Hybrid electric vehicle active suspension control method under road surface and switching excitation coupling effect

Abstract

The invention relates to the technical field of hybrid electric vehicle active suspension control, in particular to a hybrid electric vehicle active suspension control method under the road surface and switching excitation coupling effect. The road dynamic information rapid preview comprises the steps of excitation and vehicle response characteristic mapping quantity definition, binocular 3D visual identification discrete impact excitation, random road surface grade identification, continuous spectrum construction and rapid reconstruction of road time domain excitation in combination with a transformation method. According to the method, the optimal vehicle dynamic response characteristic quantity is selected through an improved distance evaluation technology, an intelligent algorithm is provided on this basis to realize accurate identification of road surface discrete impact and random road surface grades, road surface time domain excitation can be rapidly constructed in combination with a transformation method, and accurate information is provided for subsequent control input in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of active suspension control for hybrid vehicles, and particularly to an active suspension control method for hybrid vehicles under the coupled action of road surface and switching excitation. Background Art

[0002] Hybrid vehicles have the characteristics of good fuel economy, strong power output, and low dependence on infrastructure, and have now developed into one of the mainstream vehicle models in the automotive market. Different from traditional fuel vehicles, the powertrain of hybrid vehicles is composed of a complex electromechanical coupling system, and the engine and the motor jointly serve as the vehicle power source. During the driving process of hybrid vehicles, the vehicle body will be subjected to various vibrations, such as the unbalanced transient torque excitation caused by the driving mode or gear shift of the powertrain and the vibration transmitted to the vehicle body by the road surface roughness excitation. These vibrations affect the health and ride comfort of passengers and also have a negative impact on the components of the vehicle. Therefore, many researchers focus on the active suspension control system. By adopting appropriate control methods, the active suspension system can significantly improve the vibration control effect.

[0003] To ensure the stability of the control method, an effective vibration excitation signal acquisition and quantitative evaluation method is the key prerequisite. The vibration sources include both the excitation from the road surface and the transient impact and steady-state vibration of the vehicle internal powertrain. Therefore, in terms of vibration excitation signal acquisition, an effective road surface information preview method and a transient-steady state excitation coupling quantification method need to be proposed. For the evaluation of vibration signals, methods such as FFT, vibration OverallLevel, and the peak value of the original signal are often used. However, the vibration generated during the mode switching or gear shifting process in hybrid vehicles is a transient impact signal, and its characteristics cannot be accurately described by FFT and vibration OverallLevel. For the original signal peak value evaluation method, it is mainly based on the observation of the signal peak value and can be applied to the evaluation of quasi-transient impact signals, but there are subjective biases in the processing process, resulting in low processing accuracy. Therefore, how to couple and evaluate transient impact excitation and steady-state vibration signals still needs to be studied in depth.

[0004] In the aspect of active suspension control, relevant scholars have tried various control algorithms, including PID control, robust control, LQR control, and feedforward FxLMS control, etc. Among them, the feedforward FxLMS algorithm is widely used because of its strong ability to adapt to non-stationary responses and simple calculation. This algorithm obtains the reference signal through sensors and uses error monitoring to control the effect. However, if the spectrum of the reference signal does not match that of the vibration signal to be attenuated, the control effect may decrease or fail. In actual situations, there is a certain difference between the frequency of the vibration signal coupled by the powertrain and road excitation and the actual vibration frequency monitored by the error sensor, which affects the control effect of the feedforward algorithm. Therefore, it is still necessary to conduct in-depth research on selecting a more suitable control algorithm according to the broadband vibration characteristics at the vehicle body end. In order to effectively control the broadband interference unrelated to the powertrain order vibration frequency, and simultaneously control the order vibration related to the road preview reference signal frequency and the non-order vibration unrelated to it, an adaptive closed-loop control algorithm combined with preview information is designed, which can extract the reference signal from the internal and external coupled vibrations and error sensors, automatically adapt to different working conditions of the powertrain excitation and road excitation, and has the advantages of good robustness and high stability. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems existing in the prior art, and to propose an active suspension control method for a hybrid electric vehicle under the coupled action of road surface and switching excitation.

[0006] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0007] An active suspension control method for a hybrid electric vehicle under the coupled action of road surface and switching excitation, comprising the following steps:

[0008] S1. Rapid preview of road dynamic information, and the rapid preview of road dynamic information includes the definition of the mapping quantity of excitation and vehicle response characteristics, the identification of discrete impact excitation by using binocular 3D vision, the identification of random road surface grades and the construction of continuous spectra, and the rapid reconstruction of road time-domain excitation by combining transformation methods;

[0009] S2. Modeling of transient shocks in the hybrid system, and the modeling of transient shocks in the hybrid system includes the electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system, the analysis of the dynamic characteristics of the system during the mode switching process, the analysis of the system characteristics during the continuous gear shifting process, and the proposal of transient shock quantification indexes based on the transmission form;

[0010] S3. Quantification of steady-state vibrations in the hybrid system, and the quantification of steady-state vibrations in the hybrid system includes the identification of system elastic modes based on the multi-point response method, the analysis of the excitation of motor torque fluctuations under pure electric drive, the exploration of the dynamic response characteristics of the system under engine drive, and the quantification of steady-state vibration excitation under the combined drive mode;

[0011] S4. Adaptive control method for active mounts. The adaptive control method for active mounts realizes decoupling optimization of the mount structure based on excitation characteristics, proposes a dynamic performance model of the mounts using transfer characteristics, and formulates an adaptive closed-loop control algorithm by combining preview information.

[0012] Preferably, in the definition of the mapping quantity between the excitation and vehicle response characteristics, the vehicle response parameters to be used for identification are screened. Those with difficult measurement in some actual situations are removed from the training set. At the same time, since the vehicle suspension system is the main research object, the responses related to ride comfort are selected.

[0013] In the identification of discrete impact excitation using binocular 3D vision, combined with the identification requirement of the target position of the road speed bump, a binocular machine vision technical solution is selected to achieve active control of the mounts, avoid the problem of response lag when the vehicle passes over the speed bump, detect the target of the road speed bump ahead in real time, and send it to the controller through the CAN network, so as to adjust the mounts in advance.

[0014] Preferably, in the random road surface grade identification and continuous spectrum construction, after collecting the acceleration signal of the sprung mass of the vehicle, first perform low-pass filtering on the acceleration signal according to the sampling frequency and analysis requirements of the system; then use the filtered time-domain acceleration signal of the sprung mass as the input signal, select the optimal feature quantity using the distance evaluation method, calculate the value of the optimal feature quantity within a certain sampling time period, and calculate the variance, root mean square amplitude, root mean square value, and peak value within the sampling time period respectively; after obtaining the set of optimal feature quantities, input it into the classifier to obtain the road surface grade identification result; introduce the frequency-domain signal, classify it in parallel with the original time-domain signal, and calculate the mean value of the road surface grade identification output result as the road surface grade output result. A two-layer classifier is used, and the road surface grade identification output result of the first layer is used as the input of the second layer classifier, and the final road surface identification grade output is calculated.

[0015] To illustrate the road surface grade identification accuracy of the above three methods, a set H is defined to evaluate the accuracy of the identified road surface grade output y. If the output road surface grade y belongs to the set H corresponding to the actual input, the identification result is considered correct, otherwise the output is determined to be incorrect. The expression of the set H is:

[0016]

[0017] Preferably, in the rapid reconstruction of the road time-domain excitation by the combined transformation method, when constructing the road excitation of the vibration system, not only the road unevenness needs to be considered, but also the driving speed of the vehicle needs to be calculated. The spectral density function can be obtained according to the road random power spectral density. The random phase method takes the phase angle in [0, 2π], and then a time series of a certain length can be obtained according to the inverse discrete Fourier transform method. After obtaining the random road spectrum, using the inverse fast Fourier transform command, the time-domain signal can be quickly constructed. The formula for the road random power spectral density is:

[0018]

[0019] Preferably, in the electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system, a dynamic model of the power transmission system of the hybrid vehicle that can be used for steady-state and non-steady-state conditions is established, providing a theoretical support for the dynamic characteristic analysis of the power transmission system of the hybrid vehicle. It mainly considers the change of the power transmission path of the transmission system under non-steady-state conditions. By coupling the sub-model of the power system and the rigid-flexible coupling dynamic model of the dual-clutch automatic transmission, an electromechanical-rigid-flexible coupling dynamics model of the power system of the hybrid vehicle based on the P2 configuration is established.

