Active suspension control method and system based on rolling driving style recognition

Abstract

The application provides a kind of active suspension control method and system based on rolling driving style recognition, comprising: obtaining brake, accelerator pedal opening, steering wheel angle, vehicle speed information from CAN bus to construct working condition information base;Pedal opening information, steering wheel angle information are extracted and compared with preset opening threshold, to identify the current driving condition;Extract historical vehicle speed information, and perform rolling cutting on historical vehicle speed trajectory based on time window to obtain vehicle speed time series;Apply mean and standard deviation to vehicle speed time series to extract statistical features, apply EMD-DFT joint extraction to extract time sequence features, and identify driving style based on the two features;Driving condition and driving style are used as input to realize active suspension control.The application realizes accurate identification of driving style, introduces rolling recognition based on time window, improves the real-time performance of suspension control, and finally improves the comfort and steering stability of suspension control by combining human-vehicle factors.

Description

Technical Field

[0001] This invention relates to the field of automotive technology, and more specifically, to an active suspension control method and system based on rolling driving style recognition. Background Technology

[0002] High-quality vehicle development is an urgent need both domestically and internationally, with driving comfort and handling stability receiving increasing attention. As a crucial component connecting the vehicle body and powertrain, the automotive suspension system directly impacts driver comfort and overall vehicle stability during handling, the latter being directly related to driving safety.

[0003] Suspension control technology is mainly divided into two categories: semi-active suspension control and active suspension control. Semi-active suspension is a passive control system, which cannot effectively control the stiffness and damping of the suspension during vehicle start-up, steering, and braking. However, it can automatically adjust the damping parameters according to changes in vehicle vibration during operation. Active suspension is an active control system that can autonomously adjust the suspension stiffness and damping under various vehicle operating conditions. Currently, regardless of the type of suspension control technology, most research neglects the driver's personal factors.

[0004] Patent document CN110901325B (application number: CN201911197599.1) discloses an active suspension control method and system, relating to the field of vehicle suspension technology. This method observes the unevenness curves of the road ahead, presets the desired vehicle body posture, and continuously compensates for the vehicle body posture during driving, thereby improving vehicle ride comfort. While this method incorporates road surface unevenness and combines vehicle-environment information for suspension control, it fails to consider the different definitions and needs of comfort among different drivers.

[0005] Patent document CN116176197B (application number: CN202111421716.5) discloses a semi-active suspension control method and device, a storage medium, and a vehicle. This method acquires the sprung and unsprung speeds of the suspension springs, calculates their required damping, and finally uses a control circuit to control the shock absorbers. This method requires only a small amount of input data to optimize suspension control and is low-cost; however, it also fails to consider the importance of the driver, resulting in limited control effectiveness. Summary of the Invention

[0006] In view of the deficiencies in the prior art, the purpose of this invention is to provide an active suspension control method and system based on rolling driving style recognition.

[0007] An active suspension control method based on rolling driving style recognition, provided by the present invention, includes:

[0008] Step S1: Obtain brake, accelerator pedal opening, steering wheel angle and vehicle speed information from the CAN bus, and save historical data to build a vehicle operating condition information database;

[0009] Step S2: Compare the current brake pedal opening, accelerator pedal opening, and steering wheel angle with the preset corresponding thresholds to distinguish the current longitudinal driving condition and lateral driving condition of the vehicle.

[0010] Step S3: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and segment the historical vehicle speed trajectories based on a sliding time window to obtain a vehicle speed time series;

[0011] Step S4: Calculate the average and variance of the vehicle speed time series to obtain statistical features; perform Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) on the vehicle speed time series to obtain temporal features; based on the statistical and temporal features, compare with the driving style feature library and apply a classification and recognition model to identify driving style.

[0012] Step S5: Using driving conditions and driving style as inputs, apply decision rules to achieve active suspension control.

[0013] Preferably, the longitudinal driving conditions of the vehicle include acceleration, cruising, and deceleration;

[0014] The lateral driving conditions include turning left, turning right, and going straight.

