution, and easy overload or slippage of multi-motors in the existing technology under the heavy load climbing conditions of long tunnel slopes. It significantly improves the traction stability and operation safety of tunnel material transport vehicles.

[0011] The technical problem to be solved by this invention is achieved by the following technical solution:

[0012] A multi-motor cooperative control method for a tunnel transport vehicle based on a virtual spindle includes the following steps:

[0013] The multi-source status information of the tunnel transport vehicle during operation is obtained. The multi-source status information includes at least the speed and torque of each drive motor, the slope information of the road surface on which the vehicle travels, the longitudinal acceleration information, and the load distribution information of each axle or each carriage.

[0014] Based on the load distribution information and the slope information, virtual spindle speed reference value and virtual spindle torque reference value are generated;

[0015] Calculate the speed deviation of each motor from the virtual spindle speed reference value, and the torque deviation of each motor from the virtual spindle torque reference value, and combine the speed deviation and torque deviation of each motor into the comprehensive deviation of the motor.

[0016] The participation weight of each motor is adjusted according to its overall deviation, with motors having larger overall deviations receiving lower weights.

[0017] The overall deviation is compensated by weighting based on the adjusted participation weights, and each motor is controlled collaboratively according to the dual closed-loop structure of speed outer loop and torque inner loop.

[0018] The present invention is further configured such that: the multi-source state information also includes image information acquired by the camera and point cloud information acquired by the lidar;

[0019] The multi-source state information is preprocessed, and the preprocessing includes:

[0020] Use a low-pass filter to filter out high-frequency signals;

[0021] Outlier data points were removed by using sliding window mid-value filtering combined with statistical criteria.

[0022] An adaptive Kalman filter is used to estimate the state based on the slope and longitudinal acceleration information.

[0023] The processed information is mapped to a preset interval using a normalization method, wherein the mapping boundary of the normalization method is dynamically adjusted according to the slope information and the load distribution information.

[0024] The present invention is further configured such that the process noise covariance of the adaptive Kalman filter is dynamically adjusted according to the following conditions:

[0025] When the slope angle is greater than 7°, or the maximum relative deviation between the load on each axis and the average load is greater than 20%, or the longitudinal acceleration fluctuation amplitude exceeds the preset acceleration fluctuation threshold, the process noise covariance is increased.

[0026] The present invention is further configured such that the virtual spindle speed reference value is generated in the following way:

[0027] The weighting of each motor is determined based on the load on the corresponding shaft of each motor.

[0028] The base synchronous speed is obtained by multiplying the real-time speed of all motors by their corresponding weights, summing the results, and then dividing by the sum of all weights.

[0029] Calculate the product of the speed compensation coefficient, gravitational acceleration, and the sine of the current slope angle, and then divide it by the effective radius of the wheel to obtain the slope compensation term;

[0030] The reference value of the virtual spindle speed is obtained by adding the basic synchronous speed to the slope compensation term;

[0031] The speed compensation coefficient ranges from 0.05 to 0.15.

[0032] The present invention is further configured such that the virtual spindle torque reference value is generated in the following way:

[0033] The total torque requirement is estimated based on the vehicle's total mass, the aforementioned slope information, and the rolling resistance coefficient.

[0034] Divide the total torque requirement by the total number of motors to obtain the basic allocated torque;

[0035] The maximum relative deviation between the load on each axis and the average load is calculated as the load distribution non-uniformity.

[0036] The torque adjustment coefficient is determined based on the load distribution unevenness. The difference between the real-time torque of each motor and the average torque of all motors is multiplied by the torque adjustment coefficient, and then the basic distributed torque is added to obtain the virtual spindle torque reference value.

[0037] The present invention is further configured such that when the speed deviation and torque deviation of each motor are combined into the comprehensive deviation of the motor, the speed deviation and the torque deviation are multiplied by different weighting coefficients and then summed, and the weighting coefficient of the speed deviation is greater than the weighting coefficient of the torque deviation.

[0038] The weighting coefficient for the speed deviation ranges from 0.6 to 0.8, and the weighting coefficient for the torque deviation ranges from 0.2 to 0.4.

[0039] The present invention is further configured to: the comprehensive weight of each motor ( Determine using the following formula:

[0040]

[0041] in, The initial weights are determined based on the load on the corresponding shaft of the i-th motor. Let be the overall deviation of the i-th motor. This represents the maximum value of the combined deviation of all motors. The attenuation sensitivity coefficient has a value range of 0.3 to 0.7.