[0020] Among them, to explore the influence of engine excitation on the dynamic characteristics of the system, an engine power output model needs to be established to provide a theoretical basis for the dynamic characteristic analysis of the transmission system. The formula for the engine power output model is:

[0021]

[0022] Among them: R is the crank radius, D is the diameter of the upper surface of the piston, m j is the mass doing reciprocating motion, λ is the connecting rod ratio, and the gas pressure in the upper part of the cylinder can be simulated by the normal distribution function;

[0023] According to the theorem of translation of force lines, the tangential force is transformed into the force and torque acting on the crank center position. The torque M makes the crankshaft overcome the external resistance torque and rotate, that is, the driving torque generated by one cylinder of the engine. Its formula is:

[0024]

[0025] The magnetomotive force of the permanent magnet of the motor and the armature magnetomotive force act together to generate the air-gap magnetic field. The air-gap magnetic field can generate the electromagnetic force acting on the inner surface of the stator core, the electromagnetic force and electromagnetic torque acting on the rotor, and then cause the vibration of the transmission system. To explore the influence of the motor electromagnetic excitation on the dynamic characteristics of the system, a permanent magnet synchronous motor power output model is established; first, according to the Maxwell stress tensor theory, the radial force density p r (θ, t) and the tangential force density p t (θ, t) are calculated. Their formulas are:

[0026]

[0027] The coordinate transformation matrix formula is as follows:

[0028]

[0029] Where F xn 、F yn and F sn are the electromagnetic forces in the x, y, and z directions respectively after being transformed into the stationary coordinate system;

[0030] The motor excitation force and the engine excitation force act on the stator and rotor of the motor and the crankshaft of the engine respectively. The crankshaft of the engine and the rotor of the motor are coupled through the torque of the one-way clutch and the torque of the friction clutch, thus forming a mechatronic-rigid-flexible coupling dynamics model of the hybrid vehicle power system.

[0031] Preferably, in the analysis of the system characteristics during the continuous gear shifting process, the duration of the unstable state of the bearing reaction force caused by the access of the engine driving torque to the transmission system during the gear shifting of the engine driving mode is longer. This is because the duration of the unstable state of the meshing torque of the load-bearing gear pair increases, resulting in an increase in the duration of the unstable state of the bearing reaction force. During the entire gear shifting process, the fluctuation of the bearing reaction force caused by the engagement of the synchronizer is the most severe, and it is stronger than the fluctuation caused by the engagement of the third-gear synchronizer in the pure electric driving mode. This is because when the third-gear and fourth-gear synchronizers are engaged, the angular velocity difference between the driving disk and the driven disk of the synchronizer satisfies the following formula:

[0032]

[0033] Where: Δw3 represents the angular velocity difference between the driving disk and the driven disk of the third-gear synchronizer, Δw4 represents the angular velocity difference between the driving disk and the driven disk of the fourth-gear synchronizer, w3 and w4 are the angular velocities of the power sources during the gear shifting of the pure electric driving mode and the engine driving mode respectively, and i1 to i8 represent the transmission ratios of the first gear to the seventh gear and the two main reducer gear pairs respectively.

[0034] Preferably, in the transient shock quantification index proposed according to the transmission form, during the operation of the hybrid vehicle, when the brake B2 is locked, the engine and the motor E2 participate in the power output; when the brakes B1 and B2 are not locked, the engine and the motors E1 and E2 jointly output power; when the vehicle is idling and the SOC value of the power battery is low, the engine starts to charge the power battery; in all the above cases, there is an engine start-stop process. When the brake B1 is locked, the planet carrier connected to the output shaft of the engine is fixed. At this time, the motors E1 and E2 work to drive the planetary gear, which is a pure electric driving mode and there is no engine start-stop process; during the power output process, the engine needs to start and stop according to the vehicle power demand, and the resulting start-stop transient shock vibration phenomenon is relatively obvious, and the ride comfort is poor;

[0035] Furthermore, a method for evaluating the transient vibration of engine start-stop using the vibration VDV value is proposed. The vibration measurement VDV value is the integral of the function of the fourth power of the impact acceleration over a certain period of time. The fourth power of the acceleration is more sensitive to the peak value than the square, effectively highlighting the influence of the peak value. Its value gradually increases over time and can more accurately evaluate the harm and risk caused by vibration and impact to the human body. It can be applied to the processing of transient vibration signals, with the unit of m / s 1.75 , and its expression is:

[0036]

[0037] where: a(t) is the instantaneous frequency-weighted acceleration; t0 and t1 are the start and end times of the measurement.

[0038] Preferably, in the identification of the elastic mode of the system based on the multi-point response method, the mount belongs to a continuous elastic body, which basically satisfies the conditions of linearity, time-invariance, and stability. After discretization, it can be simplified into a finite multi-degree-of-freedom system with a finite number of discrete lumped parameters, and its formula is:

[0039]

[0040] In the formula: (x), (f) are the displacement array, velocity array, acceleration array, and excitation array respectively; [M], [C], and [K] are the mass matrix, damping matrix, and stiffness matrix respectively, which constitute the modal mass matrix, modal stiffness matrix, and modal damping matrix of the system, all of which are diagonal matrices, and their formula is:

[0041]

[0042] When the system enters motion from its static equilibrium position without initial velocity, through Laplace transform, it is transformed into the complex plane of the Laplace variable s = -σ + jw, and its formula is:

[0043]

[0044] In the formula, X(s) and {F(s)} are the Laplace transforms of the dynamic response and excitation force of the system respectively, and its dynamic response formula is:

[0045]

[0046]

[0047] It is called the transfer function matrix of the system, which reflects the dynamic characteristics of the system. Subsequently, substituting the Laplace variable s = jw into it, we get:

[0048]

[0049] Compared with the existing technologies, the advantages of the present invention are as follows:

[0050] 1. A fast preview method for road dynamic information such as discrete impact excitation of the road surface and random road surface grades. By means of an improved distance evaluation technique, the optimal vehicle dynamic response characteristic quantities are selected. On this basis, an intelligent algorithm is proposed to achieve accurate identification of discrete impact of the road surface and random road surface grades. Meanwhile, combined with a transformation method, the road surface time-domain excitation can be quickly constructed to provide accurate information for subsequent control inputs in real time.

[0051] 2. A quantification method for the coupling of transient impact and steady-state vibration of the powertrain of a hybrid vehicle. Based on the electromechanical-rigid-flexible coupling dynamics model of the hybrid system, the generation mechanisms of transient impact and steady-state vibration of the powertrain under different working conditions are explored. On this basis, combined with the vibration transfer path analysis method, an excitation quantification method is proposed, which together with the road excitation forms an internal and external coupling excitation as the input of the control system.

[0052] 3. An active mount structure decoupling optimization method based on excitation characteristics. By considering the action of non-proportional damping and not considering the action of damping, the decoupling criterion when the generalized excitation force only excites one system mode is deduced. Using the lumped parameter model, a multi-objective optimization model of the active mount is proposed to achieve the decoupling optimization of the active mount.