[0015] Preferably, step S2 includes:

[0016] Step S2.1: Set a threshold value based on experience, and set the accelerator pedal opening threshold value to P. thr，g Set the brake pedal opening threshold to P. thr，b Set the steering wheel angle threshold to A. thr，sw ;

[0017] Step S2.2: Extract the latest accelerator pedal opening P from the vehicle operating condition information database. g The latest brake pedal opening P b And the latest steering wheel angle A sw ;

[0018] Step S2.3: Increase the accelerator pedal opening P g With threshold P thr，g Compare the sizes, brake pedal opening P b With threshold P thr，b Compare sizes, P g Greater than P thr，g And P b Less than p thr，b For acceleration driving conditions, P g Less than P thr，g And Pb Greater than P thr，b For deceleration driving conditions, P g Less than P thr，g And P b Less than P thr，b Corresponding to cruise driving conditions, P g Greater than P thr，g And P b Greater than P thr，b The situation is a dangerous operation, which is defined as a deceleration driving condition, and the safety system is brought in to take over.

[0019] Step S2.4: Turn the steering wheel by angle A sw With threshold A thr，sw Compare the sizes, A sw Less than -A thr，sw For the left-turn driving scenario, A sw Greater than A thr，sw For right-turn driving conditions, A sw In the range (-A) thr，sw A thr，sw The corresponding straight-line driving condition is indicated within this section.

[0020] Preferably, step S3 includes:

[0021] Step S3.1: Set the sliding time window size to L;

[0022] Step S3.2: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and reversely segment the vehicle speed trajectories. That is, retain the latest L vehicle speed data to form a vehicle speed time series, and update the vehicle speed time series as the vehicle speed information is updated, so that the latest frame of vehicle speed is added to the vehicle speed time series and the oldest frame of vehicle speed is removed. Always keep the length of the vehicle speed time series the same as the length of the sliding time window L to achieve rolling segmentation.

[0023] Preferably, the joint operation of Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) in step S4 includes:

[0024] Step S4.1: Perform EMD decomposition on the vehicle speed time series to obtain the intrinsic mode function (IMF) and residual (Res), as shown in the formula:

[0025]

[0026] In the formula, X(t) represents a signal composed of a vehicle speed time series, and C i (t) is the i-th IMF, and n represents the total number of IMFs obtained after decomposing the vehicle speed time series;

[0027] Step S4.2: Perform a DFT transform on each IMF to obtain the IMF signal in the frequency domain, as shown in the formula:

[0028] C i，f (w)=DFT(C i (t))

[0029] In the formula C i，f (w) is the frequency domain signal of the i-th IMF after discrete Fourier transform;

[0030] Step S4.3: Determine the frequency domain signal C i，f The position of the maximum amplitude of (w) w i，m At this time, C i，f (w i，m That is, the dominant frequency, calculate C. i，f (w i，m The value of ) is calculated, and the main frequencies of the n IMFs are compared. The maximum value is extracted as the time series feature, and the formula is:

[0031] w i,m =argmax(abs(C i，f (w)))

[0032] f i =Freq(C i，f (w i,m ))

[0033]

[0034] In the formula, the function abs(·) calculates the magnitude of the signal, and the function argmax(·) calculates the position w of the maximum magnitude. i，m The function Freq(·) calculates the frequency of the signal, f. i It is the dominant frequency of the current signal, f i，m It is the maximum dominant frequency of this vehicle speed time series.

[0035] Preferably, the driving style feature library includes features defined by human labels and features obtained by unsupervised learning clustering algorithms based on pre-collected data from different drivers.

[0036] Preferably, based on the vehicle speed trajectory collected by the driver, the average and variance operations are performed on the collected vehicle speed trajectories to obtain the statistical average speed V. a and speed standard deviation V s ;

[0037] The maximum dominant frequency f is obtained using the EMD-IMF joint analysis method. i，m ;

[0038] The statistical average velocity V a Speed ​​standard deviation V s and the maximum dominant frequency f i，mAs input, K-means++ is used for clustering, and the original vehicle speed trajectories are labeled to construct a driving style feature library.

