An engine-motor integrated active damping control method based on principal component analysis and a motor controller

Abstract

The present disclosure relates to an engine-motor integrated active damping control method based on principal component analysis and a motor controller. Under the premise of not significantly increasing the hardware cost of the motor controller, the damping control is sunk to the inside of the motor controller to realize engine-motor integration, actively damping by using the high bandwidth advantage of the motor controller, and significantly reducing the damping waveform storage demand and the test bench calibration workload. The present scheme reduces the damping waveform storage demand and the test bench calibration workload by storing the principal component analysis waveform and the principal component function; realizes smooth damping in a continuous working condition range through interpolation reconstruction; realizes engine-motor integration by introducing the engine crankshaft position sensor signal into the motor controller to directly obtain the engine crankshaft angle in future crankshaft angle prediction.

Description

Technical Field

[0001] This disclosure relates to the field of electronic control of new energy hybrid power systems, and in particular to an integrated active vibration reduction control method and controller for engine-motor based on principal component analysis. Background Technology

[0002] In hybrid and range-extended powertrains, the engine is connected to the electric motor via a clutch, gears, or a rigid shaft. Under conditions such as start-stop, high-speed high-load power generation, and low-speed high-load operation, the engine generates significant periodic torque fluctuations, which, superimposed on the electromechanical coupling shaft, can cause the following problems: periodic oscillations in engine and motor speeds; amplification of torsional vibrations in the transmission system, coupled with vehicle NVH (noise, vibration, and harshness); perceptible vibrations, noise, and acceleration / shifting jerks by the driver; and fatigue and lifespan impacts on components such as the clutch, gears, and half-shafts over long-term operation.

[0003] Existing technology CN 113859216 A, a multi-condition active vibration reduction control method for hybrid power systems based on vibration reduction waveforms, can call offline calibrated vibration reduction waveforms in the vehicle controller according to the operating conditions and quickly inject compensation torque through the motor, achieving significant results in multi-condition vibration reduction. However, several key technical challenges remain in practical engineering applications: the vibration reduction control algorithm runs in the vehicle controller and is sent to the motor controller via CAN / FlexRay buses, resulting in a long control link. The delay caused by communication and task scheduling affects the phase alignment of the vibration reduction waveform with the engine angle; the multi-condition vibration reduction waveforms are stored in the form of a "whole waveform library," and the volume of the waveform library expands rapidly as the dimensions of operating conditions such as speed, load, and temperature increase, putting significant pressure on the controller's storage space and bench calibration workload; for continuous operating conditions outside the calibration grid, interpolation of whole waveforms for different operating conditions is usually performed, which makes it difficult to guarantee the stability and physical rationality of the waveform shape, and also lacks a systematic "model-based" description and online correction mechanism. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, the purpose of this disclosure is to construct an active vibration reduction control method that is adaptable to multiple operating conditions, can continuously interpolate, explicitly considers control delay, and has online correction capabilities, without significantly increasing the hardware cost of the motor controller. This method integrates vibration reduction control into the motor controller, achieving high-bandwidth active vibration reduction of the engine-motor integration, while significantly reducing the requirements for vibration reduction waveform storage and bench calibration workload.

[0005] To achieve the objectives of this disclosure, an integrated active vibration reduction control method for an engine-motor system based on principal component analysis is proposed. This method utilizes a current control loop or a speed loop to perform active vibration reduction control based on compensation torque. The compensation torque acquisition step includes: the motor controller directly acquires the real-time crankshaft angle via a physical line connected to the engine crankshaft position sensor; based on the current operating conditions, it looks up the principal component basis functions and calibrated principal component coefficients in the motor controller's memory; it then uses interpolation to obtain new principal component coefficients; based on the principal component basis functions and the new principal component coefficients, it reconstructs a one-cycle vibration reduction compensation waveform; and it stores the reconstructed parameters in an angle lookup table; the motor controller predicts the future crankshaft angle based on the estimated total control delay time, real-time crankshaft angle, and speed; and the motor controller looks up the compensation torque in the reconstructed vibration reduction compensation waveform table based on the predicted crankshaft angle.

[0006] In one embodiment of the above technical solution, the determination step of the principal component basis functions and calibrated principal component coefficients stored in the motor controller includes: sampling the vibration reduction compensation waveforms of each operating condition at equal intervals according to the crankshaft angle to obtain N sampling points, and constructing a waveform sample matrix based on the sampling results. Where M is the number of operating conditions; the sample matrix W is preprocessed by centering and then principal component analysis (PCA) is used to obtain... , The waveforms are for the K principal component basis functions. This is the calibrated principal component coefficient matrix corresponding to each working condition.