[0053] 4. An active suspension adaptive closed-loop control algorithm combined with preview information. The proposed closed-loop algorithm makes full use of the road surface preview information and the coupling excitation of the powertrain, effectively suppressing the narrow-band and wide-band vibrations at the body end of the active mount, thereby improving the body vibration in real time. Meanwhile, the adaptive update method of the weight vector can effectively ensure the continuous stability of the update process. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic framework diagram of the active mount control method for a hybrid vehicle under the coupling action of the road surface and switching excitation proposed by the present invention;

[0055] Figure 2 It is the electromagnetic torque characteristic diagram of the motor in the present invention;

[0056] Figure 3 It is the comparison diagram of vibration acceleration before and after active mount control in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Refer to Figures 1-3 , a method for controlling an active suspension of a hybrid electric vehicle under the coupling action of a road surface and a switching excitation, comprising the following steps:

[0059] S1. Rapid preview of road dynamic information, where the rapid preview of road dynamic information includes the definition of the mapping quantity of excitation and vehicle response characteristics, the identification of discrete impact excitation using binocular 3D vision, the identification of random road surface grades, the construction of a continuous frequency spectrum, and the rapid reconstruction of the road time-domain excitation by combining the transformation method;

[0060] S2. Modeling of transient shocks in the hybrid system, where the modeling of transient shocks in the hybrid system includes the electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system, the analysis of the system dynamics characteristics during the mode switching process, the analysis of the system characteristics during the continuous gear shifting process, and the proposal of transient shock quantization indicators based on the transmission form;

[0061] S3. Quantification of steady-state vibrations in the hybrid system, where the quantification of steady-state vibrations in the hybrid system includes the identification of system elastic modes based on the multi-point response method, the analysis of the motor torque fluctuation excitation under pure electric drive, the exploration of the system dynamic response characteristics under engine drive, and the quantification of steady-state vibration excitation under the combined drive mode;

[0062] S4. Adaptive control method for the active suspension, where the adaptive control method for the active suspension realizes the decoupling optimization of the suspension structure based on the excitation characteristics, proposes a suspension dynamic performance model using the transfer characteristics, and formulates an adaptive closed-loop control algorithm by combining the preview information.

[0063] Specifically:

[0064] The first step is to define the mapping quantity of road excitation and vehicle response characteristics.

[0065] Some vehicle responses are not easily detected in a timely manner by sensors and other devices. Therefore, the vehicle response parameters to be used for identification should be screened, and those with greater measurement difficulty in some actual situations should be removed from the training set. At the same time, the vehicle's suspension system is the main research object, so the responses related to ride comfort should also be preferentially selected. The established seven-degree-of-freedom vehicle suspension model can obtain three types of responses, acceleration, velocity, and displacement responses. According to the evaluation index of the ride comfort of the suspension system, the acceleration-type responses can be the sprung vertical accelerations z s1 、z s2 、z s3 、z s4 measured by the acceleration sensors installed at the connection part between the top of the suspension and the body. Since there is a geometric quantity relationship between such accelerations and the vehicle's center-of-mass acceleration, the center-of-mass acceleration can be indirectly obtained, as well as the unsprung vertical accelerations z u1 、z u2, z u3 , z u4 ; The speed - type responses that can be used as inputs to the identification algorithm are the roll angular velocity and pitch angular velocity of the vehicle body measured by an installed gyroscope; while the displacement - type response can be the dynamic deflection f of the suspension measured by a wire - type displacement sensor d , and finally a total of 14 vehicle responses are adopted.

[0066] When considering specific input quantities, in order to meet the time requirements of preview, it is necessary to accurately identify the road surface in a shorter time. Therefore, responses are selected and discarded when considering specific input quantities. Thus, in order to reduce the number of responses in the input scheme, an improved distance evaluation technique is selected to screen representative design schemes. According to the screening principle, vehicle responses are designed as various factors, and the format of the selected screening table is L 32 (2 31 ). The response data is input according to the following steps: First, the 32 designed schemes are input one by one for training. By calculating the root - mean - square error and other processing of the recognition result data, and at the same time calculating the mean value of the correlation coefficient, then an analysis of variance is performed on the results to select the optimal - level factors. Finally, the optimal levels with small root - mean - square error and large correlation coefficient are found to form the best input scheme.

[0067] When analyzing problems with multiple indicators such as the correlation coefficient and root - mean - square error, the comprehensive balance method needs to be adopted. Calculate and analyze according to a single indicator to find the optimal combination of its factor levels. Through analysis and comparison, comprehensive balance is carried out based on the importance of each indicator, the influence degree of factors obtained in each indicator, the quality of levels, etc. to determine the overall optimal factor - level combination. According to the verification, the optimal scheme of the identification algorithm is the dynamic deflection of the suspension, the sprung vertical acceleration, and the vehicle pitch angular acceleration.

[0068] In the second step, use binocular 3D vision to identify discrete impact excitations.

[0069] Combined with the identification requirements for the target position of the road speed bump, a binocular machine vision technology solution is selected. In order to achieve active control of the mounts and avoid the problem of response lag when the vehicle passes over the speed bump, a 3D target detection algorithm based on binocular vision is developed to detect the target of the road speed bump ahead in real - time and send it to the controller through the CAN network, so as to adjust the mounts in advance. The main hardware devices are high - definition cameras and edge - computing devices.

[0070] The 3D vision detection algorithm adopted is FCOS3D, which is a fully convolutional and single-stage detector. It is a 3D object detection algorithm adjusted and optimized based on FCOS. In terms of network structure, the backbone of FCOS3D is replaced with ResNet101, and at the same time, the 2D and 3D features are decoupled in the detection head, which is beneficial to the efficient combination of the predicted 2D information and 3D information. The backbone network of the algorithm uses ResNet-101 (Residual Neural Network-101) to further extract the high-dimensional features of the image through feature layers such as C3, C4, and C5, and uses the FPN network to fuse the target features at different scales. In order to improve the detection accuracy of the network for small targets, the FPN network starts to fuse with the same-level features from the first layer, and at the same time uses the bilinear interpolation algorithm to adjust the features of different sizes. The FPN network outputs features of different sizes for object detection at different scales, and predicts each parameter of the object pose through multiple detection heads. Combining the characteristics of the system input and the detection target, only using parameters such as offset, depth, and rotation angle to describe the specific pose of the target can more efficiently realize the real-time detection of specific road targets and the customized input of the control system.

[0071] When a large-size image is used as the input, it will bring richer feature information to the model, thereby improving the detection accuracy. However, its demand for computing power will increase significantly, resulting in a decrease in the detection frame rate. Therefore, the TensorRT technology is used to accelerate the inference speed of the model. Detection experiments are carried out using input sizes of 1 and 0.5 respectively, and the Average Precision (AP) is used to evaluate the object detection accuracy of the model. According to the detection results, when the input size is 1200*1600, the detection accuracy is the highest, reaching 55.5%, but its detection speed is relatively low; when the input size is 600*800, the detection speed is greatly improved while the detection accuracy drops slightly. Considering comprehensively, the input size of 960*1280 is selected, and the algorithm has good detection performance for speed bumps in different scenarios such as single-lane and double-lane, basically meeting the requirements of real vehicle speed bump detection.

[0072] The third step, random road surface grade identification and continuous spectrum construction

[0073] After collecting the acceleration signal of the unsprung mass of the vehicle, first, according to the sampling frequency of the system and the analysis requirements, perform low-pass filtering on the acceleration signal. Then, use the filtered acceleration time-domain signal of the unsprung mass as the input signal, select the optimal feature quantity by using the distance evaluation method, and calculate the value of the optimal feature quantity within a certain sampling time period. Calculate the variance, root mean square amplitude, root mean square value, and peak value within the sampling time period respectively. After obtaining the optimal feature quantity set, input it into the classifier to obtain the road surface grade recognition result. Introduce the frequency-domain signal, perform classification in parallel with the original time-domain signal, and calculate the mean value of the road surface grade recognition output result as the road surface grade output result. Adopt a two-layer classifier, use the road surface grade recognition output result of the first layer as the input of the second layer classifier, and calculate to obtain the final road surface recognition grade output.

