Multi-axis cooperative movement control system and method for front auxiliary frame

Abstract

The invention relates to the technical field of vehicle control, in particular to a front auxiliary frame multi-axis cooperative movement control system and method. Comprising a strain sensing module which is deployed at a key mechanical node of a front auxiliary frame and is used for collecting a strain signal generated by deformation of the frame in real time; the load prediction module is connected with the strain sensing module and is used for decoupling a multi-axis load torque component in the strain signal; the power distribution module is connected with the load prediction module and is used for generating a current instruction according to the load torque component and the motor efficiency characteristic; the quick response driving module is connected with the power distribution module, adopts a wide bandgap semiconductor device to construct an inverter circuit, and outputs a pulse width modulation signal to each shaft motor; the output end of the strain sensing module is connected with the input end of the load prediction module through a high-speed serial bus. The problem of vibration control failure caused by electromechanical coupling lag can be fundamentally solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle control, and in particular to a front subframe multi-axis cooperative movement control system and method. BACKGROUND

[0002] In a new energy vehicle multi-axis drive system, the dynamic deformation of the front subframe and motor control have a serious problem of mechanical-electrical coupling lag. The traditional control system adopts a mechanical sensing and electrical control separated architecture, and the strain signal needs to be converted through multiple stages before it can act on the motor, resulting in a system response delay of more than 10 ms. This delay is particularly prominent when the vehicle passes through an unpaved road or performs emergency braking: the high-frequency vibration energy of the vehicle frame cannot be inhibited in real time, causing the following technical defects: dynamic response mismatch: the mechanical deformation frequency is much higher than the control bandwidth, causing the torque command to lag in phase, exacerbating the torsional vibration of the vehicle frame; insufficient energy dissipation: the delay causes the vibration energy to be transmitted through the structure rather than dissipated by motor braking, accelerating the fatigue of the suspension components; cooperative control failure: the internal force coupling of each axis motor due to inconsistent response reduces the driving efficiency.

[0003] The prior art attempts to improve this problem by increasing the sampling rate of the sensor or optimizing the control algorithm, but neither of them breaks through the inherent delay bottleneck of the mechanical-electrical information link.

[0004] Based on the above problems, there is an urgent need for a technical solution that can achieve millisecond-level synchronization of frame deformation and motor control, and fundamentally solve the problem of vibration control failure caused by mechanical-electrical coupling lag. SUMMARY

[0005] The purpose of the present application is to solve the problems existing in the prior art, and a front subframe multi-axis cooperative movement control system is proposed, which comprises: A strain sensing module is arranged at the key mechanical nodes of the front subframe to collect strain signals generated by the deformation of the vehicle frame in real time; A load prediction module is connected to the strain sensing module to decouple the multi-axis load torque components in the strain signals; A power distribution module is connected to the load prediction module to generate current commands according to the load torque components and motor efficiency characteristics; A fast response drive module is connected to the power distribution module, which uses a wide bandgap semiconductor device to build an inverter circuit and outputs a pulse width modulation signal to each axis motor; The output end of the strain sensing module is connected to the input end of the load prediction module through a high-speed serial bus, the output end of the load prediction module is connected to the input end of the power distribution module through a real-time communication interface, and the output end of the power distribution module is connected to the control end of the fast response drive module through a digital isolation circuit.

[0006] Preferably, the strain sensing module comprises a distributed fiber optic strain sensor array; The distributed fiber optic strain sensor array is arranged at a torsional stress concentration area of the front subframe, and a sampling frequency of the distributed fiber optic strain sensor array is not less than 10 kHz.

[0007] Further preferably, the load prediction module is built-in with a modal decomposition algorithm; the modal decomposition algorithm separates the strain signal into vibration modal components corresponding to each motor shaft; the vibration modal components have a linear mapping relationship with the motor load torque.

[0008] Further preferably, the power distribution module stores a motor efficiency curve database; the motor efficiency curve database records copper loss and iron loss data of each motor at different current values; the power distribution module takes the minimum system total loss as an objective function to solve the optimal current command.

[0009] Further preferably, a calculation formula of the load prediction module for decoupling the i-th shaft load torque component is: ;

[0010] wherein, represents the load torque component of the i-th shaft at time t; represents a torque-strain conversion coefficient of the i-th shaft, which is obtained through finite element calibration; represents an original strain signal collected at time τ; represents an observation time window, and the value range is 50 ms-200 ms.

[0011] Further preferably, a formula for generating the optimal current command of the i-th shaft by the power distribution module is:

[0012] wherein, represents the optimal current command of the i-th shaft at time t; represents a motor torque constant of the i-th shaft; represents an energy efficiency weight factor; represents a total loss function of the i-th shaft at current I.