[0007] In one embodiment of the above technical solution, K is preferably in the range of 3 to 6.

[0008] In one embodiment of the above technical solution, the predicted crankshaft angle , The current engine crankshaft angle is read in real time by the motor controller. The engine angular velocity is read in real time by the motor controller. This is the total control delay time.

[0009] In one embodiment of the above technical solution, linear interpolation is performed based on the predicted crankshaft angle to obtain the crankshaft angle corresponding to the query compensation torque.

[0010] In one embodiment of the above technical solution, after obtaining the feedforward damping torque based on the compensation torque, the feedforward damping torque is corrected by using a feedback correction loop designed based on motor speed fluctuations. Then, the feedback correction torque and the feedforward damping torque are superimposed and then superimposed on the basic torque command of the motor, and transmitted to the current control loop.

[0011] In one embodiment of the above technical solution, the motor controller performs task-level control, reconstructs a one-cycle vibration reduction compensation waveform during low-frequency task cycles, and predicts future crankshaft angles and superimposed compensation torques by looking up tables during high-frequency task cycles.

[0012] A motor controller capable of active vibration reduction control was implemented based on any of the above methods.

[0013] Based on the above method, a motor controller system can also be obtained, comprising: a crankshaft angle acquisition module, configured to directly acquire the crankshaft angle via a physical line connected to an engine crankshaft position sensor; a vibration damping compensation waveform reconstruction module, configured to look up the principal component basis functions and calibrated principal component coefficients in the motor controller's memory based on the current operating conditions, obtain new principal component coefficients by interpolation of the calibrated principal component coefficients, reconstruct a one-cycle vibration damping compensation waveform based on the principal component basis functions and the new principal component coefficients, and store the reconstruction parameters in the form of an angle lookup table; a crankshaft angle prediction module, configured to predict the future crankshaft angle based on the estimated total control delay time, real-time crankshaft angle, and rotational speed; and a compensation torque acquisition module, configured to look up the compensation torque in the reconstructed vibration damping compensation waveform table based on the predicted crankshaft angle.

[0014] In one embodiment of the above technical solution, the motor controller further includes a feedback damping correction module, which is configured to obtain the feedforward damping torque based on the compensation torque, and then use the feedback correction loop based on the motor speed fluctuation to perform feedback correction on the feedforward damping torque. Then, the feedback correction torque and the feedforward damping torque are superimposed and then superimposed on the basic torque command of the motor, and transmitted to the current control loop or the speed loop for active damping control.

[0015] The beneficial technical effects of this disclosure are as follows: The active vibration damping control architecture, which previously relied on serial transmission of engine signals (ECU—VCU—MCU) and distributed control by the VCU and MCU, is now integrated into an engine-motor control system. The motor controller directly connects to the engine crankshaft position sensor signal and autonomously completes vibration damping control. Principal component analysis (PCA) is used to reduce the dimensionality of the multi-condition vibration damping compensation waveforms, compressing the massive waveform library into a small number of shared basis functions and condition correlation coefficients, thereby significantly reducing storage and calibration burdens. Explicit modeling of control delays is performed within the motor controller, and the phase advance of the compensation torque is achieved through crankshaft angle prediction, enabling more precise alignment of the compensation waveform with engine torque fluctuations in the angular domain. Based on the vibration damping compensation reconstructed by principal component analysis, a feedback correction loop based on motor speed fluctuations is introduced to improve adaptability to changes in operating conditions and component aging. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the hardware connection of the engine-generator (range extender) system architecture of a hybrid power system in one embodiment.

[0018] Figure 2 This is a schematic diagram of an offline test bench for acquiring torsional vibration waveforms of an engine-generator set, as described in one embodiment.

[0019] Figure 3 This is a flowchart of waveform dimensionality reduction and modeling for vibration reduction compensation based on principal component analysis in one implementation method.

[0020] Figure 4 This is a timing diagram of low-frequency and high-frequency tasks in a motor controller in one embodiment.

[0021] Figure 5 This is a schematic diagram of the online control process in one implementation method.

[0022] Figure 6 This is an overall flowchart of an integrated active vibration reduction control method for engine-motor based on principal component analysis in one implementation.

[0023] Figure 7 This is a schematic diagram of the ripple torque and vibration reduction compensation torque under typical operating conditions in one implementation method.

[0024] Figure 8 This is a comparison chart of speed fluctuations under typical operating conditions in one implementation method. Detailed Implementation

[0025] Terminology Explanation: Engine-motor integrated control: In this application, it refers to the direct connection of the engine crankshaft position sensor signal to the motor controller, which internally considers the dynamic characteristics of the engine and motor to implement torque distribution and active vibration reduction control, rather than the vehicle controller generating vibration reduction commands at the upper level.