[0039] Preferably, the classification and recognition model includes Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN), and Recurrent Neural Network (RNN).

[0040] Preferably, the decision rule described in step S5 is that there are m driving conditions and n driving styles, which can be divided into m×n cases. Based on the actual judgment of each case, an active suspension gear is set, which can achieve suspension control.

[0041] An active suspension control system based on rolling driving style recognition, provided by the present invention, includes:

[0042] Module M1: Obtains brake, accelerator pedal opening, steering wheel angle and vehicle speed information from the CAN bus, and saves historical data to build a vehicle operating condition information database;

[0043] Module M2: Compares the current brake pedal opening, accelerator pedal opening, and steering wheel angle with the corresponding preset thresholds to distinguish the current longitudinal driving conditions and lateral driving conditions of the vehicle.

[0044] Module M3: Extracts historical vehicle speed trajectories from the vehicle operating condition information database, segments the historical vehicle speed trajectories based on a sliding time window, and obtains the vehicle speed time series;

[0045] Module M4: Performs averaging and variance calculations on the vehicle speed time series to obtain statistical features; performs Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) on the vehicle speed time series to obtain temporal features; based on the statistical and temporal features, and in accordance with the driving style feature library, applies a classification and recognition model to identify driving style.

[0046] Module M5: Taking driving conditions and driving style as inputs, it applies decision rules to achieve active suspension control.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. This invention proposes a novel temporal feature extraction method. This method uses Empirical Mode Decomposition (EMD) to obtain different Intrinsic Mode Functions (IMFs), removing low-frequency IMFs to effectively mitigate clutter interference. Furthermore, by extracting the maximum dominant frequency from the IMFs using Discrete Fourier Transform (DFT) and extracting temporal features from the frequency domain, the accuracy of subsequent driving style prediction can be significantly improved.

[0049] 2. This invention proposes a driving style rolling recognition method based on a sliding time window. This method continuously updates the vehicle speed time series through a sliding time window, thereby continuously updating the driving style and effectively improving the real-time performance of suspension control.

[0050] 3. This invention proposes an active suspension control method that combines human-vehicle factors. This method incorporates the driver's personal factors and guides the active suspension control mode based on the current vehicle operating conditions, which can significantly enhance the comfort and handling stability of the vehicle suspension system. Attached Figure Description

[0051] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0052] Figure 1 This is a schematic diagram illustrating the implementation process of the method described in the embodiments of the present invention;

[0053] Figure 2 This is a schematic diagram of the rolling driving style recognition process according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the decision rules described in an embodiment of the present invention. Detailed Implementation

[0055] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0056] This invention acquires brake, accelerator pedal opening, steering wheel angle, and vehicle speed information from the CAN bus to construct a driving condition information database; it extracts pedal opening and steering wheel angle information and compares them with preset opening thresholds to identify the current driving condition; it extracts historical vehicle speed information and performs rolling segmentation based on time windows to obtain a vehicle speed time series; it applies averaging and standard deviation to the vehicle speed time series to extract statistical features, and applies Empirical Mode Decomposition (EMD)-Discrete Fourier Transform (DFT) to jointly extract temporal features, identifying driving styles based on these two features; and it uses driving conditions and driving styles as inputs to achieve active suspension control. By jointly extracting driver temporal features through EMD-DFT, accurate identification of driving styles is achieved. Simultaneously, the introduction of rolling recognition based on time windows improves the real-time performance of suspension control. Finally, by combining human-vehicle factors, it enhances the comfort and handling stability of suspension control.