[0013] Further preferably, an adaptive pulse width modulation duty cycle calculation formula of the fast response driving module is:

[0014] wherein, represents a duty cycle command of the i-th shaft at time t; represents a current tracking error; Kp, Ki, Kd represent the proportional, integral, and differential gain coefficients, which are set by Lyapunov stability theory.

[0015] A front subframe multi-axis cooperative movement control method is applied to the front subframe multi-axis cooperative movement control system in any one of the preceding embodiments, and comprises the following steps: S1: Real-time acquisition of front subframe strain signals by a distributed optical fiber strain sensor array; S2: Decoupling of each axis load torque component in the strain signal by using a modal decomposition algorithm; S3: Dynamic distribution of each axis motor current instruction with the minimum system total loss as the optimization target; S4: Output of adaptive pulse width modulation signals by a wide bandgap semiconductor inverter circuit to drive the motor; S5: The modal decomposition algorithm performs frequency domain feature extraction and axis coupling coefficient matrix operation.

[0016] Further preferably, the step of dynamically distributing the current instruction comprises: querying a pre-stored motor efficiency curve database to obtain copper loss and iron loss parameters under the current working condition; constructing a multi-objective optimization function containing load tracking error and loss weight; and solving the optimal current instruction value by using a sequential quadratic programming algorithm.

[0017] Further preferably, the step of outputting the adaptive pulse width modulation signal comprises: real-time calculation of the tracking error of the current instruction and the actual current; estimation of the motor back electromotive force interference component by a nonlinear observer; feedforward compensation of the interference component to the duty ratio control law; and output of the phase voltage by the SiC MOSFET bridge arm.

[0018] Technical effects: The four-level closed-loop architecture of the strain sensing module, the load prediction module, the power distribution module, and the fast response driving module creatively solves the three technical problems of the disconnection between mechanical and electrical control in the background technology. First, the distributed optical fiber strain sensor array with 10 kHz high-frequency sampling can capture the dynamic deformation of the frame in real time, so that the delay of mechanical load change to electrical response is shortened to 1 / 10 of the inherent period of the mechanical system, and the response lag defect of the traditional open-loop control is completely overcome. Second, the load prediction module based on the modal decomposition algorithm improves the torque decoupling accuracy to a phase difference of less than 5°, which is significantly better than the 30° deviation of the traditional static coupling model. Finally, the wide bandgap semiconductor driving module cooperates with the digital isolation technology to realize stable control at a switching frequency of 200 kHz, breaking through the bandwidth limitation of silicon-based devices.

[0019] This scheme first realizes the full-link cooperative optimization of the front subframe from mechanical deformation sensing to power electronic execution, forming a mechatronic control system with adaptive characteristics. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The application is a front subframe multi-axis cooperative movement control system connection block diagram; Figure 2 The application is a front subframe multi-axis cooperative movement control method flow chart. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0022] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0023] The conventional technical solution has the following technical problems: the existing front subframe multi-axis control system adopts an open-loop control architecture, the mechanical load changes are disconnected from the electrical control link, which causes the system to be unable to sense the dynamic deformation of the frame in real time. This mechanical-electrical information disconnection causes three major technical defects: first, the torsional vibration suppression ability is insufficient under road impact conditions, and the high-frequency vibration energy of the frame cannot be dissipated in time; second, the dynamic response has significant delay, and the current command update period is much larger than the inherent period of the mechanical system; third, the coordination efficiency of each axis motor is low, and the energy distribution strategy does not consider the actual load distribution difference. More seriously, when the vehicle passes through an unpaved road, the conventional system is prone to control instability due to response lag.

[0024] Based on this, please refer to Figure 1The embodiment provides a front subframe multi-axis cooperative movement control system, which comprises a strain sensing module arranged at a key mechanical node of the front subframe and used for collecting strain signals generated by frame deformation in real time; a load prediction module connected with the strain sensing module and used for decoupling multi-axis load torque components in the strain signals; a power distribution module connected with the load prediction module and used for generating current instructions according to the load torque components and motor efficiency characteristics; and a fast response driving module connected with the power distribution module and adopting a wide-band semiconductor device to construct an inverter circuit and output a pulse width modulation signal to each axis motor. The output end of the strain sensing module is connected with the input end of the load prediction module through a high-speed serial bus, the output end of the load prediction module is connected with the input end of the power distribution module through a real-time communication interface, and the output end of the power distribution module is connected with the control end of the fast response driving module through a digital isolation circuit.

[0025] The scheme constructs a mechanical-electrical closed-loop control system through a four-stage module chain, the strain sensing module converts frame deformation into quantifiable electrical signals, and the mechanical blind control problem of the traditional system is solved. The load prediction module adopts a dynamic coupling algorithm to analyze multi-axis torque components and eliminate mutual interference between each actuator. The power distribution module fuses motor efficiency maps and real-time load requirements to generate current instructions that take into account dynamic response and energy optimization.