[0042] The present invention is further configured such that the double closed-loop structure of the outer speed loop and the inner torque loop is:

[0043] The virtual spindle speed reference value is used as the given value of the speed outer loop, the real-time motor speed is used as the feedback value of the speed outer loop, and the output of the speed outer loop is used as the torque given intermediate value of the torque inner loop.

[0044] The sum of the given torque intermediate value and the compensation term is used as the given value of the torque inner loop, and the real-time torque of the motor is used as the feedback value of the torque inner loop. The torque inner loop outputs motor drive commands.

[0045] The present invention is further configured such that: the compensation term includes a speed deviation feedforward compensation term and a torque equalization compensation term;

[0046] The speed deviation feedforward compensation term is the result of performing proportional, integral, and derivative operations on the speed deviation, multiplied by the comprehensive weight of the motor.

[0047] The torque balance compensation term is the torque deviation multiplied by a preset balance ratio coefficient and then multiplied by the overall weight of the motor.

[0048] The sum of the torque reference value, the speed deviation feedforward compensation term, and the torque balance compensation term is used as the reference value for the inner torque loop.

[0049] A multi-motor cooperative control system for a tunnel transport vehicle based on a virtual spindle, used to execute the method according to any one of claims 1 to 9, characterized in that it comprises:

[0050] The data acquisition and signal processing module is used to acquire multi-source status information during the operation of the tunnel transport vehicle in real time, and to perform filtering, noise reduction and normalization preprocessing on the multi-source status information.

[0051] The virtual spindle construction module, connected to the data acquisition and signal processing module, is used to generate virtual spindle speed reference values ​​and virtual spindle torque reference values ​​based on the preprocessed multi-source state information.

[0052] The motor deviation calculation module, connected to the virtual spindle construction module, is used to calculate the speed deviation and torque deviation of each motor based on the virtual spindle speed reference value, the virtual spindle torque reference value, and the real-time speed and torque of each motor, and merge them into a comprehensive deviation.

[0053] The motor weight adjustment module is connected to the motor deviation calculation module and is used to dynamically adjust the participation weight of each motor according to the comprehensive deviation of each motor, wherein the motor with the larger comprehensive deviation receives a lower weight.

[0054] The collaborative control execution module is connected to the virtual spindle construction module, the motor weight adjustment module, and each drive motor, respectively. It is used to perform collaborative control on each motor according to the dual closed-loop structure of the speed outer loop and the torque inner loop based on the virtual spindle reference value and the adjusted participation weight, and output motor drive commands.

[0055] The beneficial effects of this invention are:

[0056] 1. By constructing a preprocessing system that combines multi-level filtering with adaptive Kalman filtering, high-frequency noise caused by severe road surface bumps, vehicle longitudinal vibration, and material swaying under long uphill tunnel conditions is effectively suppressed. It also eliminates transient abnormal data caused by brief wheel slippage or impact, ensuring the stability and reliability of key signals such as slope, load, and motor status. This provides a high-fidelity data foundation for subsequent control and solves the problem of control instability caused by signal contamination in existing technologies.

[0057] 2. To address the significant differences in force distribution among drive axles under heavy-load conditions with steep inclines, a method for generating virtual spindle reference values ​​based on load weighting and dynamic slope compensation was established. This virtual spindle speed reference value automatically reduces the contribution of the heavily loaded axle to the global baseline, while introducing a slope compensation term to overcome the influence of slope resistance on speed. The virtual spindle torque reference value estimates the vehicle's power demand in real time and dynamically fine-tunes it according to load distribution unevenness, thus providing a unified collaborative control baseline for all motors in the vehicle and significantly improving the speed synchronization accuracy during long uphill climbs.

[0058] 3. By merging speed deviation and torque deviation into a comprehensive deviation, and dynamically adjusting the participation weight of each motor according to the magnitude of the comprehensive deviation, the motor with the larger deviation receives a lower weight, effectively preventing the amplification of abnormal conditions. On this basis, combined with the deviation feedforward compensation in the dual closed-loop control structure, the automatic balanced distribution of load among each drive shaft is realized, avoiding the phenomenon of some motors being overloaded for a long time while other motors are lightly loaded or slipping, which significantly improves the traction stability and operational safety of the tunnel material transport vehicle under heavy-load climbing conditions. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention.

[0060] Figure 2 This is a schematic diagram showing the distribution of the sensor acquisition units of the present invention.