[0074] To illustrate the road surface grade recognition accuracy of the above three methods, define the set H to evaluate the accuracy of the recognized road surface grade output y (y r is the actual input road surface grade). If the output road surface grade y belongs to the set H corresponding to the actual input, it is considered that the recognition result is correct, otherwise it is determined to be an error output. The expression of the set H is:

[0075]

[0076] To verify the accuracy of the road surface recognition algorithm, generate eight-level road surfaces specified by the standard in sequence, set the vehicle speed to 10 m / s, collect each road surface for 10 seconds, and keep the road surface grade unchanged within 10 seconds, and collect a total of 80 seconds of road surface excitation signals. Build a simulation model, load the trained classifier into the model to verify the model accuracy. Combine the discrete impact recognition algorithm with the road surface grade identification algorithm, and the road surface information can be predicted from the time-domain data of the road surface and the response signal through the mapping relationship, eliminating the need for manual calibration and inverse models. Introduce the road surface excitation into the control strategy in the following text, so that the control system can not only respond to the impact excitation in a timely manner to avoid response lag, but also actively adjust the mounts for different conventional road surfaces, so that the control system can achieve good control effects under different road surface conditions.

[0077] The fourth step is to quickly reconstruct the road time-domain excitation by combining the transformation method

[0078] The IFFT (Inverse Fourier Transform) time-domain modeling method for road surfaces has the advantages of fast calculation speed and high simulation accuracy. Road surface unevenness is a stationary, ergodic random process with zero mean. When constructing the road surface excitation of a vibration system, not only the road surface unevenness needs to be considered, but also the driving speed of the vehicle needs to be calculated. Based on the road surface random power spectral density, the spectral density function can be obtained. The random phase method takes the phase angle in the range of [0, 2π]. Subsequently, a time series of a certain length can be obtained according to the discrete Fourier inverse transform method. After obtaining the random road surface spectrum, by using the inverse fast Fourier transform command, the time-domain signal can be quickly constructed. The formula for the road surface random power spectral density is:

[0079]

[0080] From the theory of rapid construction by IFFT, it can be obtained that the road surface unevenness power spectral density and the vehicle driving speed are the keys to establishing road surface excitation. Generally, vehicles are equipped with speed sensors, and the vehicle speed can be obtained in real time. Therefore, quickly obtaining the road surface unevenness power spectral density is the key to the road surface rapid construction method. The geometric mean values of the power spectral densities of different road surfaces are known. So, if the geometric mean value corresponding to the road surface grade is directly output under a certain vehicle response, the road surface excitation can be quickly obtained approximately. Although the accuracy and relevance are slightly lower than those of the identification method that directly outputs the road surface excitation, the speed can be greatly improved. In this way, it can help the preview controller to almost obtain the road surface excitation situation of the front wheels in real time, which plays an important role in improving ride comfort.

[0081] At the same time, due to the large number of response types and data volume, the identification process is slow. Therefore, the data volume of the response can be filtered before input. When the vehicle passes through different road surfaces at different speeds, the various responses generally have maximum values. And the extreme values often have the greatest impact on the vehicle attitude. Similarly, at a certain speed, the road surface type can generally be judged through the extreme values of the responses, and then the road surface unevenness power spectral density can be quickly obtained for road surface excitation construction. Based on the amplitudes of the vehicle response quantities at different speeds and road types obtained from a large number of simulations, the input of the identification algorithm becomes the response data that has been filtered a lot, and the output becomes the power spectral density that can be quickly used for road surface construction. Therefore, the identification process is simplified, and the data features are obvious. After training, the speed of road surface construction in actual use is greatly accelerated.

[0082] Step 5, electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system

[0083] There are multiple power sources inside a hybrid vehicle. When different power sources are driving, the dynamic response characteristics of the system are different; during the mode switching process, the system will generate shocks; during the shifting process under different driving modes, the dynamic response characteristics of the system are more complex. To fully reveal the dynamic response characteristics of the power transmission system of a hybrid vehicle under steady-state and non-steady-state conditions, it is necessary to establish a dynamic model of the power transmission system of a hybrid vehicle that can be used for steady-state and non-steady-state conditions, providing theoretical support for the dynamic characteristic analysis of the power transmission system of a hybrid vehicle. The core idea is to consider the change in the power transmission path of the transmission system under non-steady-state conditions, and establish an electromechanical-rigid-flexible coupling dynamic model of the power system of a hybrid vehicle based on the P2 configuration by coupling the sub-model of the power system and the rigid-flexible coupling dynamic model of the dual-clutch automatic transmission.

[0084] To explore the influence of engine excitation on the dynamic characteristics of the system, it is necessary to establish an engine power output model to provide a theoretical basis for the dynamic characteristic analysis of the transmission system. The formula for the engine power output model is:

[0085]

[0086] Where: R is the crank radius, D is the diameter of the upper surface of the piston, m j is the mass doing reciprocating motion, λ is the connecting rod ratio, and the gas pressure in the upper part of the cylinder can be simulated by a normal distribution function;

[0087] According to the theorem of translation of force lines, the tangential force is transformed into a force and a torque acting on the crank center position. The torque M causes the crankshaft to rotate against the external resistance torque, that is, the driving torque generated by one cylinder of the engine. Its formula is:

[0088]

[0089] The magnetomotive force of the permanent magnet of the motor and the armature magnetomotive force act together to generate an air-gap magnetic field. The air-gap magnetic field can generate electromagnetic forces acting on the inner surface of the stator core, electromagnetic forces and electromagnetic torques acting on the rotor, which will cause the vibration of the transmission system. To explore the influence of the electromagnetic excitation of the motor on the dynamic characteristics of the system, a dynamic output model of the permanent magnet synchronous motor is established; first, according to the Maxwell stress tensor theory, the radial force density p r (θ, t) and the tangential force density p t (θ, t) are calculated. Its formula is:

[0090]

[0091] The formula for the coordinate transformation matrix is:

[0092]

[0093] Where F xn 、F yn and F snThey are the electromagnetic forces in the x, y, and z directions after being transformed into the stationary coordinate system, respectively.

[0094] The motor excitation force and the engine excitation force act on the stator and rotor of the motor and the crankshaft of the engine respectively. The crankshaft of the engine and the rotor of the motor are coupled through the torque of the one-way clutch and the torque of the friction clutch, thus forming a dynamic model of the electromechanical-rigid-flexible coupling of the hybrid vehicle power system.

[0095] Step 6, Analysis of the dynamic characteristics of the system during the mode switching process

[0096] Taking the vehicle running in the first gear as an example, explore the dynamic response characteristics of the system during the process of switching from the pure electric drive mode to the engine drive mode. This mode switching process can be divided into three stages. Stage 1: The motor outputs torque externally, the engine does not output torque externally, the crankshaft speed of the engine is zero, and the vehicle runs in the pure electric drive mode. Stage 2: The friction clutch between the crankshaft and the rotor starts to slip under the action of the external pressure, and the crankshaft of the engine starts to accelerate from rest under the action of the frictional torque of the friction clutch. To ensure the stability of the input torque of the transmission, the motor needs to increase the output torque to offset the resistance torque received by the driven disk of the friction clutch fixedly connected to the motor rotor. Stage 3: After the crankshaft speed of the engine and the rotor speed of the motor are synchronized, the one-way clutch locks, the friction clutch disengages, the engine ignites, and starts to output torque externally. The driving torque of the motor becomes zero, and the engine drives the motor rotor and the transmission to rotate, and the vehicle enters the engine drive mode.