[0026] The wide-band semiconductor driving module ensures the integrity of the control signal through a digital isolation interface, and its nanosecond-level switching characteristics break through the physical limitations of traditional silicon-based devices to achieve a breakthrough in system bandwidth.

[0027] In particular, the high-speed serial bus adopts an LVDS differential transmission protocol, which can still maintain the stability of signal transmission in a strong electromagnetic interference environment; the real-time communication interface is based on a time-triggered Ethernet protocol, which ensures the determinism and timeliness of the control instructions; and the digital isolation circuit selects a capacitive coupling isolation chip with an insulation voltage of more than 5kV, which completely blocks the conduction of power ground noise to the control system.

[0028] The above technical scheme achieves the following effects: the conversion delay of mechanical deformation information to electrical control instructions is shortened to less than 1 / 10 of the inherent period of the mechanical system, so that the frame torsional vibration energy is actively suppressed in the initial stage; the real-time decoupling capability of multi-axis load torque ensures that each motor shaft accurately distributes driving energy according to actual requirements, avoiding local overload caused by the traditional average distribution strategy; and the cooperative design of the wide-band semiconductor device and the digital isolation technology enables the system to maintain stable control performance at a switching frequency of 200kHz, solving the electromagnetic compatibility problem caused by high-frequency switching.

[0029] The closed-loop architecture fundamentally changes the passive response mode of the traditional open-loop control and realizes the full-link cooperative optimization of the front subframe system from perception, decision-making and execution.

[0030] The conventional technical solution has the following technical problems: the conventional resistance strain gauge sensor has three technical limitations in the vehicle vibration environment: first, the sampling frequency is usually lower than 1 kHz, which cannot capture the high-frequency vibration mode of the vehicle frame; second, the position of the distributed points is determined depending on experience, which is difficult to cover the stress concentration area; third, the electromagnetic compatibility is poor, and the inverter switching noise is easily coupled to the measurement signal. These problems result in low signal-to-noise ratio of the input signal of the load prediction module, which seriously affects the precision of multi-axis collaborative control. Especially in the motor regenerative braking working condition, the measurement error of the conventional sensor can cause phase distortion of the torque decoupling result.

[0031] Therefore, the strain sensing module includes a distributed fiber strain sensor array; the distributed fiber strain sensor array is arranged at the torsional stress concentration area of the front subframe; and the sampling frequency of the distributed fiber strain sensor array is not lower than 10 kHz.

[0032] The scheme uses fiber Bragg grating technology to realize strain measurement, and each sensing unit works independently through wavelength coding, avoiding the crosstalk problem of conventional electrical sensors. The sensor array is topologically optimized according to the stress contour determined by finite element analysis, and key parts such as the connection point of the subframe and the suspension and the torque beam node are monitored. The sampling frequency of more than 10 kHz can completely capture the dynamic deformation characteristics of the vehicle frame in the 0-5 kHz vibration frequency band.

[0033] In specific implementation, the fiber sensing network uses wavelength division multiplexing technology, and a single optical fiber can be connected with up to 20 sensing points, greatly simplifying the wiring complexity. The optical signal demodulator has a built-in temperature compensation algorithm to eliminate the influence of environmental temperature changes on strain measurement.

[0034] The above technical solution achieves the following effects: high-frequency sampling capability completely retains the time-frequency domain characteristics of the dynamic deformation of the vehicle frame, providing high-fidelity original data for modal analysis; anti-electromagnetic interference characteristics ensure that pure strain signals can still be obtained in the high-speed switching environment of SiC devices; the optimized point arrangement strategy improves the capture efficiency of the monitoring network for strain changes in key parts by more than 3 times. Compared with the conventional point-type sensing scheme, the distributed architecture can reduce the number of sensors by 40% under the same accuracy requirement, significantly reducing the system complexity and maintenance cost.

[0035] The conventional technical solution has the following technical problems: the existing load prediction algorithm uses static inter-axis coupling coefficients, which cannot adapt to the dynamic working condition changes of the vehicle frame. This simplified process has three technical defects: first, it ignores the vibration mode coupling effect, and the prediction error increases sharply near the resonance frequency; second, it does not consider the time-varying stiffness characteristics caused by material creep; third, it lacks a compensation mechanism for temperature drift. These problems cause a significant phase difference between the torque decoupling result and the actual load, which seriously restricts the performance of multi-axis collaborative control.

[0036] Based on this, the load prediction module is built-in modal decomposition algorithm; the modal decomposition algorithm separates the strain signal into the vibration modal component corresponding to each motor shaft; the vibration modal component and the motor load torque have a linear mapping relationship.

[0037] This scheme solves the inherent defects of traditional static coupling model through real-time modal analysis technology. The algorithm first performs short-time Fourier transform on the original strain signal to extract the energy distribution of the characteristic frequency band; then, based on the modal participation matrix established by the vehicle frame finite element model, the mixed signal is decomposed into the modal component dominated by each shaft; finally, according to the pre-calibrated modal-torque conversion coefficient, a linear relationship between the vibration component and the load torque is established.