[0061] Figure 3This is an overall flowchart of the method of the present invention. Detailed Implementation

[0062] To facilitate a clear understanding of the technical means, creative features, objectives, and effects of this invention, the invention will be further described below in conjunction with specific illustrations.

[0063] The present invention relates to a multi-motor cooperative control method and system for tunnel transport vehicles based on a virtual spindle, which aims to solve the technical defects of existing multi-motor cooperative control methods under long tunnel slopes and heavy-load climbing conditions, such as low synchronization accuracy, uneven torque distribution, and easy overload or slippage.

[0064] like Figure 1 As shown, a multi-motor cooperative control system for a tunnel transport vehicle based on a virtual spindle includes a data acquisition and signal processing module 11, a virtual spindle construction module 22, a motor deviation calculation module 23, a motor weight adjustment module 24, and a cooperative control execution module 31.

[0065] Among them, the sensor processing module 11, the virtual spindle construction module 22, the motor deviation calculation and consistency evaluation module 23, and the motor weight adjustment module 24 are all integrated on the vehicle main controller (VMC), while the collaborative control execution module 31 is integrated on the motor controller (MCU) of each axle, forming a control architecture of centralized decision-making and distributed execution.

[0066] The data acquisition and signal processing module 11 is used to acquire multi-source status information during the operation of the tunnel transport vehicle in real time, and to perform filtering, noise reduction, and normalization preprocessing on this information. This module includes a sensor acquisition unit 111 and a signal processing unit 112.

[0067] like Figure 2 As shown, the sensor acquisition unit 111 is responsible for real-time acquisition of multi-source data, which specifically includes a motor speed sensor 1106, a current sensor 1105, a vehicle speed sensor 1107, a slope sensor 1103, a pressure sensor 1108, an IMU inertial measurement unit, a camera 1101, and a lidar 1102.

[0068] Motor speed sensor 1106 and current sensor 1105 are arranged at the output end of each drive motor 1104. The motor speed sensor is used to measure the motor speed, and the current sensor is used to calculate the motor output torque based on the phase current.

[0069] The speed sensor 1107 and the slope sensor 1103 are located at the front of the tractor unit to detect the speed of the train and the slope of the incline. The pressure sensors 1108 are distributed under each car and above the drive shaft to detect the load distribution and material offset.

[0070] The IMU (Inertial Measurement Unit) is an integrated sensor used to measure the motion state of the vehicle, including longitudinal acceleration. The camera 1101 is arranged on the top of the vehicle and at the front of the vehicle to collect images of the material surface in real time, which helps to determine the material offset, accumulation state and distribution uniformity.

[0071] The lidar 1102 is arranged at the front of the vehicle and on the sides of each carriage to accurately scan the surrounding environment, distance to obstacles and material outlines of the vehicle group, and realize the auxiliary verification of load distribution and multi-sensor fusion of slope information. The signal processing unit 112 is responsible for preprocessing the data obtained by the acquisition unit 111. The specific process will be detailed in the following method.

[0072] The virtual spindle construction module 22 is connected to the data acquisition and signal processing module 11 and is used to generate virtual spindle speed reference value and virtual spindle torque reference value based on the preprocessed multi-source state information. The module includes a virtual spindle speed reference unit 221 and a virtual spindle torque reference unit 222. The virtual spindle speed reference unit is used to generate the vehicle speed reference, and the virtual spindle torque reference unit is used to generate the vehicle torque reference.

[0073] The motor deviation calculation module 23 is connected to the virtual spindle construction module 22. It is used to calculate the speed deviation and torque deviation of each motor based on the virtual spindle speed reference value, the virtual spindle torque reference value, and the real-time speed and torque of each motor, and then merge them into a comprehensive deviation.

[0074] The motor weight adjustment module 24 is connected to the motor deviation calculation module 23 and is used to dynamically adjust the participation weight of each motor according to the comprehensive deviation of each motor, wherein the motor with the larger comprehensive deviation receives a lower weight.

[0075] The collaborative control execution module 31 is connected to the virtual spindle construction module 22, the motor weight adjustment module 24, and each drive motor 1104, respectively. It is used to perform collaborative control on each motor according to the dual closed-loop structure of speed outer loop and torque inner loop based on the virtual spindle reference value and the adjusted participation weight, and finally output motor drive command.