[0026] Vibration reduction compensation waveform: Under a certain operating condition of the engine, the superimposed waveform of motor torque is designed to counteract the excitation of engine torque fluctuation on the electromechanical coupling shaft. It is usually defined in the crankshaft angle domain of the engine, and its amplitude and phase are opposite to the torque fluctuation.

[0027] Control delay and predicted crankshaft angle: Control delay refers to the time delay from sensor sampling and control calculation to the actual application of motor torque. In this invention, the angle increment corresponding to this delay is estimated by estimating the instantaneous angular velocity of the engine, thereby obtaining the predicted crankshaft angle at the actual moment when the compensation torque takes effect.

[0028] In existing technology, crankshaft position sensor signals are only connected to the engine controller (ECU). The ECU is responsible for collecting and processing crankshaft signals (calculating speed, angle, combustion phase, etc.), and then uploading the engine status information to the upper-level vehicle control unit (VCU) via the CAN bus. Vibration damping control logic is typically deployed in the VCU. The VCU looks up tables or calculates the damping compensation torque based on engine operating conditions, and then sends it down to the motor controller (MCU) via the bus. The MCU does not participate in the vibration damping compensation calculation based on crankshaft angle information; it only executes the compensation torque command (base torque + vibration damping compensation torque) issued by the upper-level controller. The main drawback of this approach is: (1) Long control link makes it difficult to accurately consider control delay: Most solutions deploy the vibration reduction control logic on the vehicle controller (VCU) or engine controller (ECU). The vibration reduction compensation torque needs to be transmitted to the motor controller via CAN or FlexRay. Through different task cycles, communication cycles and current loop responses, there is an uncertain delay between the compensation torque and the engine crankshaft angle, making it difficult to achieve high-precision phase alignment in the angle domain.

[0029] (2) The vibration reduction waveform library is huge and has high requirements for storage and calibration: under multiple speed, multiple load and multiple state working conditions, a whole vibration reduction waveform is stored separately for each working condition. The amount of waveform library data grows exponentially with the working condition dimension; the controller Flash capacity is limited and it is difficult to maintain the huge waveform library for a long time; in order to ensure accuracy, a lot of bench testing and waveform calibration are required, which is a lot of work and costly.

[0030] (3) Insufficient interpolation and modeling for continuous operating conditions: Non-calibrated operating conditions are usually interpolated by interpolating the complete waveforms of adjacent operating conditions, lacking a unified low-dimensional model description; interpolation may cause waveform shape distortion and unclear physical meaning; lack of systematic model update and online correction mechanism makes it difficult to adapt to factors that affect operating conditions and response, such as engine aging and changes in fuel quality.

[0031] (4) The high-speed control capability of the motor controller is not fully utilized: the main vibration reduction logic is in the VCU layer, and the vibration reduction bandwidth is limited by the control frequency of the VCU (usually tens of Hz). The motor controller only passively receives torque commands and cannot give full play to the high-frequency response advantage of the motor.

[0032] (5) Some current technologies believe that under coaxial conditions, the motor position sensor can be used directly to replace the crankshaft signal, but ignore the influence of engine combustion phase information and torsional elasticity on instantaneous angle consistency.

[0033] Although in some hybrid systems the engine and motor are typically coaxially connected, and the motor itself is equipped with a position sensor, the motor position sensor only reflects the mechanical or electrical angle of the motor rotor. It cannot distinguish the combustion phase in the engine's four-stroke working cycle, nor can it provide a crankshaft angle reference that is strictly aligned with engine cylinder events. Furthermore, in actual drive systems with torsional elasticity and backlash, there is a dynamic deviation between the motor rotor angle and the instantaneous angle of the engine crankshaft, which is particularly significant under torsional vibration conditions.

[0034] Based on this, this invention shortens the control link by introducing the engine crankshaft position sensor signal into the motor controller, providing an accurate and independent engine phase reference for active vibration reduction control, thereby improving the phase accuracy and robustness of vibration reduction compensation. In terms of specific control methods, this scheme performs principal component analysis on the vibration reduction compensation waveforms under multiple operating conditions, and reduces the dimensionality of the vibration reduction compensation waveforms based on principal component analysis to achieve waveform compression. In waveform reconstruction, an adaptive operating condition strategy is used to interpolate the principal component coefficients in the operating condition dimension, achieving smooth vibration reduction within a continuous operating condition range. Furthermore, a method for predicting future crankshaft angle calculations and compensation waveform values ​​that explicitly considers control delay is used to achieve advance phase compensation in the angle domain. Finally, a one-cycle vibration reduction compensation waveform is reconstructed through a low-frequency task cycle, and the compensation torque is superimposed by looking up a table at the predicted angle position during a high-frequency task cycle, achieving a balance between real-time performance and computational load. This scheme replaces the original method of storing complete waveforms according to operating conditions with a structure of "a small number of basis functions + a small number of coefficients per operating condition," reducing the storage requirements within the controller from hundreds of KB to tens of KB or even lower, adapting to the resource constraints of existing motor control MCUs.

