Automobile stabilizer bar control system and method based on machine learning

Abstract

The invention relates to the technical field of vehicle control, in particular to an automobile stabilizer bar control system and method based on machine learning. The automobile stabilizer bar control method comprises the steps that driving state data and suspension stroke change data of a target automobile are collected in real time; judging whether the target vehicle is in a steering working condition or not according to the driving state data and the suspension travel change data; and if the target vehicle is in the steering working condition, the driving state data and the suspension stroke change data are input into a pre-established machine learning prediction model to output the optimal torsional rigidity value of the stabilizer bar, and the rigidity of the stabilizer bar is adjusted to the optimal torsional rigidity value. According to the method, the nonlinear coupling influence relation of multiple factors such as the vehicle speed, the load, the suspension state and the road surface condition on body roll is automatically captured in real time through the machine learning model, so that the optimal stabilizer bar rigidity parameter is output, the stabilizer bar control precision is improved, and the safety risk of whole vehicle driving is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle control, and particularly relates to an anti-roll bar control system and control method for an automobile based on machine learning. Background Art

[0002] For a traditional passive anti-roll bar of an automobile, the left and right suspensions of the automobile are rigidly connected, so that the bumps and vibrations on one side of the vehicle body are transmitted to the other side, reducing the comfort of the occupants; in addition, since the stiffness of the passive anti-roll bar is fixed, when the vehicle is in an extreme state, the vehicle will have excessive roll or even rollover accidents.

[0003] In the related art, generally an active anti-roll bar is used for adjustment. The active anti-roll bar can adjust the stiffness of the anti-roll bar through a hydraulic or electric actuator. When the vehicle is turning, the system will increase the stiffness of the anti-roll bar to reduce the roll of the vehicle body and improve the stability and controllability of the vehicle; while when driving straight or turning at a low speed, the system will reduce the stiffness to increase the flexibility of the suspension and improve the riding comfort. However, in the related art, the accuracy and response speed of the active anti-roll bar control method are still not ideal enough, and its control parameters are usually fixed, making it difficult to adapt to the dynamic changes of the system. After the vehicle has been driving for a long time, due to wear and other reasons, some inertias will change, resulting in a decrease in precision control and an increase in safety risks. Summary of the Invention

[0004] Aiming at the problem that the control parameters of the active anti-roll bar control method in the related art for the stiffness are fixed and it is difficult to adapt to the dynamic changes of the system.

[0005] In a first aspect, an embodiment of the present application provides an anti-roll bar control method for an automobile based on machine learning, and the anti-roll bar control method for an automobile includes:

[0006] Real-time collect the driving state data and suspension stroke change data of the target vehicle;

[0007] Judge whether the target vehicle is in a turning condition according to the driving state data and the suspension stroke change data;

[0008] If the target vehicle is in a turning condition, input the driving state data and the suspension stroke change data into a pre-built machine learning prediction model to output the optimal torsional stiffness value of the anti-roll bar, and adjust the stiffness of the anti-roll bar to the optimal torsional stiffness value.

[0009] In combination with the first aspect, in an implementation manner, before inputting the driving state data and the suspension stroke change data into the pre-built machine learning prediction model, it further includes:

[0010] Collect the complete driving data of the sample vehicle;

[0011] Train the SVR model using the complete driving data to generate a machine learning prediction model.

[0012] Combined with the first aspect, in one embodiment, the training of the SVR model using the complete driving data includes:

[0013] Divide the complete driving data into a training data group and a test data group;

[0014] Use the training data group to train the SVR model, and then test the prediction results of the SVR model through the test data group until the prediction results meet the preset requirements.

[0015] Combined with the first aspect, in one embodiment, after dividing the complete driving data into a training data group and a test data group, it further includes: performing normalization processing on the training data group and the test data group.

[0016] Combined with the first aspect, in one embodiment, the collection of the complete driving data of the sample vehicle includes: collecting the complete driving data of multiple sample vehicles within a preset time under different road conditions.