[0076] like Figure 3 As shown, based on the above system, this invention provides a multi-motor cooperative control method for a tunnel transport vehicle based on a virtual spindle, comprising the following steps:

[0077] S1: Steps for acquiring and preprocessing multi-source state information.

[0078] In actual tunnel material transport vehicle operation, especially under long uphill slopes, sudden gradient changes, or heavy load conditions, the operating status of each drive motor, the overall vehicle posture, and the load distribution change drastically and rapidly. The raw sensor signals contain a large amount of noise and outliers, such as high-frequency interference signals caused by uneven steep road surfaces, severe longitudinal vibration of the heavily loaded vehicle body, and material swaying with the slope, as well as instantaneous abnormal data points caused by brief wheel slippage, instantaneous material displacement, or vibration impact. Directly using these for control would lead to system instability. Therefore, the system acquires multi-source status information during the tunnel transport vehicle's operation in real time. This information includes at least the speed and torque of each drive motor, the gradient information of the road surface, the longitudinal acceleration information, and the load distribution information of each axle or each compartment.

[0079] Specifically, the sensor acquisition unit acquires data in real time at a sampling frequency of no less than 100Hz. Motor speed sensors and current sensors are respectively installed at the output terminal of each drive motor. The motor speed sensor is used to directly measure the motor speed, and the current sensor is used to calculate the motor output torque based on the phase current.

[0080] Speed ​​and gradient sensors are located at the front of the tractor unit to detect the speed of the train and the gradient of the incline. Pressure sensors are located under each carriage and above the drive shaft to detect load distribution and material offset. The inertial measurement unit (IMU) is an integrated sensor used to measure the motion state of the train, including longitudinal acceleration.

[0081] To further enhance the robustness of load distribution and road surface information perception, multi-source state information also includes image information acquired by cameras and point cloud information acquired by lidar. Cameras are deployed on the top and front of the vehicle to collect images of the material surface in real time, assisting in the judgment of material offset, accumulation state and distribution uniformity. LiDAR is deployed at the front of the vehicle and the sides of each vehicle to accurately scan the surrounding environment, obstacle distance and material outline, realizing the auxiliary verification of load distribution and multi-sensor fusion of slope information. These camera image features and lidar point cloud features are also incorporated into the subsequent signal processing flow.

[0082] To ensure data quality, the system performs a series of preprocessing steps on all the collected raw multi-source state information. First, a low-pass filter is used to remove high-frequency noise and retain effective low-frequency information. The cutoff frequency of the low-pass filter is set to 50Hz. The filter has the largest flat amplitude response in the passband and a steep roll-off characteristic in the stopband. It can effectively filter out high-frequency interference signals caused by uneven road surface, severe longitudinal vibration of heavy-load vehicle body and material swaying with the slope when the tunnel transport vehicle is climbing. At the same time, it completely retains key low-frequency dynamic information such as slow slope increase and gradual change of load distribution.

[0083] Secondly, a sliding window midpoint filter combined with statistical criteria is used to remove outlier data points. Specifically, a sliding window with a length of 10 sampling points is set, the standard deviation of the data within the window is calculated, and outlier points exceeding 3 times the standard deviation within the window are removed, thereby quickly removing instantaneous abnormal data caused by brief wheel slippage, instantaneous material displacement, or vibration impact during the uphill process.

[0084] To further address the sudden changes in slope and strong vibration interference under climbing conditions, an adaptive Kalman filter is used for state estimation based on slope information and longitudinal acceleration information. This adaptive Kalman filter algorithm dynamically adjusts the process noise covariance and measurement noise covariance based on the current slope angle of the train, load distribution unevenness, and longitudinal acceleration.

[0085] Specifically, when the slope angle is greater than 7°, or the maximum relative deviation between the load on each axis and the average load is greater than 20%, or the longitudinal acceleration fluctuation amplitude exceeds the preset acceleration fluctuation threshold, the system determines the current working condition as a strong interference condition and increases the process noise covariance accordingly. The load distribution unevenness is obtained in real time by the pressure sensor, thereby effectively suppressing interference and ensuring the high accuracy and smoothness of the slope and acceleration signals.

[0086] All filtered and denoised multi-source state information, including motor speed, phase current, estimated torque, vehicle speed, slope, load, longitudinal acceleration, camera image features, and lidar point cloud features, are normalized to map the processed information to a preset range. For example, the minimum-maximum normalization method is used to map it to the [0, 1] range. The mapping boundary (i.e., minimum and maximum values) of this normalization method is dynamically adjusted according to the slope information and load distribution information: during the uphill heavy load stage, the system significantly expands the dynamic range of torque, current, and load signals to adapt to the large fluctuations in signal amplitude caused by the sharp increase in slope resistance, thereby retaining higher resolution in subsequent processing, which is convenient for subsequent virtual spindle construction and multi-motor deviation calculation.