[0035] The following description, in conjunction with the accompanying drawings, clearly and completely describes how the technical solution of this case is implemented. Obviously, the described embodiments are only a part of the embodiments of this case, and not all of them. Based on the embodiments in this case, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0036] See Figure 1This invention employs an integrated engine-motor active vibration reduction control architecture. The system includes an engine, an engine crankshaft position sensor, a motor position sensor, a motor controller (MCU), and a vehicle controller (VCU). The engine and the ISG motor are connected via a clutch, gears, or a rigid shaft. The engine crankshaft position sensor provides information on the engine crankshaft angle and speed. The motor position sensor, such as a rotary transformer or encoder, provides information on the motor rotor angle and speed.

[0037] In the integrated engine-motor active vibration reduction control architecture, the engine crankshaft position sensor signal is directly connected to the motor controller. The motor controller executes the vibration reduction control logic internally at a high frequency cycle, without the need for secondary transmission through the VCU. This shortens the control link, improves the vibration reduction bandwidth and phase accuracy, and reduces the time and calculation time of the two communication conversions from ECU to VCU to MCU. The engine ECU can still be used to complete the fuel injection and ignition control of the engine body. The vibration reduction control logic of concern in this invention is completed independently by the motor controller.

[0038] The active vibration reduction control method of this application will be described in detail below.

[0039] First, the test bench was used to test the torque waveform and phase on the electromyographic coupling shaft of the engine-motor system under multiple operating conditions. The principle of the same phase frequency and opposite amplitude was adopted to construct a vibration reduction compensation waveform dataset for multiple operating conditions.

[0040] See Figure 2 This diagram illustrates an offline test bench experiment for acquiring torsional vibration waveforms of an engine-generator set. The typical operating conditions selected on the engine-generator bench include, but are not limited to: the entire engine start-up process (several typical start-up speed trajectories); idling and low-speed high-load operating conditions; several high-speed, high-load, high-efficiency power generation operating conditions; and common key operating conditions such as start-stop, rapid acceleration, and smooth cruising. Under each operating condition, the data acquisition instrument collects: engine crankshaft angle and engine speed; motor rotor angle and motor speed; and electromechanical coupling shaft torque (which can be obtained through a torque sensor or high-precision torque estimation method; cost can be temporarily disregarded during the bench test).

[0041] The measured shaft torque signal is converted to the engine crankshaft angle domain to obtain a torque fluctuation waveform with one or more cycles. To counteract torque fluctuations, a corresponding damping compensation torque needs to be generated. The core construction principle is that "the compensation torque is the same as the fluctuating torque in frequency and phase, but opposite in amplitude," thus constructing the corresponding damping compensation waveform. The specific generation methods include: 1) Frequency domain adjustment: Based on wave spectrum analysis, directly construct the inverse spectrum and perform inverse Fourier transform; 2) Time domain optimization: Using the measured wave waveform as input, iteratively generate the optimal compensation signal through algorithms such as adaptive filtering; 3) Direct inversion: When the wave waveform can be accurately obtained, directly invert and superimpose it.

[0042] Secondly, based on the principal component analysis method, the vibration reduction compensation waveform under multiple working conditions is decomposed into principal component basis functions and principal component coefficient matrices, and the principal component basis functions and coefficient matrices are stored in the motor controller.

[0043] See Figure 3 The vibration reduction compensation waveforms for each operating condition are sampled at equal intervals according to the crankshaft angle. For example, the crankshaft angle (CA) from 0 to 720° is divided into N sampling points, and a waveform sample matrix is ​​constructed based on the sampling results. ,in The number of operating conditions is given. The sample matrix W is preprocessed by centering, and principal component analysis (PCA) is used to obtain: , The waveforms are for the K principal component basis functions. It is applicable to all operating conditions. This represents the principal component coefficient matrix corresponding to each working condition. K is determined based on preset indicators and their corresponding thresholds. The determined K principal component basis functions already contain most of the waveform energy, thus significantly reducing the data dimensionality without significantly reducing vibration reduction accuracy. For example, if the preset indicator is the cumulative variance contribution rate, combined with its corresponding preset threshold, the range of K is determined to be 3–6.