[0017] Combined with the first aspect, in one embodiment, the real-time collection of the driving state data and the suspension stroke change data of the target vehicle includes:

[0018] Real-time monitor the suspension stroke change data through a strain sensor, and obtain the body lateral acceleration, yaw rate, steering wheel angle and rotational speed through the controller area network of the target vehicle.

[0019] Combined with the first aspect, in one embodiment, after the real-time collection of the driving state data and the suspension stroke change data of the target vehicle, it further includes:

[0020] Preprocess the driving state data and the suspension stroke change data to filter out high-frequency noise.

[0021] Combined with the first aspect, in one embodiment, when determining whether the target vehicle is in a steering condition based on the driving state data and the suspension stroke change data, it further includes:

[0022] When the target vehicle is in a flat road driving condition, drive the stabilizer bar to reduce the stiffness to a preset threshold;

[0023] When the target vehicle is in a bumpy road driving condition, disconnect the connection of the stabilizer bar.

[0024] In a second aspect, an embodiment of the present application provides an automotive stabilizer bar control system, which includes:

[0025] A data acquisition module, which is used to collect the driving state data and the suspension stroke change data of the target vehicle in real time;

[0026] A road condition classification module, which is used to determine whether the target vehicle is in a steering condition according to the driving state data and the suspension stroke change data;

[0027] A stiffness prediction module, which is used to input the driving state data and the suspension stroke change data into a pre-built machine learning prediction model when the target vehicle is in a steering condition, so as to output the optimal torsional stiffness value of the stabilizer bar.

[0028] Combined with the second aspect, in an implementation manner, the data acquisition module includes:

[0029] A strain sensor, which is used to be installed at the connection between the suspension spring and the shock absorber;

[0030] A controller area network, which is used to read the lateral acceleration, yaw rate, steering wheel angle and rotational speed of the body of the target vehicle.

[0031] The beneficial effects brought by the technical solution provided by the embodiment of the present application include:

[0032] Through the machine learning model, the present application can automatically capture the non-linear coupling influence relationship of vehicle body roll affected by multiple factors such as vehicle speed, load, suspension state, and road conditions in real time, so as to output the optimal stabilizer bar stiffness parameters, improve the control accuracy of the stabilizer bar, and reduce the safety risk of the whole vehicle during driving. Description of the Drawings

[0033] Figure 1 It is a schematic flowchart of the method for controlling an automotive stabilizer bar in an embodiment of the present application;

[0034] Figure 2 It is a flowchart of machine learning prediction in an embodiment of the present application;

[0035] Figure 3 It is a schematic hardware structure diagram of the automotive stabilizer bar control device involved in the solution of the embodiment of the present application. Detailed Embodiments

[0036] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0037] In related technologies, the accuracy and response speed of the active stabilizer bar control method are still not ideal enough. Its control parameters are usually fixed and it is difficult to adapt to the dynamic changes of the system. After a vehicle has been driving for a long time, due to wear and other reasons, some inertias will change, resulting in a decrease in precision control and an increase in safety risks.

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail with reference to the accompanying drawings.

[0039] In a first aspect, as Figure 1 shown, an embodiment of this application provides a vehicle stabilizer bar control method based on machine learning. The vehicle stabilizer bar control method includes:

[0040] Step S1: Collect the driving state data and suspension stroke change data of the target vehicle in real time.

[0041] It should be noted that when the vehicle is turning, the system will increase the stiffness to reduce the body roll and improve the stability and handling of the vehicle; while when driving straight or turning at low speed, the system will reduce the stiffness to increase the flexibility of the suspension and improve the riding comfort. Therefore, it is necessary to collect the driving state data and suspension stroke change data of the target vehicle in real time for subsequent steps to determine the current driving state of the target vehicle.

[0042] Step S1 specifically includes:

[0043] Step S1a: Collect the initial data of the driving state data and suspension stroke change data of the target vehicle.