[0087] S2: After the above preprocessing, the system obtains stable, reliable, and uniform real-time status information. Next, the system proceeds to the step of generating virtual spindle reference values. In the existing technology, due to the lack of a unified benchmark that can simultaneously reflect the vehicle's power requirements and real-time operating conditions, each motor controller can only adjust based on local information, resulting in the inability to effectively coordinate the differences in force on each axle during climbing.

[0088] To address this issue, the system generates dynamic virtual spindle speed and torque reference values ​​based on load distribution and slope information. This virtual spindle is not a physical entity, but rather a mathematical benchmark representing the optimal operating state of the entire vehicle, calculated in real-time within the vehicle's main controller (VMC). Specifically, in the tunnel transport vehicle's uphill climbing scenario, the calculation methods for the dynamic virtual spindle's speed and torque reference values ​​are optimized to account for changes in slope resistance and load distribution.

[0089] The virtual spindle speed reference value is generated as follows: the system first determines the weight of each motor based on the load of the corresponding axis. The weight of the motor corresponding to the axis with higher load is appropriately reduced to reduce the impact of heavy load lag on the global reference.

[0090] Then, multiply the real-time speeds of all motors by their corresponding weights, sum the results, and divide by the sum of all weights to obtain the base synchronous speed, which is the weighted average of all motor speeds. The weight Based on the load distribution of each axis, the weight of motors with higher loads is appropriately reduced. This step makes the load distribution more uniform; that is, the contribution of the real-time speed of the heavily loaded axis to the final reference speed is deliberately reduced to avoid the system as a whole converging towards an overload state.

[0091] Next, the system introduces a slope compensation term. ,in For speed compensation coefficient, It is the acceleration due to gravity. For real-time slope, The effective radius of the wheel, The linear acceleration component caused by the slope, and the velocity compensation coefficient. It is an empirical value, ranging from 0.05 to 0.15, used to adjust the intensity of the influence of slope on the speed reference value.

[0092] Finally, the basic synchronous speed is added to the slope compensation term to obtain the virtual spindle speed reference value, the calculation formula of which is expressed as follows: ,in This is a reference value for the virtual spindle speed. Let i be the actual speed of the i-th motor. For the corresponding weights, For speed compensation coefficient, It is the acceleration due to gravity. For real-time slope, This is the effective radius of the wheel.

[0093] The method for generating the virtual spindle torque reference value focuses on estimating and dynamically allocating the vehicle's power demand. The system first estimates the total torque demand based on the vehicle's total mass (obtained by adding the load measured by pressure sensors to the vehicle's own weight), gradient information, and rolling resistance coefficient. The calculation takes into account gradient resistance, rolling resistance, and additional resistance; that is, it first estimates the total torque requirement of the train set. ,in , This is the rolling resistance coefficient, with a value ranging from 0.01 to 0.05, which should be adjusted according to the actual working conditions at the site. For other additional resistances, including air resistance and tunnel ventilation resistance, the total mass m of the train is obtained by adding the pressure sensor's input to the train's own weight.

[0094] Then, divide the total torque requirement by the total number of motors to obtain the base distributed torque. Here, n represents the total number of motors, which is a baseline value for average distribution. However, due to uneven load distribution across shafts, average distribution cannot meet actual needs. Therefore, the system further introduces dynamic adjustment. The system performs feedback fine-tuning based on the real-time torque of each motor. Specifically, it first calculates the maximum relative deviation between the load on each shaft and the average load as the load distribution unevenness, i.e. ,in It is the real-time load measured by the pressure sensor of the j-th carriage or axle. It is the average load of all carriages.

[0095] A torque adjustment coefficient is determined based on the load distribution unevenness, for example... Finally, the difference between the real-time torque of each motor and the average torque of all motors is multiplied by the torque adjustment coefficient and then added to the base distributed torque to obtain the final virtual spindle torque reference value. ,Right now ,in Let be the real-time torque of the i-th motor. The torque is the average value of all motor torques. In this way, as the load distribution unevenness increases, the torque adjustment coefficient also increases, so that the virtual spindle torque reference value can more sensitively follow the actual load changes, providing a benchmark that is more in line with actual needs for subsequent deviation calculations.