[0038] In particular, the algorithm introduces a recursive least squares method to update the conversion coefficient online, automatically compensating for the effects of material aging and temperature drift. During implementation, the modal database stores the modal parameters of the vehicle frame under 20 typical working conditions, and the optimal parameter set is automatically selected by the working condition recognition algorithm.

[0039] The above technical solutions achieve the following effects: dynamic modal decomposition controls the phase difference between the torque prediction result and the actual load within 5°, which is significantly better than the 30° deviation of traditional methods; the online parameter updating function adapts to the slow change of the vehicle frame stiffness with service time, extending the system calibration period to more than 1000 hours; the multi-condition modal database ensures that the algorithm maintains prediction accuracy under special conditions such as climbing, braking, and steering. This technology fundamentally solves the torque cross-interference problem caused by vibration coupling and provides reliable load information input for the power distribution module.

[0040] The traditional technical solutions have the following technical problems: the existing power distribution strategy has three technical limitations: first, it relies on a simplified motor loss model and does not consider the nonlinear characteristics of copper loss and iron loss; second, the optimization goal is single and cannot balance dynamic response and energy efficiency; third, fixed weight coefficients are difficult to adapt to complex working condition changes. These problems cause the system to frequently deviate from the optimal operating point in actual operation, resulting in serious waste of electrical energy and uneven motor temperature rise.

[0041] Based on this, the power distribution module stores a motor efficiency curve database; the motor efficiency curve database records the copper loss and iron loss data of each motor at different current values; the power distribution module takes the minimum total loss of the system as the objective function to solve the optimal current command.

[0042] The scheme constructs a full-parameter motor efficiency model, and accurately calculates the loss distribution of any working point through a three-dimensional interpolation algorithm. The database contains the complete loss characteristics of the motor in the temperature range of-40℃ to 125℃ and the load rate interval of 0-100%. Each working point is obtained by dynamic dynamometer calibration. The optimization algorithm uses Pareto front analysis to find the optimal compromise among dynamic response, energy efficiency, thermal balance and other multiple objectives.

[0043] In implementation, the real-time scheduler dynamically adjusts the optimization weight according to the working condition parameters such as vehicle acceleration and battery SOC, and realizes adaptive switching of the control strategy. In particular, the database is provided with an aging correction factor to automatically compensate for performance degradation according to the running time of the motor.

[0044] The above technical scheme achieves the following effects: the full-condition loss model makes the current distribution accuracy reach the motor MAP test standard, and the time proportion of the system average working efficiency entering the high efficiency area of the motor is improved to more than 85%; the multi-objective optimization algorithm improves the regenerative braking energy recovery rate by 50% compared with the traditional scheme under the premise of ensuring dynamic response; the thermal balance control function controls the temperature difference between the motors to be within 15℃, significantly prolonging the service life of the electric drive system. The technology realizes fine control of energy management and breaks through the technical dilemma of traditional schemes in energy efficiency and performance.

[0045] The traditional technical scheme has the following technical problems: the existing torque estimation method has three technical defects: first, the static transfer function is used, which cannot reflect the dynamic correlation between the differential characteristics of the strain signal and the load; second, the time correlation of historical data is ignored, resulting in prediction lag in transient conditions; third, there is no temperature compensation mechanism for strain-torque conversion coefficients. These problems cause obvious torque control deviation in the system during dynamic processes such as acceleration and braking.

[0046] Based on this, the calculation formula of the load prediction module for decoupling the i-axis load torque component is:

[0047] Wherein, represents the load torque component of the i-axis at time t; represents the torque-strain conversion coefficient of the i-axis; represents the original strain signal collected at time τ; represents the observation time window.

[0048] The scheme establishes a dynamic torque observation model through the integral operation of the second derivative of the strain signal. The second derivative processing highlights the acceleration component in the signal, accurately reflecting the load mutation characteristics; the time window integration smooths random noise and retains effective dynamic information. The conversion coefficient The parameters including material stiffness, sensor position sensitivity, etc. are determined by combining finite element analysis and physical calibration. In implementation, the system monitors the temperature of the vehicle frame in real time, and dynamically corrects the conversion coefficient through the temperature-stiffness relationship curve. The observation time window According to the current vehicle speed adaptive adjustment, a smaller value is taken at high speed to improve the response speed, and a larger value is taken at low speed to enhance the anti-interference ability.

[0049] The formula establishes the dynamic correlation between the strain signal and the load torque, and solves the phase lag problem of the static transfer function. Second-order differentiation : five-point central difference method is used for calculation, and the difference step is , which eliminates zero drift and retains 0-5kHz frequency band information. Time window integration : Take the first-order torsional period of the vehicle frame (usually 20-50ms), and reduce the spectral leakage by weighting with the Hanning window. Calibration by impact hammer test, apply known torque , adjust so that the formula output and error <3%. Temperature compensation: , is the thermal expansion coefficient of the material.