[0044] The principal component basis functions Φ and the principal component coefficients for each calibration condition are stored as calibration results, forming a vibration reduction compensation waveform model based on principal component analysis. Several principal component basis function waveforms are stored in the non-volatile memory of the motor controller. Principal component coefficients corresponding to each calibration operating point A calibration grid is constructed using engine speed and load characteristics as indexes; operating condition identification logic and parameters are also included. The original method of storing complete waveforms for each operating condition is replaced with a structure of "a few basis functions + a few coefficients per operating condition," reducing the controller's waveform-related storage requirements from hundreds of KB to tens of KB or even lower, thus adapting to the resource constraints of existing motor control MCUs. When adding new operating conditions, only waveform acquisition and principal component coefficient calculation are required; there is no need to store a complete waveform for each operating condition, saving bench testing and calibration time.

[0045] Next, based on the current operating conditions, the motor controller looks up the principal component basis functions and calibrated principal component coefficients in the motor controller's memory, uses interpolation to obtain new principal component coefficients, and reconstructs a one-cycle vibration reduction compensation waveform based on the principal component basis functions and the new principal component coefficients. The reconstructed parameters are stored in the form of an angle lookup table.

[0046] When the vehicle is running, the motor controller identifies the grid cell containing the current operating condition based on the current engine speed and load characteristics; it then performs bilinear interpolation or higher-dimensional interpolation on the principal component coefficients at the vertices of that grid cell to obtain a set of principal component coefficients corresponding to the current operating condition. The principal component coefficients obtained by interpolation are used to construct the compensation waveform, thereby obtaining the vibration reduction compensation waveform for continuous operating conditions and achieving smooth vibration reduction within the continuous operating range. Furthermore, interpolation in the principal component coefficient space results in a waveform that is smoother in shape and more physically plausible, avoiding shape distortion and instability caused by directly interpolating the entire waveform.

[0047] In this invention, the construction and retrieval of the vibration reduction compensation waveform table uses engine speed and load characterization parameters as operating condition indexes. Engine speed is calculated from the crankshaft angle signal collected by the crankshaft position sensor and is used to characterize the engine's operating frequency characteristics. The load characterization parameter characterizes the average load level experienced by the engine at the corresponding speed; it is not limited to a direct measurement value from a specific sensor, but can be obtained from equivalent parameters that reflect the engine's average output torque or load level.

[0048] In the range extender operating mode of the engine-generator set, the engine output power is mainly consumed by the electromagnetic load of the generator. At this time, the load characteristic can be obtained online by the motor controller based on the generator electromagnetic torque, power generation, or equivalent load torque calculated from the electric power and speed, which is used to reflect the average load level borne by the engine.

[0049] In parallel drive mode, the engine output torque and the motor torque work together on the transmission system and vehicle load. The load characteristics can be obtained from the actual engine torque, target torque, or relative load provided by the engine controller or vehicle controller, or from the equivalent engine load calculated from the vehicle's required torque using the transmission ratio and efficiency. These load characteristics are used to update the amplitude or principal component coefficients of the low-frequency vibration reduction compensation waveform and do not participate in high-frequency phase alignment calculations; therefore, they can be obtained via communication.

[0050] It should be noted that the load characterization parameters in this invention are only used for the operating condition index and parameter selection of the vibration reduction compensation waveform. Their function is to distinguish the changes in the intensity of engine torque fluctuations at different speeds, rather than for real-time reconstruction of the engine's instantaneous torque waveform. Therefore, in practical applications, appropriate load characterization parameters can be selected based on the system architecture and signal availability without affecting the effectiveness of the method of this invention.

[0051] Next, the motor controller estimates the total control delay time and predicts the future crankshaft angle based on the real-time crankshaft angle, the rotational speed, and the total control delay time.

[0052] The motor controller estimates the total delay time through calibration or online identification. The total delay time calculation includes sensor sampling, signal filtering, control calculation, and current loop response. The future effective angle is calculated based on the current engine angle and angular velocity. The value is taken from the vibration reduction compensation waveform at the predicted angle position to achieve advance phase compensation in the angle domain.

[0053] Predicted crankshaft angle And map it to the range of 0~720°CA or 0~360°CA. The current engine crankshaft angle is read in real time by the motor controller. The engine angular velocity is read in real time by the motor controller. This is the total control delay time.

[0054] This invention explicitly estimates the total delay time in the motor controller and predicts the angle value in the crankshaft angle domain, ensuring that the compensation torque takes effect near the target phase of engine torque fluctuations. Compared to schemes based solely on the current angle value, this results in more stable vibration reduction. The estimation of the total delay time can be achieved through offline calibration, online identification, or model-based estimation methods; as long as the delay between the measured torque and its actual application is explicitly modeled within the controller and predicted in the angle domain, it falls within the scope of this invention.