[0057] Example 1

[0058] According to the present invention, an active suspension control method based on rolling driving style recognition is provided, such as... Figure 1 As shown, it includes:

[0059] Step S1: Obtain brake and accelerator pedal opening, steering wheel angle, and vehicle speed information from the CAN bus, and save historical data to build a vehicle operating condition information database. Only the latest data for accelerator and brake pedal opening and steering wheel angle information needs to be saved; historical data is not required.

[0060] Step S2: Compare the current brake and accelerator pedal openings and steering wheel angles with preset threshold values ​​to distinguish between the current longitudinal and lateral driving conditions of the vehicle. In other words, set the brake and accelerator pedal opening thresholds to P. thr，g P thr，b The current accelerator pedal opening P is respectively set. g Brake pedal opening P b Compared with the brake and accelerator pedal opening thresholds, the current longitudinal driving condition of the vehicle is distinguished, and the steering wheel angle threshold is set as A. thr，sw The steering wheel angle threshold and the steering wheel angle A are then compared. sw In comparison, the lateral driving conditions of the current vehicle are distinguished. The longitudinal driving conditions of the vehicle include acceleration, cruising, and deceleration. The lateral driving conditions include left turn, right turn, and straight driving. Step S2 includes:

[0061] Step S2.1: Set a threshold value based on experience, and set the accelerator pedal opening threshold value to P. thr，g Set the brake pedal opening threshold to P. thr，b Set the steering wheel angle threshold to A. thr，sw .

[0062] Step S2.2: Extract the latest accelerator pedal opening P from the vehicle operating condition information database. g The latest brake pedal opening P b And the latest steering wheel angle A sw

[0063] Step S2.3: Increase the accelerator pedal opening P g With threshold P thr，g Compare the sizes, brake pedal opening P b With threshold Pthr，b Compare sizes, (P) g >P thr，g )&(P b <P thr，b (P) corresponds to the acceleration driving condition. g <P thr，g )&(P b >Pthr，b (P) corresponds to the deceleration driving condition. g <P thr，g )&(P b >P thr，b This corresponds to the cruise driving condition. (P) not mentioned in step S23 g >P thr，g )&(P b >P thr，b The situation is a dangerous operation, which is defined as a deceleration driving condition, and the safety system is brought in to take over.

[0064] Step S2.4: Turn the steering wheel by angle A sw With threshold A thr，sw Compare sizes, (A) sw <-A thr，sw (A) corresponds to the left-turn driving condition. sw >A thr，sw (-A) corresponds to the right turn driving condition. thr，sw <A sw <A thr，sw This corresponds to the straight-ahead driving condition.

[0065] Step S3: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and segment the historical vehicle speed trajectories based on a sliding time window to obtain a vehicle speed time series. Step S3 includes:

[0066] Step S3.1: Set the sliding time window size to L.

[0067] Step S3.2: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and reversely segment the vehicle speed trajectories. That is, retain the latest L vehicle speed data to form a vehicle speed time series, and update the vehicle speed time series as the vehicle speed information is updated, so that the latest frame of vehicle speed is added to the vehicle speed time series and the oldest frame of vehicle speed is removed. Always keep the length of the vehicle speed time series the same as the length of the sliding time window L to achieve rolling segmentation.

[0068] Step S4: Average and variance calculations are performed on the vehicle speed time series to obtain statistical features; Empirical Mode Decomposition (EMD)-Discrete Fourier Transform (DFT) is performed on the vehicle speed time series to obtain temporal features; based on the statistical and temporal features, and referring to the driving style feature database, a classification and recognition model is applied to identify the driving style. The specific implementation process of the Empirical Mode Decomposition (EMD)-Discrete Fourier Transform (DFT) operation in Step S4 is as follows:

[0069] Step S4.1: Perform EMD decomposition on the vehicle speed time series obtained in step S3.2 to obtain the intrinsic mode function (IMF) and residual (Res). The specific formula is as follows:

[0070]

[0071] In the formula, X(t) represents a signal composed of a vehicle speed time series, and C i (t) The i-th IMF, where n represents the total number of IMFs obtained after decomposing the vehicle speed time series.