[0050] The effects achieved by the above technical solutions include: the dynamic torque observation model shortens the prediction response time under sudden load to within 5ms, completely following the inherent vibration period of the vehicle frame; the temperature compensation mechanism controls the conversion coefficient drift within ±2%, ensuring measurement consistency in the entire temperature range; the adaptive time window algorithm effectively suppresses measurement noise caused by road roughness while ensuring accuracy. The technology realizes real-time and accurate observation of mechanical load, and provides a feedback signal with high dynamic characteristics for the closed-loop control system.

[0051] The existing current distribution algorithm only considers the load torque tracking requirement, without considering the motor loss characteristics as an optimization target, resulting in mutual restriction between system energy efficiency optimization and dynamic response performance. Fixed weight coefficients cannot adapt to complex working condition changes, and are prone to current command oscillation under sudden load, while the regenerative braking energy recovery efficiency is limited.

[0052] Based on this, the formula for generating the optimal current command of the i-th axis by the power distribution module is: ;

[0053] Among them, represents the optimal current command of the i-th axis at time t; represents the motor torque constant of the i-th axis; represents the energy efficiency weight factor; Let I represent the total loss function of the i-th axis under current I.

[0054] This formula constructs a multi-objective optimization problem, which solves the problem that energy efficiency and dynamic performance are difficult to balance in traditional current distribution.

[0055] First item The L2 norm of the load tracking error reflects the actual torque. With motor output torque The sum of squared deviations. Among them... It is the first The torque constant of the shaft motor was calibrated through a no-load-locked rotor test.

[0056] Second item This is a penalty item for wear and tear. Including copper loss and iron loss ,in For junction temperature-dependent winding resistance, , The coefficients for hysteresis and eddy current losses are... denoted as magnetic flux density. It is a dynamic weighting factor that adaptively adjusts within the range of 0.1-1.0 based on battery SOC and acceleration requirements.

[0057] The preprocessed conjugate gradient (PCG) method is used for iterative solution. The preprocessing matrix is ​​taken as the diagonal element of the Hessian matrix to ensure convergence within a 5ms control period.

[0058] Inequality constraints The process is handled by the projection gradient method, where the boundary values ​​are determined based on the demagnetizing current of the motor's permanent magnet and the inverter's maximum current-carrying capacity.

[0059] The technical advantages of this scheme are reflected in the fact that by constructing a multi-objective optimization function that includes a load tracking error squared term and loss weighting, it achieves synergistic optimization of torque control accuracy and system energy efficiency. (Loss function) Accurate modeling of the nonlinear characteristics of motor copper and iron losses ensures that the current command is always within the globally optimal operating range. Energy efficiency weighting factor. Its dynamic adjustment function can adaptively balance response speed and energy loss according to operating conditions, maximizing regenerative braking energy recovery rate while ensuring dynamic performance. This formula fundamentally solves the technical contradiction of energy efficiency and response performance being mutually exclusive in traditional solutions, forming a closed-loop optimization system with adaptive characteristics.

[0060] It is worth mentioning that: the loss function The model was built using a combination of offline testing and online parameter identification, and includes a winding resistance temperature drift compensation term to ensure accuracy across the entire temperature range. Energy efficiency weighting factor. The value range of is determined through Lyapunov stability analysis to avoid system instability during optimization. The optimization solution process adopts the preprocessed conjugate gradient method to ensure that the optimal solution for n motor shafts is calculated within a 5ms period, meeting real-time requirements.

[0061] The technical effects achieved by the above embodiments include: the multi-objective optimization function enables the system to maintain a smooth transition of current command under sudden load conditions, effectively suppressing the overshoot phenomenon inherent in traditional PID control; the fine modeling of the loss model enables the operating points of each motor shaft to automatically avoid the low efficiency region, significantly reducing the heat loss of the system operation; the adaptive adjustment function of the weight factor ensures that the vehicle automatically prioritizes power performance or energy efficiency requirements under special conditions such as climbing and braking, realizing the intelligent evolution of the control strategy.

[0062] Traditional technical solutions suffer from the following problems: Conventional PWM control uses fixed gain parameters, making it difficult to cope with electromagnetic interference and resonance issues introduced by the high-speed switching of SiC devices. The current loop response speed is limited by the dead-time compensation accuracy of power devices, resulting in tracking lag in microsecond-level dynamic processes, which leads to a decrease in the accuracy of multi-axis collaborative control.

[0063] Based on this, the formula for calculating the adaptive pulse width modulation duty cycle of the fast response driving module is:

[0064] in, The instruction indicating the duty cycle of the i-th axis at time t; Indicates current tracking error; Represents the proportional, integral, and differential gain coefficients.