[0055] To achieve a balance between real-time performance and computational load, this invention adopts a task hierarchical structure of low-frequency reconstruction + high-frequency table lookup. In the low-frequency task cycle, a one-cycle vibration reduction compensation waveform is reconstructed, and in the high-frequency task cycle, compensation torque is superimposed based on the predicted angle position table lookup.

[0056] Combination Figure 4The task timing and implementation method are as follows. The task timing of the low-frequency (e.g., 50-100 Hz) reconstruction task is as follows: operating condition identification, principal component coefficient interpolation, one-cycle vibration reduction compensation waveform reconstruction, and storage of reconstruction parameters in the form of an angle lookup table. The task timing of the high-frequency (e.g., 1-2 kHz) lookup table task is as follows: real-time reading of engine crankshaft angle and angular velocity, calculation and prediction of future crankshaft angle, lookup value in the compensation waveform table, and superposition of compensation torque and motor basic torque. The cycles of the low-frequency and high-frequency tasks can be flexibly adjusted according to controller resources. The compensation waveform can be stored in the angle domain or calculated point by point in the high-frequency task; as long as the function of "parameter update in the operating condition dimension + high-frequency angle domain compensation value retrieval" is achieved, it is within the protection scope of this invention.

[0057] To address the varying requirements for operating condition determination and phase alignment in vibration reduction compensation, the controller employs different processing methods for crankshaft angle and speed signals: On one hand, it performs cyclic averaging or filtering on the engine speed signal to remove periodic fluctuations caused by the combustion process, thus characterizing the engine's current operating condition (low-frequency task); on the other hand, it retains the real-time crankshaft angle and instantaneous engine speed signals without cyclic averaging, reflecting the crankshaft's immediate motion state (high-frequency task). The engine speed signal can be obtained by the motor controller based on the crankshaft position sensor's acquisition of the engine crankshaft angle signal, calculated from the sampling interval between adjacent angles.

[0058] Based on the different processing results, the motor controller executes corresponding vibration reduction control functions in different calculation update cycles: it updates the morphological parameters of the vibration reduction compensation waveform using the averaged or filtered speed information, and predicts and corrects the effective phase of the compensation torque using real-time angle and instantaneous speed information. The specific implementation process is as follows.

[0059] The cyclic averaging or filtering of the engine speed signal is performed based on the engine's operating cycle: the instantaneous speed is averaged within one or more complete engine cycles, or a low-pass filter with a cutoff frequency lower than the combustion order frequency is used to effectively filter out speed fluctuations within the cycle, retaining only the slowly varying components that change with operating conditions; that is, low-frequency tasks are processed within this cycle. Since the averaging or filtering results change little within a single engine cycle, their update frequency does not need to be synchronized with the crankshaft angle and can be completed within a shorter calculation update cycle. Depending on the engine speed range and the controller's computing power, the calculation cycle for the speed averaging and vibration reduction compensation waveform morphology update can be set to the order of several milliseconds, corresponding to a task frequency of tens of hertz.

[0060] In the low-frequency operation cycle of the motor controller, such as the 50-100 Hz current loop control cycle, based on the principal component coefficients and principal component basis functions obtained by interpolation, a one-cycle vibration reduction compensation waveform is reconstructed in the crankshaft angle domain according to the following formula: And store it in the angle domain table in the form of a lookup table.

[0061] During high-frequency task cycles, such as the 1–2 kHz current loop control cycle, the motor controller reads the current engine crankshaft angle in real time. and engine angular velocity The predicted crankshaft angle corresponding to the moment when the compensation torque actually takes effect is calculated according to the following formula: And map it to the range of 0~720°CA or 0~360°CA.

[0062] In the reconstructed vibration reduction compensation waveform table, according to Retrieve the compensation torque from the corresponding index. This is then superimposed on the motor's base target torque command and output to the current control loop.

[0063] The above processing procedure can be found in [link / reference]. Figure 5 Through the above processing, explicit compensation for control delay is achieved in the crankshaft angle domain, so that the damping compensation torque can be more accurately aligned with the engine torque fluctuation in phase.