[0097] By plotting the time-domain diagram of the angular velocity of the transmission system components during the process of switching from the pure electric drive mode to the engine drive mode, explore the angular velocity fluctuations of the rotor and the crankshaft and the stable state of the system after the mode switching is completed. After the system starts to stabilize, due to the increase in the system speed, the changing trend of the angular velocity fluctuations of the rotor and the crankshaft after stabilization. Then calculate the time-domain diagrams of the torque of the driving disk of the odd-gear clutch and the meshing force torque of the driving wheel of the first-gear bearing gear pair during the mode switching process. In the pure electric drive mode, due to the small torque fluctuation of the motor, the amplitudes of the torque of the driving disk of the clutch and the meshing force torque of the first gear are both small. After switching to the engine drive mode, the amplitudes of the torque of the driving disk of the clutch and the meshing force torque of the first gear both increase first, then decrease, and then increase again. This is because after the mode switching is completed, the system will gradually tend to the stable state from the unstable state, resulting in the torque amplitude increasing first and then decreasing. And because the vehicle speed gradually increases, the torque amplitude increases again. The amplitude of the torque of the driving disk of the clutch is greater than the amplitude of the meshing force torque of the first gear, because the system damping can attenuate the vibration, resulting in a smaller torque amplitude at the rear end of the transmission chain. Therefore, in the process of designing the power system, it is necessary to consider the influence of the vibration noise caused by the knocking of the non-bearing gear pair on the input shaft during the mode switching process on the dynamic characteristics of the system.

[0098] Step 7, Analysis of the system characteristics during the continuous gear shifting process

[0099] Taking the vehicle shifting from first gear to second gear to third gear as an example, the dynamic characteristics of the shifting process system are explored. The shifting process can be divided into five stages, namely: (1) The vehicle is running in first gear, and the first and second gear synchronizers are both in the engaged state. (2) The clutch oil pressure of the odd gear gradually decreases, and the clutch oil pressure of the even gear gradually increases. When the oil pressure of the odd gear decreases to zero, the motor drive torque becomes zero, so that the active plate and the driven plate of the even gear clutch are quickly synchronized. After the speed is synchronized, the motor drive torque becomes the required torque, the clutch oil pressure of the even gear increases to the holding oil pressure, and the shifting from first gear to second gear is completed. (3) The vehicle is running in second gear, and the third gear synchronizer starts to pre-gear. After the synchronizer friction pair speed is synchronized, the third gear synchronizer is engaged to complete the pre-gear. (4) The clutch oil pressure of the even gear gradually decreases, and the clutch oil pressure of the odd gear gradually increases. When the oil pressure of the even gear decreases to zero, the motor drive torque becomes zero, so that the active plate and the driven plate of the odd gear clutch are quickly synchronized. After the speed is synchronized, the motor drive torque becomes the required torque, the odd-numbered gear clutch oil pressure rises to the holding oil pressure, and the second gear to third gear shift is completed. (5) The vehicle runs in third gear.

[0100] Compared with the shifting in pure electric drive mode, the unstable state of the bearing support reaction force caused by the engine driving torque connected to the transmission system in the engine driving mode shifting lasts longer. This is because the duration of the unstable state of the meshing torque of the load-bearing gear pair increases, which makes the duration of the unstable state of the bearing support reaction force increase. During the entire shifting process, the fluctuation of the bearing support reaction force caused by the synchronizer engagement is the most violent, and is stronger than the fluctuation caused by the third gear synchronizer engagement in the pure electric drive mode shifting. This is because when the third and fourth gear synchronizers are engaged, the angular velocity difference between the synchronizer active plate and the driven plate satisfies the following formula:

[0101]

[0102] Among them: Δw3 represents the angular velocity difference between the master and driven plates of the third gear synchronizer, Δw4 represents the angular velocity difference between the master and driven plates of the fourth gear synchronizer, w3 and w4 are the angular velocities of the power source for shifting in pure electric drive mode and engine drive mode, respectively, and i1 to i8 represent the transmission ratios of the first to seventh gears and the two main reducer gear pairs, respectively.

[0103] Step 8: Propose quantitative indicators of transient impact based on the transmission form

[0104] During the operation of a hybrid vehicle, when the brake B2 is locked, the engine and the motor E2 participate in power output; when the brakes B1 and B2 are not locked, the engine and the motors E1 and E2 jointly output power; when the vehicle is idling and the SOC value of the power battery is low, the engine starts to charge the power battery; in all the above cases, there is an engine start-stop process. When the brake B1 is locked, the planet carrier connected to the engine output shaft is fixed. At this time, the motors E1 and E2 work to drive the planetary gears, which is a pure electric driving mode and there is no engine start-stop process. During the power output process, the engine needs to start and stop according to the vehicle power demand, and the resulting start-stop transient impact vibration phenomenon is relatively obvious, and the ride comfort is poor.

[0105] Based on the above, a method for evaluating the engine start-stop transient vibration using the vibration VDV value is proposed. The vibration measurement VDV value is the integral of the function of the fourth power of the impact acceleration over a certain period of time. The fourth power of the acceleration is more sensitive to the peak value than the square, effectively highlighting the influence of the peak value. Its value gradually increases over time and can more accurately evaluate the harm and risk caused by vibration and impact to the human body. It can be applied to the processing of transient vibration signals, and the unit is m / s1.75 , and its expression is:

[0106]

[0107] where: a(t) is the instantaneous frequency-weighted acceleration; t0 and t1 are the start and end times of the measurement. In order to clearly and intuitively analyze the start-stop state of the engine and determine the time node of the engine start-stop, combined with the engine speed time-domain signal read by the CAN communication protocol, judge and identify the start and stop states of the engine, find out the time period of the engine start-stop, read the vibration data time-domain signal in the same time period, process the signal in this time history, obtain the VDV value in the start-stop state, and then analyze the start-stop characteristics of the engine and the impact of vibration on the vehicle interior.

[0108] Step 9: Identify the elastic mode of the system based on the multi-point response method

[0109] The mount belongs to a continuous elastic body, which basically satisfies the conditions of linearity, time-invariance, and stability. After discretization, it can be simplified into a finite number of discrete lumped parameter multi-degree-of-freedom systems. Its formula is:

[0110]

[0111] In the formula: (x), (f) are the displacement array, velocity array, acceleration array, and excitation array respectively; [M], [C], and [K] are the mass matrix, damping matrix, and stiffness matrix respectively, which constitute the modal mass matrix, modal stiffness matrix, and modal damping matrix of the system, all of which are diagonal matrices. Its formula is:

[0112]

[0113] When the system enters motion from its static equilibrium position without an initial velocity and is transformed into the complex plane of Laplace variable s = -σ + jω by Laplace transform, the formula is as follows:

[0114]

[0115] In the formula, X(s) and {F(s)} are the Laplace transforms of the dynamic response and excitation force of the system respectively, and its dynamic response formula is:

[0116]

[0117]

[0118] It is called the transfer function matrix of the system, which reflects the dynamic characteristics of the system. Subsequently, substituting the Laplace variable s = jω into it, we get:

[0119]

[0120] It can be seen from this that each element of the frequency response function matrix contains the modal parameters m i , k i , c i or ω i , σ i of each order of the system. Each row or column contains the modal vectors (x) of each order of the system. Therefore, when it is necessary to identify the natural frequency f 0i =(ω 0i =2πf 0i ), damping ratio ξ i (ξ i =σ i / ω 0i ), after obtaining any element H ef (ω) in the frequency response function matrix by combining multi-point responses, elastic modal analysis is carried out based on the frequency domain.