[0096] S3: After obtaining the global unified benchmark of virtual spindle speed reference value and torque reference value, the system enters the multi-dimensional deviation calculation and fusion step.

[0097] Traditional methods only compare speed differences, ignoring the fundamental cause of synchronization errors—torque imbalance. Therefore, in the scenario of a tunnel transport vehicle climbing a slope, the system calculates the speed deviation between each motor's speed and the virtual spindle speed reference value, as well as the torque deviation between each motor's torque and the virtual spindle torque reference value.

[0098] Speed ​​deviation ,in Let i be the actual speed of the i-th motor. This is a reference value for the virtual spindle speed;

[0099] Torque deviation is ,in The real-time torque of the i-th motor is estimated from the current. This is a reference value for the virtual spindle torque.

[0100] Then, the system merges the speed deviation and torque deviation of each motor into the comprehensive deviation of that motor. During the merging, the speed deviation and torque deviation are multiplied by different weighting coefficients and then summed. Since maintaining the synchronization of wheel speeds to prevent slippage is the primary goal under heavy load climbing conditions, the weighting coefficient of the speed deviation is set to be greater than the weighting coefficient of the torque deviation.

[0101] Specifically, the weighting coefficient of the speed deviation The value ranges from 0.6 to 0.8, representing the weighting coefficient for torque deviation. The value range is 0.2 to 0.4, and can be adjusted according to the climbing and heavy-load conditions, with speed synchronization as the primary consideration and torque balance as a secondary consideration, taking into account the overall deviation. The calculation formula is This overall deviation The overall deviation of the i-th motor from the ideal state (virtual spindle) is quantified, including both synchronization error and load balancing error.

[0102] S4: After obtaining the comprehensive deviation of each motor, the system enters the dynamic adjustment step of the participation weight.

[0103] In a multi-motor system, different motors have different "contributions" or "reliability" to the overall control due to their different locations and loads. If a motor that has already shown a large deviation is given a high weight, it may drag the entire system into an unstable state. Therefore, the system adjusts the participation weight of each motor according to its overall deviation. The basic principle of adjustment is to give lower weights to motors with larger overall deviations.

[0104] First, the system determines the initial participation weight of the motor based on the measured load values ​​of each axis, that is, it determines the initial weight of the i-th motor according to the load of the axis corresponding to it. The calculation formula is: ,in The load on the axle corresponding to the i-th motor is collected in real time by a pressure sensor, and m is the total number of drive motors in the vehicle. This initial weight reflects the physical basis of the load on each axle. The heavier the load, the higher its initial weight.

[0105] Then, the system calculates the overall deviation. The initial weights are dynamically adjusted to determine the final comprehensive weight of each motor. Determine by the following formula: .

[0106] in, This represents the maximum value of the combined deviation of all motors. This is the attenuation sensitivity coefficient, also known as the weight adjustment coefficient, used to control the degree of influence of deviation on the weight. Its value ranges from 0.3 to 0.7. The physical meaning of this formula is that when the overall deviation of a motor... The larger it is, the higher its weight. Compared to the initial weights The greater the attenuation, the more reliable the system will trust motors whose operating state is closer to the virtual spindle in subsequent control stages, while doubting motors with excessive deviations, thereby preventing the abnormal state of individual motors from polluting the global control reference.

[0107] S5: Finally, the system enters the collaborative control execution step.

[0108] After obtaining the dynamically adjusted participation weights, the system performs weighted compensation for the overall deviation based on these adjusted weights, and performs coordinated control of each motor according to a dual closed-loop structure of speed outer loop and torque inner loop. Traditional single-loop control cannot simultaneously satisfy the two coupled objectives of speed synchronization and torque balance, while the dual closed-loop structure provides independent control channels for these two objectives. Specifically, the dual closed-loop structure is as follows: using the virtual spindle speed reference value... The given value for the outer speed loop is the motor's real-time speed. This is the feedback value for the outer speed loop. The controller for the outer speed loop (usually a PI controller) calculates the intermediate torque setpoint based on the difference between the setpoint and the feedback value. This intermediate value represents the torque that the motor theoretically needs to output to eliminate speed deviations. Then, the sum of the intermediate torque setpoint and the compensation term is used as the setpoint for the inner torque loop, with the real-time motor torque as the reference value. This is the feedback value for the inner torque loop. The controller of the inner torque loop (usually a PI controller) outputs the final motor drive command (such as a current or voltage signal) to achieve fast and accurate torque tracking.