[0065] This formula enables high-precision current tracking driven by SiC devices, overcoming the shortcomings of traditional PID parameters that are fixed.

[0066] Proportional Term : With switching frequency Positive correlation, designed as Through the gate drive resistor Adjust the dynamic response.

[0067] Integral term : Pick ,in With a target bandwidth of 5kHz, For the dynamic inductance of the motor, an anti-saturation algorithm is used to limit the integral output within ± Within the range.

[0068] Differential term : , the switching noise is suppressed by a second-order Butterworth filter (cut-off frequency ) wherein is the winding resistance.

[0069] The motor impedance parameters are identified by a recursive least square (RLS) algorithm every 10 ms , and the conduction temperature , is reduced by 20% to avoid thermal runaway of the SiC device.

[0070] By dynamically adjusting the PID control parameters, the working characteristics of the SiC device at a switching frequency of 200 kHz or above are adapted, and the control mismatch problem of the traditional IGBT driving scheme in the high frequency domain is solved. The integral term uses an anti-windup algorithm to avoid current overshoot, and the differential term embeds a low-pass filter link to suppress switching noise interference. The control parameters are automatically adjusted by online identification of the system transfer function, ensuring that the best dynamic response can be maintained under different bus voltages and load inertias. The control law allows the current loop bandwidth to be increased to 5 kHz or above, enabling microsecond-level tracking of the nominal current command.

[0071] It is worth mentioning that the proportional gain is positively related to the switching frequency, which is dynamically adjusted by real-time monitoring of the SiC device junction temperature by FPGA; the integral time constant is adaptively updated according to the change in motor inductance parameters, avoiding a decrease in phase margin caused by parameter mismatch; the differential term introduces acceleration feedforward compensation to offset the back electromotive force disturbance when the load suddenly changes. The control algorithm is implemented in a digital signal processor, with the execution cycle strictly synchronized with the PWM carrier signal to ensure timing accuracy.

[0072] The technical effects achieved by the above embodiments include: the adaptive PID control law effectively suppresses the current ripple amplification phenomenon caused by the switching process of the SiC device, making the THD index better than that of the traditional scheme; the microsecond-level current tracking capability ensures the phase synchronization accuracy of multi-axis motors under high-speed commutation conditions; the parameter self-tuning function eliminates the need for manual debugging, significantly reducing system maintenance costs. The control algorithm is deeply designed in collaboration with the SiC device, breaking through the technical barriers in the fields of power electronics and motion control.

[0073] The existing control method processes mechanical signal acquisition, electrical control and driving execution as independent links, lacking a cross-domain collaboration mechanism. The conversion process from the vehicle frame strain signal to the motor current command involves multiple information losses, resulting in system response delay and inability to meet high dynamic condition requirements.

[0074] Based on this, please refer to Figure 2The embodiment provides a front subframe multi-axis cooperative movement control method, which is applied to the front subframe multi-axis cooperative movement control system and comprises the following steps. S1: collecting a front subframe strain signal in real time through a distributed fiber strain sensor array; S2: adopting a modal decomposition algorithm to decouple each axis load torque component in the strain signal; S3: dynamically distributing each axis motor current instruction with the minimum system total loss as an optimization target; S4: outputting an adaptive pulse width modulation signal to drive the motor based on a wide band gap semiconductor inverter circuit; S5: the modal decomposition algorithm performs frequency domain feature extraction and axis coupling coefficient matrix operation.

[0075] The technical effect of the scheme is that a full-link closed-loop control system from mechanical deformation sensing to power electronic driving is established, and information attenuation in the traditional hierarchical control architecture is eliminated. The modal decomposition algorithm accurately separates the vibration modal components corresponding to each motor axis by calculating the axis coupling coefficient matrix in real time, and provides high-fidelity input for load torque prediction. The total loss optimization target function considers the electrical system efficiency and mechanical dynamic performance, and realizes the best matching of cross-domain parameters. The wide band gap semiconductor driving scheme fully utilizes the high-speed switching characteristics of SiC devices, and the control delay is compressed to less than 1 / 10 of the mechanical system inherent frequency, which fundamentally avoids electromechanical coupling oscillation.

[0076] It is worth mentioning that: the distributed fiber sensor array adopts wavelength demodulation technology, the spatial resolution is 1cm, the sampling frequency is 20kHz, and the high-order vibration mode of the frame can be captured. The modal decomposition algorithm is realized in a digital signal processor, and the coupling coefficient matrix is updated every 5ms to adapt to the dynamic changes of the load. The total loss optimization process integrates the motor efficiency map and real-time temperature data to ensure the accuracy of the model parameters. The SiC driving circuit adopts double pulse test to calibrate the dead time, and the switching loss is reduced by 60% compared with the IGBT scheme.