[0072] Step S4.2: Perform a DFT transform on each IMF to obtain the IMF signal in the frequency domain. The specific formula is as follows:

[0073] C i，f (w)=DFT(C i (t))

[0074] In the formula C i，f (w) is the frequency domain signal of the i-th IMF after discrete Fourier transform.

[0075] Step S4.3: Determine the frequency domain signal C i，f The position of the maximum amplitude of (w) w i，m At this time, C i，f (w i，m That is, the dominant frequency, calculate C. i，f (w i，m The value of ) is calculated, and the main frequencies of the n IMFs are compared. The maximum value is extracted as the time series feature. The specific formula is as follows:

[0076] w i,m =argmax(abs(C i,f (w)))

[0077] f i =Freq(C i，f (w i，m ))

[0078]

[0079] In the formula, the function abs(·) calculates the magnitude of the signal, and the function argmax(·) calculates the position w of the maximum magnitude. i，m The function Freq(·) calculates the frequency of the signal, f. i It is the dominant frequency of the current signal, f. i，m It is the maximum dominant frequency of the vehicle speed time series.

[0080] like Figure 2 As shown, the driving style feature library and driving style rolling recognition of the present invention are specifically described as follows:

[0081] The driving style feature library mentioned in step S4 can be defined by human labels or obtained by pre-collecting data from different drivers and performing unsupervised learning clustering algorithms, such as the K-means++ algorithm. More specifically, the collected data from different drivers mainly includes vehicle speed trajectories during driving. The collected speed trajectories are averaged and varianced to obtain the statistical average speed V. a and speed standard deviation V s And using the above-mentioned EMD-IMF joint analysis method, the maximum dominant frequency f is obtained. i，m Using these three quantities as input features, K-means++ is applied for clustering. As a preferred embodiment of the present invention, the number of clusters is 3, and the original vehicle speed trajectory segments are labeled to construct a driving style feature library.

[0082] The classification and recognition model described in step S4 can employ machine learning algorithms such as Support Vector Machine (SVM) and Random Forest (RF), or deep learning algorithms such as Artificial Neural Network (ANN) and Recurrent Neural Network (RNN). More specifically, the driving style feature condition database is used as input to pre-train the adopted SVM until the loss function stabilizes. The trained SVM is then deployed in the vehicle MCU to realize the driving style recognition function.

[0083] Step S5: Using driving conditions and driving style as input, apply decision rules to achieve active suspension control. The decision rules in Step S5 specifically mean that if there are m driving conditions and n driving styles, then m×n cases can be identified. Based on the actual judgment of each case, an active suspension gear is determined, thus achieving suspension control. For example, ... Figure 3 The decision-making rules shown are divided into 27 situations based on 9 driving conditions and 3 driving styles.

[0084] This invention aims to accurately identify driving style by jointly extracting driver temporal features through EMD-DFT, and introduces rolling recognition based on time windows to continuously update driving style, thereby enhancing the real-time performance of active suspension control.

[0085] This invention implements active suspension control based on rolling driving style recognition. It effectively extracts clutter interference during time-series feature extraction, significantly improving the accuracy of driving style prediction; by rolling updates of the vehicle speed time series, it effectively improves the real-time performance of suspension control; and by combining human-vehicle factors to guide the active suspension control mode, it significantly enhances the performance and reliability of the vehicle suspension system. Given the crucial role of active suspension control in vehicle comfort and handling stability, this invention has significant practical value.

[0086] Example 2

[0087] The present invention also provides an active suspension control system based on rolling driving style recognition. The active suspension control system based on rolling driving style recognition can be implemented by executing the process steps of the active suspension control method based on rolling driving style recognition. That is, those skilled in the art can understand the active suspension control method based on rolling driving style recognition as a preferred embodiment of the active suspension control system based on rolling driving style recognition.