[0064] To address the discrepancy between the principal component analysis model and the actual torque fluctuations, this invention adds a feedback correction loop to the feedforward vibration reduction compensation. This includes: designing a feedback controller (such as a PI controller with a resonant element, a frequency-selective filter, etc.) targeting the main vibration frequency components based on engine and motor speed fluctuation signals, according to signal availability, and calculating a small-amplitude feedback correction torque; superimposing the feedback correction torque with the feedforward vibration reduction compensation torque before applying it to the target torque command of the motor, thus realizing a composite vibration reduction control structure with feedforward as the primary component and feedback as a fallback. The feedback correction loop helps improve robustness to factors such as changes in operating conditions and actuator aging. The feedback controller can adopt different forms such as repetitive control, resonant control, and state feedback. The observations can be based on motor speed, motor current, or estimated torque. As long as a small online correction is made around the vibration reduction compensation reconstructed from the principal component, it can be considered a reasonable variation of this invention.

[0065] In summary, this disclosure proposes an active vibration reduction control method applicable to multiple operating conditions of a hybrid power system, the steps of which are as follows: Figure 6The process includes: testing the torque waveform and phase on the electromechanical coupling shaft of the engine-motor system under multiple operating conditions on a test bench; constructing a multi-condition vibration reduction compensation waveform dataset using the principle of the same phase frequency and opposite amplitude; decomposing the vibration reduction compensation waveform under multiple operating conditions into principal component basis functions and principal component coefficient matrices based on principal component analysis; storing the principal component basis functions and coefficient matrices into the motor controller; identifying the current operating condition online based on the engine speed and load characteristics; interpolating the principal component coefficients for different operating conditions to obtain the principal component coefficients corresponding to the current operating condition; and obtaining the interpolated principal component coefficients. The vibration compensation waveform is reconstructed from the basis function into a vibration reduction compensation waveform in the crankshaft angle domain. The motor controller estimates the total delay time based on calibration. Based on the real-time read crankshaft angle, engine angular velocity, and estimated total delay time, the future crankshaft angle is predicted. Based on the predicted crankshaft angle, the compensation torque value is indexed in the reconstructed vibration compensation waveform table and superimposed on the motor's basic torque command. Using the speed fluctuation signals of the engine and motor, a feedback controller is designed to provide feedback correction to the feedforward vibration reduction torque. The feedback correction torque and the feedforward vibration reduction torque are superimposed on the motor's basic torque command and transmitted to the current control loop to achieve active vibration reduction control.

[0066] Figure 7 This is a schematic diagram of the fluctuation torque and vibration reduction compensation torque under the range-extended drive condition in an embodiment of the present invention. Figure 8 This is a schematic diagram comparing the speed fluctuations before and after vibration reduction under the extended-range drive condition in an embodiment of the present invention. As can be seen from the comparison, the vibration reduction control effect is obvious.

[0067] Based on the above embodiments, those skilled in the art can design a motor controller according to the active vibration reduction control method proposed in this disclosure. In one embodiment, the motor controller includes a crankshaft angle acquisition module; a vibration reduction compensation waveform reconstruction module; a crankshaft angle prediction module; a compensation torque acquisition module; and a current loop / speed loop control module. Some motor controllers also include a feedback vibration reduction correction module. The vehicle controller provides information such as basic torque requirements and engine start-stop commands. Specifically, the crankshaft angle acquisition module is configured to be directly acquired by the motor controller via a physical line connected to the engine crankshaft position sensor. The vibration reduction compensation waveform reconstruction module is configured to look up the principal component basis functions and calibrated principal component coefficients in the motor controller's memory based on the current operating conditions, obtain new principal component coefficients by interpolation of the calibrated principal component coefficients, reconstruct a one-cycle vibration reduction compensation waveform based on the principal component basis functions and the new principal component coefficients, and store the reconstruction parameters in the form of an angle lookup table. The crankshaft angle prediction module is configured to predict the future crankshaft angle based on the estimated total control delay time, real-time crankshaft angle, and speed. The compensation torque acquisition module is configured to look up the compensation torque in the reconstructed vibration reduction compensation waveform table based on the predicted crankshaft angle. The current loop / speed loop control module is configured to perform active vibration reduction control based on the compensation torque. In some embodiments, the motor controller also includes a feedback vibration reduction correction module, which is configured to, after obtaining the feedforward vibration reduction torque based on the compensation torque, use a feedback correction loop based on motor speed fluctuations to correct the feedforward vibration reduction torque, and then superimpose the feedback correction torque and the feedforward vibration reduction torque onto the motor's basic torque command, which is then transmitted to the current control loop or speed loop for active vibration reduction control.

[0068] The method of this invention is applicable to the powertrain controller (especially the motor controller MCU) and vehicle control system of hybrid / range-extended vehicles. This method can be extended to hybrid systems with different topologies, such as series, parallel, series-parallel, and range-extended hybrids, as long as there is torsional coupling between the engine and the motor.