[0109] The compensation term here is the key to achieving high-precision collaborative control. It consists of two parts: speed deviation feedforward compensation term and torque balance compensation term. The system directly couples the multidimensional deviation to the dual closed-loop controller to achieve active deviation suppression and load balancing.

[0110] Specifically, a deviation feedforward compensation term is introduced into the speed loop. This item is for the speed deviation. The result of the proportional-integral-derivative (PID) calculation is then multiplied by the overall weight of the motor. The purpose is to precisely adjust the motor speed using a PID controller, enabling it to follow the target speed as quickly and accurately as possible while minimizing errors. The physical meaning of this is that for motors with high overall weighting, the system considers their state reliable and therefore provides stronger feedforward compensation to accelerate their speed convergence to the virtual spindle. Conversely, for motors with low overall weighting, the compensation is weakened to prevent abnormal states from being amplified.

[0111] Introducing a load balancing compensation term into the torque loop This item is the torque deviation. Multiply by the preset balance ratio coefficient Then multiply by the overall weight of the motor. This step directly introduces torque imbalance information into the control loop, actively suppressing torque differences. Therefore, the final torque setpoint for each motor... Provide an intermediate value for the torque. Feedforward compensation term for speed deviation Torque balance compensation item The sum of the three is In this way, the comprehensive weight is not only used to adjust the generation of the virtual spindle, but also directly participates in the compensation calculation of each motor's independent controller, forming a complete, closed-loop adaptive collaborative control logic, which effectively ensures the synchronous drive and automatic load balancing of multiple motors under complex climbing conditions.

[0112] Through the aforementioned dual-closed-loop collaborative control strategy based on comprehensive weights, the system achieves the use of virtual spindle reference values ​​as a unified benchmark, feedforward compensation for multi-dimensional deviations, and adaptive adjustment of comprehensive weights, thereby effectively ensuring synchronous drive and automatic load balancing of multiple motors under complex climbing conditions.

[0113] This method effectively solves the technical defects of existing technologies in tunnels with long slopes and heavy loads, such as low synchronization accuracy of multi-motor speeds, uneven torque distribution, and susceptibility to overload or slippage. It significantly improves the traction stability, operating efficiency, and overall safety of tunnel material transport vehicles in complex slope environments.

[0114] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and various changes and modifications can be made without departing from the spirit and scope of the invention, all of which fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-motor cooperative control method for a tunnel transport vehicle based on a virtual spindle, characterized in that, Includes the following steps: The multi-source status information of the tunnel transport vehicle during operation is obtained. The multi-source status information includes at least the speed and torque of each drive motor, the slope information of the road surface on which the vehicle travels, the longitudinal acceleration information, and the load distribution information of each axle or each carriage. Based on the load distribution information and the slope information, virtual spindle speed reference value and virtual spindle torque reference value are generated; Calculate the speed deviation of each motor from the virtual spindle speed reference value, and the torque deviation of each motor from the virtual spindle torque reference value, and combine the speed deviation and torque deviation of each motor into the comprehensive deviation of the motor. The participation weight of each motor is adjusted according to its overall deviation, with motors having larger overall deviations receiving lower weights. The overall deviation is compensated by weighting based on the adjusted participation weights, and each motor is controlled collaboratively according to the dual closed-loop structure of speed outer loop and torque inner loop.

2. The method according to claim 1, characterized in that, The multi-source state information also includes image information acquired by the camera and point cloud information acquired by the lidar; The multi-source state information is preprocessed, and the preprocessing includes: Use a low-pass filter to filter out high-frequency signals; Outlier data points were removed by using sliding window mid-value filtering combined with statistical criteria. An adaptive Kalman filter is used to estimate the state based on the slope and longitudinal acceleration information. The processed information is mapped to a preset interval using a normalization method, wherein the mapping boundary of the normalization method is dynamically adjusted according to the slope information and the load distribution information.

3. The method according to claim 2, characterized in that, The process noise covariance of the adaptive Kalman filter is dynamically adjusted according to the following conditions: When the slope angle is greater than 7°, or the maximum relative deviation between the load on each axis and the average load is greater than 20%, or the longitudinal acceleration fluctuation amplitude exceeds the preset acceleration fluctuation threshold, the process noise covariance is increased.