[0077] The technical effects achieved by the above embodiment include: the full-link closed-loop control shortens the response time of the system to road impact to 1 / 5 of the traditional scheme, significantly improves the vehicle dynamic stability; the cross-domain optimization algorithm automatically balances mechanical vibration suppression and electrical system efficiency, prolongs the service life of key components; the collaborative design of SiC driving and control algorithm breaks through the bandwidth limitation of traditional electromechanical systems, and provides a new technical path for high-performance chassis control.

[0078] The traditional technical scheme has the following technical problems: the dynamic current distribution process depends on experience rules or simplified models, and cannot accurately reflect the nonlinear characteristics of motor loss. The fixed step optimization algorithm is easy to fall into local optimal solution under complex constraint conditions, resulting in that the current distribution result deviates from the global optimal working point.

[0079] Based on this, the step of dynamically allocating the current instruction includes: querying a pre-stored motor efficiency curve database to obtain copper loss and iron loss parameters under the current working condition; constructing a multi-objective optimization function containing load tracking error and loss weighting; and solving the optimal current instruction value by using a sequential quadratic programming algorithm.

[0080] The technical effects of this scheme are reflected in: the motor efficiency curve database is established by combining offline testing and online parameter identification, and records the loss characteristics under different current, speed and temperature conditions, providing accurate data support for optimization. The multi-objective optimization function organically combines the L2 norm of torque tracking error and the total loss term through a weight factor to form a judgment standard that takes into account dynamic performance and energy efficiency. The sequential quadratic programming algorithm uses a feasible direction method to handle inequality constraints, ensuring that the solution is always within the safe operating area of the motor, and achieves super-linear convergence speed through the Hessian matrix approximation, meeting the real-time control cycle requirements.

[0081] It is worth mentioning that: the motor efficiency curve database contains loss characteristic data in the temperature range of -40℃ to 125℃, and an independent data page is established every 10℃, and the matching data page is selected in real time through the junction temperature sensor. The multi-objective optimization function introduces a regularization term to prevent overfitting, and the weight factor is dynamically adjusted according to the vehicle speed and acceleration. The sequential quadratic programming algorithm is implemented in parallel computing in FPGA, and a 12-dimensional optimization problem can be solved in a single control cycle.

[0082] The technical effects achieved by the above embodiments include: the optimization algorithm based on the accurate loss model shifts the system average operating point to the high efficiency area, significantly reducing the temperature rise during long time operation; the sequential quadratic programming algorithm ensures that the global optimal solution is obtained under complex constraint conditions, avoiding the premature convergence problem of traditional gradient descent method; the cooperative design of database and optimization algorithm realizes the automatic adaptation of the control strategy to time-varying factors such as motor aging and parameter drift, greatly improving the system reliability.

[0083] The traditional technical scheme has the following technical problems: the voltage spikes and electromagnetic interference caused by high-speed switching of SiC devices deteriorate the signal-to-noise ratio of the current sampling signal, and traditional PID control cannot effectively suppress high-frequency disturbances. The motor back electromotive force forms strong disturbance during fast commutation, which destroys the stability of multi-axis cooperative control.

[0084] Based on this, the step of outputting the adaptive pulse width modulation signal includes: calculating the tracking error of the current instruction and the actual current in real time; estimating the motor back electromotive force disturbance component through a nonlinear observer; feeding forward the disturbance component to the duty ratio control law for compensation; and driving the SiC MOSFET bridge arm to output the phase voltage.

[0085] The technical effects of the scheme are embodied in that the nonlinear observer reconstructs the back electromotive force disturbance in real time based on the electromechanical energy conversion model of the motor through a sliding mode variable structure algorithm, and solves the estimation lag problem of the traditional linear observer under high dynamic conditions. The feedforward compensation channel and the feedback control law form a composite control architecture, the feedforward term offsets more than 90% of the back electromotive force disturbance, and the feedback term processes the remaining error and model uncertainty. The SiC MOSFET drive uses active Miller clamp technology to suppress bridge arm cross talk, and cooperates with the RC buffer circuit to absorb voltage spikes, ensuring that clean current sampling signals can still be obtained under high switching frequency. The scheme makes the anti-interference ability of the control system under 200 kHz switching frequency more than 5 times that of the traditional scheme.

[0086] It is worth mentioning that: the nonlinear observer operates in the alpha-beta coordinate system, avoiding the coupling error introduced by Park transformation, and the update frequency is synchronized with the PWM carrier. The feedforward compensation quantity is processed by low-pass filtering, and the cutoff frequency is set to 3 times the control bandwidth, which retains the useful frequency band and suppresses the switching noise. The SiC drive circuit adopts Kelvin source connection mode, reduces the gate loop parasitic inductance, and reduces the switching loss by 30% compared with the traditional layout.