[0069] The entire solution can be implemented on the existing motor controller hardware platform through software upgrades, without the need to add expensive cylinder pressure sensors or additional torque sensors. The hardware modifications are minor, the cost is controllable, and it is suitable for mass production applications.

[0070] The principal component analysis method in this scheme can be replaced by nonnegative matrix decomposition, wavelet basis functions, B-spline basis functions, or other data-driven basis function learning methods. As long as several representative basis functions can be extracted from the vibration reduction compensation waveform under multiple working conditions, and waveform reconstruction can be achieved with a small number of coefficients, it can be considered an equivalent variation of the present invention.

[0071] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein. The scope of the invention is defined by the appended claims.

Claims

1. A method for integrated active vibration reduction control of an engine-motor system based on principal component analysis, which utilizes a current control loop or a speed loop for active vibration reduction control based on compensated torque, characterized in that... The steps for obtaining the compensated torque include: The motor controller directly obtains the real-time crankshaft angle through a physical line connected to the engine crankshaft position sensor; Based on the current operating conditions, the principal component basis functions and calibrated principal component coefficients are obtained by looking up a table in the motor controller's memory. New principal component coefficients are obtained by interpolation of the calibrated principal component coefficients. Based on the principal component basis functions and the new principal component coefficients, the one-cycle vibration reduction compensation waveform is reconstructed, and the reconstructed parameters are stored in the form of angle lookup table. The motor controller predicts the future crankshaft angle based on the estimated total control delay time, real-time crankshaft angle, and speed. The motor controller looks up the reconstructed vibration reduction compensation waveform table based on the predicted crankshaft angle to obtain the compensation torque.

2. The method according to claim 1, characterized in that, The determination steps, including using the principal component basis functions stored in the motor controller and the calibrated principal component coefficients, include: The vibration reduction compensation waveforms for each operating condition are sampled at equal intervals according to the crankshaft angle to obtain N sampling points. A waveform sample matrix is ​​then constructed based on the sampling results. , where M is the number of operating conditions; The sample matrix W is preprocessed by centering, and principal component analysis (PCA) is used to obtain... , The waveforms are the basis functions of the K principal components. This is the calibrated principal component coefficient matrix corresponding to each working condition.

3. The method according to claim 2, characterized in that, The preferred range for K is 3 to 6.

4. The method according to claim 1, characterized in that, Predicted crankshaft angle , The current engine crankshaft angle is read in real time by the motor controller. The engine angular velocity is read in real time by the motor controller. This is the total control delay time.

5. The method according to claim 4, characterized in that, Linear interpolation is performed based on the predicted crankshaft angle to obtain the crankshaft angle corresponding to the query compensation torque.

6. The method according to claim 1, characterized in that, After obtaining the feedforward damping torque based on the compensation torque, the feedforward damping torque is corrected by using a feedback correction loop designed based on motor speed fluctuations. Then, the feedback correction torque and the feedforward damping torque are superimposed and then superimposed on the basic torque command of the motor, which is then transmitted to the current control loop.

7. The method according to claim 1, characterized in that, The motor controller performs task-level control, reconstructing the vibration reduction compensation waveform for one cycle during low-frequency task cycles, and predicting the future crankshaft angle and superimposing the compensation torque by looking up a table during high-frequency task cycles.

8. A motor controller, characterized in that, The motor controller implements active vibration reduction control using any one of the methods in claims 1 to 7.

9. A motor controller system, characterized in that, The motor controller system includes: The crankshaft angle acquisition module is configured to be acquired directly by the motor controller via a physical line connected to the engine crankshaft position sensor; The vibration reduction compensation waveform reconstruction module is configured to look up the principal component basis functions and calibrated principal component coefficients in the memory of the motor controller based on the current working conditions, use interpolation to obtain new principal component coefficients for the calibrated principal component coefficients, and reconstruct a one-cycle vibration reduction compensation waveform based on the principal component basis functions and the new principal component coefficients. The reconstruction parameters are stored in the form of angle lookup table. The crankshaft angle prediction module is configured to predict the future crankshaft angle based on the estimated total control delay time, real-time crankshaft angle, and speed. The compensation torque acquisition module is configured to look up the compensation torque in the reconstructed vibration reduction compensation waveform table based on the predicted crankshaft angle.

10. The motor controller system according to claim 9, characterized in that, It also includes a feedback damping correction module, which is configured to obtain the feedforward damping torque based on the compensation torque, and then use the feedback correction loop based on the motor speed fluctuation to correct the feedforward damping torque. Then, the feedback correction torque and the feedforward damping torque are superimposed and then superimposed on the basic torque command of the motor, and transmitted to the current control loop or speed loop for active damping control.

Images