4. The method according to claim 1, characterized in that, The virtual spindle speed reference value is generated as follows: The weighting of each motor is determined based on the load on the corresponding shaft of each motor. The base synchronous speed is obtained by multiplying the real-time speed of all motors by their corresponding weights, summing the results, and then dividing by the sum of all weights. Calculate the product of the speed compensation coefficient, gravitational acceleration, and the sine of the current slope angle, and then divide it by the effective radius of the wheel to obtain the slope compensation term; The reference value of the virtual spindle speed is obtained by adding the basic synchronous speed to the slope compensation term; The speed compensation coefficient ranges from 0.05 to 0.

15.

5. The method according to claim 1, characterized in that, The virtual spindle torque reference value is generated as follows: The total torque requirement is estimated based on the vehicle's total mass, the aforementioned slope information, and the rolling resistance coefficient. Divide the total torque requirement by the total number of motors to obtain the basic allocated torque; The maximum relative deviation between the load on each axis and the average load is calculated as the load distribution non-uniformity. The torque adjustment coefficient is determined based on the load distribution unevenness. The difference between the real-time torque of each motor and the average torque of all motors is multiplied by the torque adjustment coefficient, and then the basic distributed torque is added to obtain the virtual spindle torque reference value.

6. The method according to claim 1, characterized in that, When the speed deviation and torque deviation of each motor are combined into the comprehensive deviation of the motor, the speed deviation and the torque deviation are multiplied by different weighting coefficients and then summed, and the weighting coefficient of the speed deviation is greater than the weighting coefficient of the torque deviation. The weighting coefficient for the speed deviation ranges from 0.6 to 0.8, and the weighting coefficient for the torque deviation ranges from 0.2 to 0.

4.

7. The method according to claim 1, characterized in that, The overall weight (w_i) of each motor is determined by the following formula: in, The initial weights are determined based on the load on the corresponding shaft of the i-th motor. Let be the overall deviation of the i-th motor. This represents the maximum value of the combined deviation of all motors. The attenuation sensitivity coefficient has a value range of 0.3 to 0.

7.

8. The method according to claim 1, characterized in that, The dual closed-loop structure of the outer speed loop and the inner torque loop is as follows: The virtual spindle speed reference value is used as the given value of the speed outer loop, the real-time motor speed is used as the feedback value of the speed outer loop, and the output of the speed outer loop is used as the torque given intermediate value of the torque inner loop. The sum of the given torque intermediate value and the compensation term is used as the given value of the torque inner loop, and the real-time torque of the motor is used as the feedback value of the torque inner loop. The torque inner loop outputs motor drive commands.

9. The method according to claim 8, characterized in that, The compensation terms include a speed deviation feedforward compensation term and a torque balance compensation term: The speed deviation feedforward compensation term is the result of performing proportional, integral, and derivative operations on the speed deviation, multiplied by the comprehensive weight of the motor. The torque balance compensation term is the torque deviation multiplied by a preset balance ratio coefficient and then multiplied by the overall weight of the motor. The sum of the torque reference value, the speed deviation feedforward compensation term, and the torque balance compensation term is used as the reference value for the inner torque loop.

10. A multi-motor cooperative control system for a tunnel transport vehicle based on a virtual spindle, used to execute the method according to any one of claims 1 to 9, characterized in that, include: The data acquisition and signal processing module is used to acquire multi-source status information during the operation of the tunnel transport vehicle in real time, and to perform filtering, noise reduction and normalization preprocessing on the multi-source status information. The virtual spindle construction module, connected to the data acquisition and signal processing module, is used to generate virtual spindle speed reference values ​​and virtual spindle torque reference values ​​based on the preprocessed multi-source state information. The motor deviation calculation module, connected to the virtual spindle construction module, is used to calculate the speed deviation and torque deviation of each motor based on the virtual spindle speed reference value, the virtual spindle torque reference value, and the real-time speed and torque of each motor, and merge them into a comprehensive deviation. The motor weight adjustment module is connected to the motor deviation calculation module and is used to dynamically adjust the participation weight of each motor according to the comprehensive deviation of each motor, wherein the motor with the larger comprehensive deviation receives a lower weight. The collaborative control execution module is connected to the virtual spindle construction module, the motor weight adjustment module, and each drive motor, respectively. It is used to perform collaborative control on each motor according to the dual closed-loop structure of the speed outer loop and the torque inner loop based on the virtual spindle reference value and the adjusted participation weight, and output motor drive commands.

Images