[0121] Step 10, Analysis of motor torque fluctuation excitation under pure electric drive

[0122] Power source excitation is one of the important internal excitations of the vehicle. Hybrid vehicles have different driving modes in different speed-torque ranges, and the influence of the power source on the dynamic characteristics of the system is different in different driving modes. To explore the influence of the power source on the dynamic characteristics of the system, the vehicle is driven in pure electric mode, and the dynamic response characteristics of the system in this mode are analyzed. First, a motor response model is constructed, and the electromagnetic torque characteristic diagram of the motor is as shown in Figure 2 . To facilitate the analysis of the frequency components of the spectrum, define f dDenoted as the fundamental frequency of the motor and f nd Denote the rotational frequency of the motor, and the defining formulas are respectively:

[0123]

[0124] When the vehicle runs in the pure electric drive mode, the engine does not output torque externally, the friction clutch and the one-way clutch are both disengaged, and the vehicle runs in the pure electric drive mode driven by the P2 motor. At this time, the torque fluctuation contains multiple frequency components, mainly including: the fundamental frequency component is related to the fundamental frequency of the motor and is the main power frequency component. The rotational frequency component is related to the rotational frequency and is the low-frequency vibration caused by rotor imbalance, slotting effect, etc. The high-frequency harmonics are the harmonics introduced by non-linear characteristics, switching control, etc., and need to be particularly concerned about in electromagnetic design and control strategies. On this basis, a mathematical model of the dynamic equation of the motor is constructed, including the voltage equation and the mechanical motion equation, which are respectively:

[0125]

[0126] Compared with not considering the motor torque fluctuation, when considering the motor torque fluctuation, the mean value of the meshing torque remains unchanged, but the amplitude increases. The main frequency components of the meshing torque are the first-gear meshing frequency and its multiples, and the amplitude corresponding to the double fundamental frequency of the motor is the largest, because the amplitude of the double fundamental frequency of the motor in the motor torque fluctuation is the largest. This shows that the vibration caused by the motor torque fluctuation is the main reason for the increase in the amplitude of the meshing torque of the load-bearing gear pair, and the influence of the motor torque fluctuation on the dynamic characteristics of the load-bearing gear pair should be fully considered in the design process.

[0127] The eleventh step, exploring the dynamic response characteristics of the system under engine drive

[0128] When the vehicle is in the engine drive mode, the engine outputs torque externally, the rotational speed of the inner shaft of the one-way clutch is greater than that of the outer shaft, the one-way clutch is locked, and the engine driving torque drives the motor rotor and the transmission to rotate. Define f e Denote the fundamental frequency of the four-cylinder engine, f ne Is the rotational frequency of the engine, and the defining formulas are respectively:

[0129]

[0130] To analyze the dynamic characteristics of the system under the engine drive mode in the steady state condition and explore the influence of the engine torque fluctuation and the torque fluctuation when the engine and the motor act simultaneously on the dynamic characteristics of the system, taking the third-gear engine drive condition as an example, a dynamic analysis of the dynamic characteristics of the transmission system is carried out. In the crank connecting rod mechanism, the piston motion is converted into the rotational motion of the crankshaft through the crank and the connecting rod, and the piston acceleration Is:

[0131]

[0132] where \(w\) is the angular velocity of the crankshaft. The instantaneous torque generated by the engine is the product of the piston force and the crank radius, taking into account the angles of the crank and the connecting rod. Based on this, the overall mechanical motion equation of the engine is calculated to describe the relationship between the engine output torque, the load torque, the damping coefficient, and the moment of inertia:

[0133]

[0134] where \(J\) is the overall moment of inertia. When considering the engine torque fluctuation compared to not considering it, the frequency components of the meshing torque increase by the fundamental frequency of the four-cylinder engine and its harmonics. Different from the meshing torque spectrum of the load-bearing gear pair, in the meshing torque spectrum of the non-load-bearing gear pair, the amplitudes of the fundamental frequency of the engine and its harmonics are not much different from the amplitude of the meshing frequency. This is because the non-load-bearing gear pair does not directly transmit the driving torque of the power source, and the vibration caused by the more fluctuating engine torque is more likely to attenuate when transmitted to the position of the non-load-bearing gear pair. Therefore, the engine torque fluctuation has a greater impact on the load-bearing gear pair, so the impact of the engine torque fluctuation on the load-bearing gear pair should be focused on during the design process.

[0135] Step 12, quantification of steady-state vibration excitation under the combined drive mode

[0136] When the vehicle is under combined drive, the engine outputs torque externally, and the motor compensates for the torque that the engine cannot provide to meet the driver's needs. Under the combined drive mode (the engine and the motor drive together), the vibration characteristics of the power system are complex. Vibration excitation may lead to a decline in system performance, a reduction in ride comfort, and even cause structural fatigue. Therefore, quantifying and analyzing these vibration excitations is crucial for designing the mount control strategy.

[0137]

[0138]

[0139]

[0140] To quantify the vibration excitation of the power system in the combined drive mode, it is first necessary to identify and analyze the vibration sources. By using Fourier transform to convert the time-domain signal to the frequency domain, the main vibration frequency components are identified. At the same time, through experiments and simulations, the modal parameters of the system are obtained to determine the natural frequencies and vibration modes of the system, and the possible resonance frequencies are identified. Through these analysis methods, the vibration characteristics of each component in the power system and their effects on the overall system can be clarified, thus identifying the main vibration sources. On the basis of vibration source identification, quantifying the intensity of vibration is the key to further analyzing the vibration excitation characteristics. Three quantification methods are proposed, including root mean square (RMS), peak value, and vibration acceleration level analysis, and their formulas are as follows:

[0141]

[0142] In addition, the analysis of attenuation characteristics is equally crucial, especially the transmission path and energy attenuation characteristics of vibration from the source to the target location. Transmission path analysis quantifies the vibration transmission characteristics by calculating the transfer function of the system. The transfer function can represent the relationship between the input and output signals to understand how vibration is transmitted through the system and its impact on key locations. The modal damping ratio is another important indicator for analyzing attenuation characteristics, reflecting the vibration attenuation of each mode to identify the high-risk vibration modes of the system. Combining these attenuation analysis results, the vibration resistance of the system can be effectively evaluated, and a theoretical basis can be provided for optimizing the suspension control strategy by adjusting system parameters to reduce the transmission and impact of vibration.

[0143] The thirteenth step is to achieve decoupling optimization of the mounting structure based on the excitation characteristics

[0144] Using the single-mode decoupling optimization idea, the generalized excitation force is made to excite at most one natural frequency to minimize the possibility of resonance and reduce the resonance frequency band. The independent variable y for optimization is the installation positions of each mounting point and the installation XYZ Euler angles [θ xi , θ yi , θ zi , with a total of 18 independent variables. When the influence of damping on decoupling is not considered, there will be only one objective function for the system; when damping is considered, the objective function becomes two. The independent variable y, the objective function g(y), and the inequality constraints and equality constraints can be uniformly written as: T When the influence of damping on decoupling is not considered, there will be only one objective function for the system; when damping is considered, the objective function becomes two. The independent variable y, the objective function g(y), and the inequality constraints and equality constraints can be uniformly written as:

[0145]

[0146]

[0147] u i (y) = u i i = 1, 2... d

[0148]

[0149] Wherein: w1 and w2 are weighting functions, g1(y) is an optimization criterion considering only stiffness, and g2(y) is an optimization criterion considering only damping. When w2 = 0, it is a decoupled optimization without considering the damping effect; when w1 ≠ 1 and w2 ≠ 0, it is a decoupled optimization considering the damping effect. Taking the weighting functions w1 = 1 and w2 = 0, the optimization problem is the optimization method of the torque shaft without considering the damping effect. By running, the optimal objective function value and the corresponding optimal independent variable optimization Θ i and optimization P i are obtained, and g1 = 1.0000 and g2 = 1.0000, that is, the purpose of exciting only one system mode by one excitation force is achieved.