[0088] An active suspension control system based on rolling driving style recognition, provided by the present invention, includes:

[0089] Module M1: Obtains brake, accelerator pedal opening, steering wheel angle and vehicle speed information from the CAN bus, and saves historical data to build a vehicle operating condition information database.

[0090] Module M2: Compares the current brake, accelerator pedal opening, and steering wheel angle with preset thresholds to distinguish between the vehicle's longitudinal and lateral driving conditions. Longitudinal driving conditions include acceleration, cruising, and deceleration. Lateral driving conditions include left turns, right turns, and straight driving. Module M2 includes: Module M2.1: Sets thresholds based on experience, setting the accelerator pedal opening threshold to P. thr，g Set the brake pedal opening threshold to P. thr，b Set the steering wheel angle threshold to A. thr，sw Module M2.2: Extracts the latest accelerator pedal opening P from the vehicle operating condition information database. g The latest brake pedal opening P b And the latest steering wheel angle A sw Module M2.3: Adjusts the accelerator pedal opening P. g With threshold P thr，g Compare the sizes, brake pedal opening P b With threshold P thr，b Compare sizes, P g Greater than P thr，g And P b Less than P thr，b For acceleration driving conditions, P g Less than P thr，g And P b Greater than P thr，b For deceleration driving conditions, P g Less than P thr，g And P b Less than A thr，b Corresponding to cruise driving conditions, P g Greater than A thr，g And P b Greater than P thr，bThe situation is a dangerous operation; it is defined as a deceleration driving condition, and the safety system takes over. Module M2.4: Adjusts the steering wheel angle A. sw With threshold A thr，sw Compare the sizes, A sw Less than -A thr，sw For the left-turn driving scenario, A sw Greater than A thr，sw For right-turn driving conditions, A sw In the range (-A) thr，sw A thr，sw The corresponding straight-line driving condition is indicated within this section.

[0091] Module M3: Extracts historical vehicle speed trajectories from the vehicle condition information database and segments these trajectories using a sliding time window to obtain a vehicle speed time series. Module M3 includes: Module M3.1: Sets the sliding time window size to L. Module M3.2: Extracts historical vehicle speed trajectories from the vehicle condition information database and segments the trajectories in reverse, retaining the latest L vehicle speed data to form a vehicle speed time series. As vehicle speed information is updated, the time series is supplemented with the latest frame of vehicle speed and removed from the oldest frame, always maintaining the length of the vehicle speed time series the same as the sliding time window length L, thus achieving rolling segmentation.

[0092] Module M4: Performs averaging and variance calculations on the vehicle speed time series to obtain statistical features. It then performs a joint operation of Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) on the vehicle speed time series to obtain temporal features. Based on the statistical and temporal features, and referring to a driving style feature library, a classification and recognition model is applied to identify driving styles. The joint operation of Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) in Module M4 includes: Module M4.1: Performs EMD decomposition on the vehicle speed time series to obtain the Intrinsic Mode Function (IMF) and residuals (Res), with the following formula:

[0093]

[0094] In the formula, X(t) represents a signal composed of a vehicle speed time series, and C i (t) The i-th IMF, where n represents the total number of IMFs obtained after decomposing the vehicle speed time series. Module M4.2: Perform DFT transformation on each IMF to obtain the IMF signal in the frequency domain, using the formula:

[0095] C i，f (w)=DFT(C i (t))

[0096] In the formula C i，f (w) is the frequency domain signal of the i-th IMF after discrete Fourier transform. Module M4.3: Calculate the frequency domain signal C. i，f The position of the maximum amplitude of (w) wi，m At this time, C i，f (w i，m That is, the dominant frequency, calculate C. i，f (w i，m The value of ) is calculated, and the main frequencies of the n IMFs are compared. The maximum value is extracted as the time series feature, and the formula is:

[0097] w i,m =argmax(abs(C i,f (w)))

[0098] f i =Freq(C i,f (w i,m ))

[0099]

[0100] In the formula, the function abs(·) calculates the magnitude of the signal, and the function argmax(·) calculates the position w of the maximum magnitude. i，m The function Freq(·) calculates the frequency of the signal, f. i It is the dominant frequency of the current signal, f i，m It is the maximum dominant frequency of this vehicle speed time series.