[0087] The technical effects achieved by the above embodiments include: the composite control strategy shortens the recovery time of the current loop under the condition of back electromotive force mutation to within 10 mu s, ensuring the multi-axis phase synchronization accuracy; the nonlinear observer accurately identifies the back electromotive force waveform when the position sensor fails, realizing smooth switching of sensorless control; the SiC drive optimization design controls the electromagnetic interference level below the CISPR25 Class3 limit, meeting the most stringent vehicle-grade EMC requirements. The technical scheme provides key control protection for high-power-density electric drive systems.

[0088] The above is only a preferred embodiment of the present application, and does not limit the form of the present application in other forms. Any person skilled in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application still belongs to the protection scope of the present application.

Claims

1. A front subframe multi-axis cooperative movement control system, characterized by, The application relates to a vehicle front subframe active control system. The application comprises: a strain sensing module arranged at key mechanical nodes of the front subframe for collecting strain signals generated by the deformation of the front subframe in real time; a load prediction module connected to the strain sensing module for decoupling multi-axis load torque components in the strain signals; a power distribution module connected to the load prediction module for generating current instructions according to the load torque components and motor efficiency characteristics; a fast response driving module connected to the power distribution module, which adopts a wide-band semiconductor device to build an inverter circuit and outputs a pulse width modulation signal to each-axis motor; 2. The front subframe multi-axis cooperative movement control system according to claim 1, characterized in that, an output end of the strain sensing module is connected to an input end of the load prediction module through a high-speed serial bus, an output end of the load prediction module is connected to an input end of the power distribution module through a real-time communication interface, and an output end of the power distribution module is connected to a control end of the fast response driving module through a digital isolation circuit. The strain sensing module comprises a distributed optical fiber strain sensor array; 3. The front subframe multi-axis coordinated movement control system of claim 1, wherein, arrangement positions of the distributed optical fiber strain sensor array cover the torsional stress concentration area of the front subframe, and the sampling frequency of the distributed optical fiber strain sensor array is not lower than 10 kHz.

4. The front subframe multi-axis coordinated movement control system of claim 1, wherein, The load prediction module is provided with a modal decomposition algorithm; the modal decomposition algorithm separates the strain signals into vibration modal components corresponding to each-axis motor; the vibration modal components have a linear mapping relationship with the motor load torque.

5. The front subframe multi-axis coordinated movement control system of claim 1, wherein, The power distribution module stores a motor efficiency curve database; the motor efficiency curve database records the copper loss and iron loss data of each motor under different current values; and the power distribution module takes the minimum system total loss as an objective function to solve the optimal current instruction. ； wherein, represents the load torque component of the ith axis at time t; represents the torque-strain conversion coefficient of the ith axis, obtained by finite element calibration; represents the original strain signal collected at time τ represents the observation time window, with a value range of 50ms-200ms.

6. The front subframe multi-axis coordinated movement control system of claim 4, wherein, The calculation formula of the decoupled i-axis load torque component of the load prediction module is: ； wherein, represents an optimal current command of the i-th axis at time t; represents a motor torque constant of the i-th axis; represents an energy efficiency weight factor; represents a total loss function of the i-th axis at current I.

7. The front subframe multi-axis coordinated movement control system of claim 1, wherein, The formula for generating the optimal i-axis current instruction of the power distribution module is: ； wherein, denotes the duty cycle command of the i-th axis at time t; denotes the current tracking error; denotes the proportional, integral, derivative gain coefficients, tuned by Lyapunov stability theory.

8. A front subframe multi-axis cooperative movement control method, applied to the front subframe multi-axis cooperative movement control system according to any one of claims 1-6, characterized in that, The adaptive pulse width modulation duty cycle calculation formula of the fast response driving module is: The application comprises the following steps: S1: collecting the strain signals of the front subframe in real time through the distributed optical fiber strain sensor array; S2: decoupling the load torque components of each axis in the strain signals through the modal decomposition algorithm; S3: dynamically distributing the current instructions of each-axis motor by taking the minimum system total loss as an optimization target; S4: outputting the adaptive pulse width modulation signal to drive the motor based on the wide-band semiconductor inverter circuit; 9. The front subframe multi-axis cooperative movement control method according to claim 8, characterized in that, S5: the modal decomposition algorithm performs frequency domain feature extraction and axis coupling coefficient matrix operation.

10. The front subframe multi-axis cooperative movement control method of claim 8, wherein, The step of dynamically distributing the current instructions comprises the following steps: querying the pre-stored motor efficiency curve database to obtain the copper loss and iron loss parameters under the current working condition; constructing a multi-objective optimization function containing a load tracking error and loss weight; and solving the optimal current instruction value by using a sequential quadratic programming algorithm. The step of outputting the adaptive pulse width modulation signal comprises the following steps: calculating the tracking error of the current instruction and the actual current in real time; estimating the motor back electromotive force interference component through a nonlinear observer; feeding forward the interference component to the duty cycle control law for compensation; and driving the SiC MOSFET bridge arm to output a phase voltage.

Images