[0150] Except for the mechanical part, the magnetic circuit is the core component of the active mount. Since the designed magnetic circuit structure is symmetric, 1 / 2 of the magnetic circuit structure is selected for optimization to reduce the calculation. In order to control the output force of the magnetorheological damper, it must be adjusted by changing the magnetic field strength. According to the required maximum magnetic field strength H, and combining Ampere's circuital theorem ∮Hdl = NI in the magnetic medium and B = μH, the equivalent relationship between the number of turns of the coil and the current is obtained. Then, considering various factors such as the cross-sectional area of the magnetic circuit and the part assembly condition, the relevant dimensional parameters of the magnetic circuit structure are determined.

[0151] The fourteenth step is to propose a dynamic performance model of the mount using the transfer characteristics

[0152] The active dynamic characteristics of the active mount refer to the frequency response characteristics with the voltage excitation applied to the actuator when both the powertrain end and the vehicle body end are fixed, with the voltage excitation as the input and the force received at the vehicle body end as the output. The transfer characteristic from the input voltage of the actuator to the force received at the vehicle body end is called the secondary channel transfer characteristic in the active mount control. Based on the mechanical model, considering the relationship between the active force f a (t) of the oscillating coil actuator and the excitation current i(t), i.e., f a (t) = Bli(t) = k M i(t), the mathematical model of the active mount is established as follows:

[0153]

[0154] When the active mount actuator is working, since the flow resistance of the liquid in the inertia channel increases, resulting in almost no flow, at this time, y2 = 0 can be set to obtain the mathematical model of the active mount in the middle and high frequency bands. Then, the intermediate variables p1 and f a are eliminated, and combining the relationship among the decoupling membrane line stiffness k3, the pump liquid piston area A3, and the dynamic volume stiffness K3, i.e., Obtain the transfer function expression of the secondary channel:

[0155]

[0156] The relevant parameters of the active mount identified by the parameter identification theory are used to obtain the theoretical frequency response curve of the active characteristics of the active mount. By comparing with the experimental data, the accuracy of the identification of the relevant parameters of the active mount and the correctness of the establishment of the relevant mathematical model are shown. The derived transfer function expression of the secondary channel of the active mount can be applied to the control algorithm of the active mount.

[0157] Step 15: Develop an adaptive closed-loop control algorithm in combination with preview information

[0158] The algorithm mainly includes three subsystems, namely the feedforward subsystem, the error signal separation subsystem, and the feedback subsystem. The feedforward subsystem is used to process the road surface preview excitation information and the powertrain order vibration signal. Its reference signals are the sine component x fs (n) and the cosine component x fc (n), and the corresponding filters are W fs (z) and W fc (z); the error signal separation subsystem is used for decoupling the error signal. During the convergence process of its filter, the feedforward and feedback subsystems respectively obtain their corresponding error signals. The residual vibration signal e(n) at the body end of the active mount is:

[0159] e(n) = d(n) + v(n) - s(n) * y(n)

[0160] where s(n) is the impulse response of the secondary channel S z , and the control signal y(n) input to the active mount is:

[0161] y(n) = y f (n) + y b (n)

[0162] where y f (n) is the control signal generated by the feedforward subsystem, and y b (n) is the control signal generated by the feedback subsystem, and they are respectively expressed as:

[0163]

[0164]

[0165] where x fs (n) and x fc (n) are respectively the sine component and the cosine component of the reference signal in the feedforward subsystem, and They are the weight vectors of the corresponding filters. x b (n) is the reference signal vector in the feedback subsystem, is the weight vector of the filter corresponding to the reference signal vector in the feedback system. The adaptive update formula can be expressed as:

[0166]

[0167] In the formula, by introducing the squared Euclidean norm of the vibration signal, the convergence coefficient in the weight update formula is normalized, which can effectively avoid the problem that when the reference signal is relatively large, the step size of the weight iteration is too large, or when it is too small, the correction term is too large, thus ensuring the continuous stability of the adaptive update process. The control result is as Figure 3 shown, and it can be seen that the vibration acceleration is effectively suppressed.

[0168] In the present invention, for the fast preview method of road dynamic information such as road surface discrete impact excitation and random road surface grade, the optimal vehicle dynamic response characteristic quantity is selected through the improved distance evaluation technology. On this basis, an intelligent algorithm is proposed to achieve the accurate identification of road surface discrete impact and random road surface grade. At the same time, combined with the transformation method, the road surface time-domain excitation can be quickly constructed to provide accurate information for the subsequent control input in real time.

[0169] The quantization method for the transient impact and steady-state vibration coupling of the powertrain of a hybrid vehicle, based on the electromechanical-rigid-flexible coupling dynamics model of the hybrid system, explores the generation mechanism of the transient impact and steady-state vibration of the powertrain under different working conditions. On this basis, combined with the vibration transfer path analysis method, an excitation quantization method is proposed, which together with the road excitation forms the internal and external coupling excitation as the input of the control system.

[0170] The decoupling optimization method for the active suspension structure based on excitation characteristics, by considering the action of non-proportional damping and not considering the action of damping, deduces the decoupling criterion when the generalized excitation force only excites one system mode, and using the lumped parameter model, proposes a multi-objective optimization model for the active suspension to achieve the decoupling optimization of the active suspension.

[0171] The active suspension adaptive closed-loop control algorithm combined with preview information. The proposed closed-loop algorithm makes full use of the road surface preview information and the coupling excitation of the powertrain, effectively suppresses the narrowband and broadband vibrations at the body end of the active suspension, thus improving the body vibration in real time. At the same time, the adaptive update method of the weight vector can effectively ensure the continuous stability of the update process.

[0172] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent replacements or changes should be covered within the protection scope of the present invention.

Claims

1. A control method for an active suspension of a hybrid vehicle under the coupling action of a road surface and a switching excitation, characterized in that, It includes the following steps: S1. Rapid preview of road dynamic information. The rapid preview of road dynamic information includes the definition of the mapping quantity of excitation and vehicle response characteristics, the identification of discrete impact excitation using binocular 3D vision, the identification of random road surface grades, the construction of continuous spectra, and the rapid reconstruction of road time-domain excitation by combining the transformation method; S2. Transient impact modeling of the hybrid system. The transient impact modeling of the hybrid system includes the electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system, the analysis of the system dynamics characteristics during the mode switching process, the analysis of the system characteristics during the continuous gear shifting process, and the proposal of transient impact quantification indexes according to the transmission form; S3. Steady-state vibration quantification of the hybrid system. The steady-state vibration quantification of the hybrid system includes the identification of the system elastic modes based on the multi-point response method, the analysis of the motor torque fluctuation excitation under pure electric drive, the exploration of the system dynamic response characteristics under engine drive, and the steady-state vibration excitation quantification under the combined drive mode; S4. Adaptive control method for active mounts. The adaptive control method for active mounts realizes the decoupling optimization of the mount structure based on the excitation characteristics, proposes the dynamic performance model of the mount using the transfer characteristics, and formulates the adaptive closed-loop control algorithm by combining the preview information.

2. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, characterized in that In the definition of the mapping quantity of excitation and vehicle response characteristics, the vehicle response parameters required for identification are screened. Some parameters that are difficult to measure in actual situations are removed from the training set. At the same time, the vehicle suspension system is the main research object, so the responses related to ride comfort are selected; In the identification of discrete impact excitation using binocular 3D vision, combined with the identification requirement of the target position of the road speed bump, the binocular machine vision technology solution is selected to realize the active control of the mount, and at the same time, the problem of response lag when the vehicle passes over the speed bump is avoided. The target of the road speed bump ahead is detected in real time and sent to the controller through the CAN network, so as to adjust the mount in advance.

3. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, characterized in that In the identification of random road surface grades and the construction of continuous spectra, after collecting the acceleration signal of the sprung mass of the vehicle, first, according to the sampling frequency and analysis requirements of the system, the acceleration signal is low-pass filtered; then, the filtered time-domain acceleration signal of the sprung mass is used as the input signal, and the optimal feature quantity is selected using the distance evaluation method. At the same time, the value of the optimal feature quantity within a certain sampling time period is calculated, and the variance, root mean square amplitude, root mean square value, and peak value within the sampling time period are calculated respectively; after obtaining the set of optimal feature quantities, it is input into the classifier to obtain the road surface grade recognition result; the frequency-domain signal is introduced and classified in parallel with the original time-domain signal, and the mean value of the road surface grade recognition output result is obtained as the road surface grade output result. A two-layer classifier is used, and the road surface grade recognition output result of the first layer is used as the input of the second-layer classifier, and the final road surface recognition grade output is calculated and obtained; To illustrate the road surface grade recognition accuracy of the above three methods, a set H is defined to evaluate the accuracy of the recognized road surface grade output y. If the output road surface grade y belongs to the set H corresponding to the actual input, the recognition result is considered correct, otherwise, the output is determined to be incorrect. The expression of the set H is:

4. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, wherein, In the fast reconstruction of road time-domain excitation by the combined transformation method, when constructing the road excitation of the vibration system, not only the road unevenness needs to be considered, but also the driving speed of the vehicle needs to be calculated. According to the road surface random power spectral density, the spectral density function can be obtained. The random phase method takes the phase angle in the range of [0, 2π]. Subsequently, a time series of a certain length can be obtained according to the inverse discrete Fourier transform method. After obtaining the random road surface spectrum, by using the inverse fast Fourier transform command, the time-domain signal can be quickly constructed. The formula for the road surface random power spectral density is as follows:

5. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, characterized in that, In the electromechanical-rigid-flexible coupling dynamics modeling of the hybrid system, a dynamic model of the power transmission system of the hybrid vehicle that can be used for steady-state and non-steady-state conditions is established, providing a theoretical support for the dynamic characteristic analysis of the power transmission system of the hybrid vehicle. It mainly considers the change of the power transmission path of the transmission system under non-steady-state conditions. By coupling the sub-model of the power system and the rigid-flexible coupling dynamic model of the dual-clutch automatic transmission, an electromechanical-rigid-flexible coupling dynamics model of the power system of the hybrid vehicle based on the P2 configuration is established. Among them, to explore the influence of engine excitation on the dynamic characteristics of the system, an engine power output model needs to be established, providing a theoretical basis for the dynamic characteristic analysis of the transmission system. The formula for the engine power output model is as follows: Where: R is the crank radius, D is the diameter of the upper surface of the piston, m j is the mass doing reciprocating motion, λ is the connecting rod ratio, and the gas pressure in the upper part of the cylinder can be simulated by the normal distribution function; According to the force line translation theorem, the tangential force is transformed into a force and a torque acting on the crank center position. The torque M makes the crankshaft rotate to overcome the external resistance torque, that is, the driving torque generated by one cylinder of the engine. The formula is as follows: The magnetomotive force of the motor's permanent magnet and the armature magnetomotive force act together to generate an air-gap magnetic field. The air-gap magnetic field can generate electromagnetic forces acting on the inner surface of the stator core, electromagnetic forces and electromagnetic torques acting on the rotor, which in turn cause vibrations in the drive system. To explore the influence of the motor's electromagnetic excitation on the dynamic characteristics of the system, a dynamic output model of the permanent magnet synchronous motor is established. First, according to the Maxwell stress tensor theory, the radial force density p r (θ, t) and the tangential force density p t (θ, t) are calculated, and their formulas are as follows: The formula for the coordinate transformation matrix is as follows: Among which F xn , F yn and F sn are the electromagnetic forces in the x, y, and z directions respectively after being transformed into the stationary coordinate system; The motor excitation force and the engine excitation force act on the stator and rotor of the motor and the crankshaft of the engine respectively. The crankshaft of the engine and the rotor of the motor are coupled through the one-way clutch torque and the friction clutch torque, thus forming an electromechanical-rigid-flexible coupling dynamics model of the power system of the hybrid vehicle.

6. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, wherein In the analysis of the system characteristics during the continuous gear shifting process, the unstable state duration of the bearing reaction force caused by the access of the engine driving torque to the transmission system during the engine driving mode shift is longer. This is because the unstable state duration of the meshing torque of the load-bearing gear pair increases, resulting in an increase in the unstable state duration of the bearing reaction force. During the entire shifting process, the fluctuation of the bearing reaction force caused by the engagement of the synchronizer is the most intense, and it is stronger than the fluctuation caused by the engagement of the third-gear synchronizer during the pure electric driving mode shift. This is because when the third-gear and fourth-gear synchronizers are engaged, the angular velocity difference between the driving disk and the driven disk of the synchronizer satisfies the following formula: Where: Δw3 represents the angular velocity difference between the driving and driven disks of the third-gear synchronizer, Δw4 represents the angular velocity difference between the driving and driven disks of the fourth-gear synchronizer, w3 and w4 are the angular velocities of the power sources during the pure electric driving mode shift and the engine driving mode shift respectively, and i1 to i8 represent the transmission ratios of the first to seventh gears and the two main reducer gear pairs.

7. The active suspension control method for a hybrid vehicle under the coupling action of road surface and switching excitation according to claim 1, characterized in that, In the transient shock quantification index proposed based on the transmission form, during the operation of a hybrid vehicle, when the brake B2 is locked, the engine and the motor E2 participate in power output; when the brakes B1 and B2 are not locked, the engine and the motors E1 and E2 jointly output power; when the vehicle is idling and the SOC value of the power battery is low, the engine starts to charge the power battery; in all of the above cases, there is an engine start-stop process. When the brake B1 is locked, the planet carrier connected to the engine output shaft is fixed. At this time, the motors E1 and E2 work to drive the planetary gears, which is a pure electric driving mode and there is no engine start-stop process; during the power output process, the engine needs to start and stop according to the vehicle power demand, and the resulting start-stop transient shock vibration phenomenon is relatively obvious, and the ride comfort is poor; Furthermore, a method for evaluating the transient vibration during engine start-stop using the vibration VDV value is proposed. The vibration measurement VDV value is the integral of the function of the fourth power of the impact acceleration over a certain period of time. The fourth power of the acceleration is more sensitive to the peak value than the square, effectively highlighting the influence of the peak value. Its value gradually increases over time and can more accurately evaluate the harm and risk caused by vibration and shock to the human body. It can be applied to the processing of transient vibration signals, with the unit of m / s 1.75 , and its expression is as follows: Where: a(t) is the instantaneous frequency-weighted acceleration; t0 and t1 are the measurement start and end times.

8. The active suspension control method for a hybrid electric vehicle under the coupling action of road surface and switching excitation according to claim 1, wherein, In the identification of the elastic mode of the system based on the multi-point response method, the mount belongs to a continuous elastic body, which basically satisfies the conditions of linearity, time-invariance, and stability. After discretization, it can be simplified into a finite number of discrete lumped parameter multi-degree-of-freedom systems, and its formula is: where: (x), (f) are the displacement array, velocity array, acceleration array, and excitation array respectively; [M], [C], and [K] are the mass matrix, damping matrix, and stiffness matrix respectively, forming the modal mass matrix, modal stiffness matrix, and modal damping matrix of the system, all of which are diagonal matrices, and their formulas are: When the system enters motion from its static equilibrium position without an initial velocity, through Laplace transform, it is transformed into the complex plane of the Laplace variable s = -σ + jw, and its formula is: X(s) in the formula 、 {F(s)} are the Laplace transforms of the dynamic response and the excitation force of the system respectively, and its dynamic response formula is: This is called the transfer function matrix of the system, which reflects the dynamic characteristics of the system. Subsequently, substituting the Laplace variable s = jw into it, we get:

Images