[0101] The driving style feature library includes features defined by human labels and features obtained through unsupervised learning clustering algorithms based on pre-collected data from different drivers. Based on the driver's speed trajectory, the collected speed trajectories are averaged and their variances are calculated to obtain the statistical measure, average speed V. a and speed standard deviation V s Based on the EMD-IMF joint analysis method, the maximum dominant frequency f is obtained. i，m The statistical average velocity V will be obtained. a Speed ​​standard deviation V s and maximum main frequency f i，m As input, K-means++ is used for clustering, and the original vehicle speed trajectories are labeled to construct a driving style feature library. The classification and recognition models include Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN), and Recurrent Neural Network (RNN).

[0102] Module M5: Taking driving conditions and driving styles as input, it applies decision rules to achieve active suspension control. The decision rules for Module M5 are as follows: given m driving conditions and n driving styles, there are n×n possible scenarios. Based on the actual judgment of each scenario, an active suspension gear is determined, thus achieving suspension control.

[0103] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0104] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An active suspension control method based on rolling driving style recognition, characterized in that, include: Step S1: Obtain brake, accelerator pedal opening, steering wheel angle and vehicle speed information from the CAN bus, and save historical data to build a vehicle operating condition information database; Step S2: Compare the current brake pedal opening, accelerator pedal opening, and steering wheel angle with the preset corresponding thresholds to distinguish the current longitudinal driving condition and lateral driving condition of the vehicle. Step S3: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and segment the historical vehicle speed trajectories based on a sliding time window to obtain a vehicle speed time series; Step S4: Calculate the average and variance of the vehicle speed time series to obtain statistical characteristics; Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) are jointly performed on the vehicle speed time series to obtain time series features; based on the statistical and time series features, and in accordance with the driving style feature database, a classification and recognition model is applied to identify driving style. Step S5: Using driving conditions and driving style as inputs, apply decision rules to achieve active suspension control; The driving style feature library includes features defined by human labels and features obtained by unsupervised learning clustering algorithms based on pre-collected data from different drivers. Based on the vehicle speed trajectory collected by the driver, the average and variance of the collected speed trajectories are calculated to obtain the statistical quantity, average speed. and speed standard deviation ; The maximum dominant frequency was obtained using the EMD-IMF joint analysis method. ; Average speed of the statistics Speed ​​standard deviation and the maximum main frequency As input, K-means++ is used for clustering, and the original vehicle speed trajectories are labeled to construct a driving style feature library.

2. The active suspension control method based on rolling driving style recognition according to claim 1, characterized in that, The vehicle's longitudinal driving conditions include acceleration, cruising, and deceleration; The lateral driving conditions include turning left, turning right, and going straight.

3. The active suspension control method based on rolling driving style recognition according to claim 2, characterized in that, Step S2 includes: Step S2.1: Set a threshold based on experience; set the accelerator pedal opening threshold as follows: Set the brake pedal opening threshold to Set the steering wheel angle threshold to ; Step S2.2: Extract the latest accelerator pedal opening from the vehicle operating condition information database. The latest brake pedal opening and the latest steering wheel angle ; Step S2.3: Increase the accelerator pedal opening. With threshold Compare the size of the brake pedal opening. With threshold Compare sizes, Greater than and Less than Corresponding to acceleration driving conditions, Less than and Greater than Corresponding to deceleration driving conditions, Less than and Less than Corresponding to cruise driving conditions, Greater than and Greater than The situation is a dangerous operation, which is defined as a deceleration driving condition, and the safety system is brought in to take over. Step S2.4: Turn the steering wheel angle With threshold Compare sizes, Less than - For left-turn driving conditions, Greater than Corresponding to right turn driving conditions, In the range (- , The corresponding straight-line driving condition is indicated within this section.

4. The active suspension control method based on rolling driving style recognition according to claim 1, characterized in that, Step S3 includes: Step S3.1: Set the sliding time window size to... ; Step S3.2: Extract historical vehicle speed trajectories from the vehicle operating condition information database, and reverse-engineer the vehicle speed trajectories to segment them, i.e., retain the latest ones. The vehicle speed data is used to form a vehicle speed time series, which is updated as vehicle speed information changes. The latest frame of vehicle speed is added to the vehicle speed time series, while the oldest frame of vehicle speed is removed. The length of the vehicle speed time series is always kept consistent with the length of the sliding time window. The same principle applies to achieving rolling segmentation.

5. The active suspension control method based on rolling driving style recognition according to claim 1, characterized in that, The joint operation of Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) in step S4 includes: Step S4.1: Perform EMD decomposition on the vehicle speed time series to obtain the intrinsic mode function (IMF) and residual (Res), as shown in the formula: In the formula This represents a signal composed of a time sequence of vehicle speeds. No. One IMF, This indicates that the time series of vehicle speeds was decomposed into a total of [number] values. One IMF; Step S4.2: Perform a DFT transform on each IMF to obtain the IMF signal in the frequency domain, as shown in the formula: In the formula It is the first The frequency domain signal of an IMF after discrete Fourier transform; Step S4.3: Determine the frequency domain signal Position of maximum amplitude ,at this time That is, the main frequency, calculate The value, and for The main frequencies of each IMF are compared, and the maximum value is extracted as the time series feature. The formula is as follows: In the formula, the function To find the magnitude of a signal, the function is... Find the position of the maximum modulus. ,function Find the frequency of the signal. It is the dominant frequency of the current signal. It is the maximum dominant frequency of this vehicle speed time series.

6. The active suspension control method based on rolling driving style recognition according to claim 1, characterized in that, The classification and recognition models include Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN), and Recurrent Neural Network (RNN).

7. The active suspension control method based on rolling driving style recognition according to claim 1, characterized in that, The decision rule described in step S5 is that there exists Various driving conditions and Different driving styles are then categorized. In each of these situations, by determining the appropriate active suspension settings based on the actual conditions, suspension control can be achieved.

8. An active suspension control system based on rolling driving style recognition, characterized in that, include: Module M1: Obtains brake, accelerator pedal opening, steering wheel angle and vehicle speed information from the CAN bus, and saves historical data to build a vehicle operating condition information database; Module M2: Compares the current brake pedal opening, accelerator pedal opening, and steering wheel angle with the corresponding preset thresholds to distinguish the current longitudinal driving conditions and lateral driving conditions of the vehicle. Module M3: Extracts historical vehicle speed trajectories from the vehicle operating condition information database, segments the historical vehicle speed trajectories based on a sliding time window, and obtains the vehicle speed time series; Module M4: Performs averaging and variance calculations on the vehicle speed time series to obtain statistical characteristics; Empirical Mode Decomposition (EMD) and Discrete Fourier Transform (DFT) are jointly performed on the vehicle speed time series to obtain time series features; based on the statistical and time series features, and in accordance with the driving style feature database, a classification and recognition model is applied to identify driving style. Module M5: Takes driving conditions and driving style as inputs and applies decision rules to achieve active suspension control; The driving style feature library includes features defined by human labels and features obtained by unsupervised learning clustering algorithms based on pre-collected data from different drivers. Based on the vehicle speed trajectory collected by the driver, the average and variance of the collected speed trajectories are calculated to obtain the statistical quantity, average speed. and speed standard deviation ; The maximum dominant frequency was obtained using the EMD-IMF joint analysis method. ; Average speed of the statistics Speed ​​standard deviation and the maximum main frequency As input, K-means++ is used for clustering, and the original vehicle speed trajectories are labeled to construct a driving style feature library.

Images