[0044] Furthermore, the collection of suspension stroke change data can use a strain sensor installed at the connection between the suspension spring and the shock absorber, with a sampling frequency of 1 kHz for real-time monitoring of suspension stroke changes, and the accuracy of the strain sensor is within ±0.1% FS. The collection of driving state data can read the body lateral acceleration, yaw rate, steering wheel angle (accuracy ±0.5°), and rotational speed (sampling frequency 100 Hz) through the CAN bus (i.e., Controller Area Network).

[0045] Step S1b: Preprocess the driving state data and suspension stroke change data to filter out high-frequency noise.

[0046] It is worth noting that the data collected in step S1a has interference such as noise, so it needs to be preprocessed first to improve the accuracy of subsequent judgment and prediction steps.

[0047] Specifically, a sliding window is used to filter out high-frequency noise, and the data of multiple sensors is fused through Kalman filtering. Among them, the size of the sliding window can be set to 0.5 s and the overlap rate is 50%.

[0048] Step S2. Determine whether the target vehicle is in a steering condition based on the driving state data and the suspension stroke change data.

[0049] It should be noted that the road condition classification needs to dynamically determine the vehicle state based on the steering wheel angle change rate and the suspension stroke change amount.

[0050] In a specific embodiment of the present application, the judgment process includes:

[0051] Judgment condition 1. If the angle change rate is less than 5° / s and the suspension stroke change is less than 5 mm, it is determined that the target vehicle is driving on a flat road surface.

[0052] Judgment condition 2. If the angle change rate is less than 5° / s and the suspension stroke change is greater than 10 mm, it is determined that the target vehicle is driving on a bumpy road surface.

[0053] Judgment condition 3. If the angle change rate is greater than 5° / s, it is determined that the target vehicle is in a steering condition.

[0054] Step S3. Adjust the stabilizer bar stiffness according to the condition of the target vehicle.

[0055] Condition 1. If the target vehicle is on a flat road surface, adjust the stabilizer bar stiffness to a preset threshold. The preset threshold can be selected as 50 N·m / °.

[0056] Condition 2. When the target vehicle is in a bumpy road driving condition, disconnect the stabilizer bar connection.

[0057] Condition 3. If the target vehicle is in a steering condition, input the driving state data and the suspension stroke change data into a pre-built machine learning prediction model to output the optimal torsional stiffness value of the stabilizer bar, and adjust the stabilizer bar stiffness to the optimal torsional stiffness value.

[0058] Furthermore, the input features of the machine learning prediction model include: body roll angle, lateral acceleration (generally in the range of ±2g), vehicle speed (generally in the range of 0 - 200 km / h), steering angle, and suspension stroke change amount. The output target of the machine learning prediction model is the optimal torsional stiffness of the stabilizer bar, and its general range is 50 - 250 N·m / °.

[0059] Preferably, the machine learning prediction model has an online update function. In some optional embodiments, the machine learning prediction model is incrementally updated every 24 hours, and historical data is stored in the edge computing unit (storage capacity ≥ 1TB).

[0060] It should be noted that the traditional method for determining the stabilizer bar stiffness relies on a mathematical model, while machine learning is data-driven and does not require an accurate physical model, which has more advantages in complex systems. For different driving modes (such as sport, comfort, economy), machine learning can predict the adaptable stiffness strategy. For example, the stiffness is increased in the sport mode to enhance handling, and the stiffness is reduced in the comfort mode to filter vibrations. Further, the traditional method relies on repeated tests of physical prototypes or high-precision simulation models, which is time-consuming and costly. Machine learning builds models through historical data or virtual simulation data, quickly generating prediction models and reducing the number of experimental iterations. Body roll is affected by the non-linear coupling of multiple factors such as vehicle speed, load, suspension state, and road conditions, and it is difficult for traditional control theory to accurately model. Machine learning can automatically capture these complex relationships. Finally, the vehicle faces uncertainties such as component aging, load changes, and tire wear during actual driving, and the machine learning model can dynamically adjust the prediction strategy through online learning.

[0061] Combined with the above-mentioned working condition three, before inputting the driving state data and the suspension stroke change data into the pre-built machine learning prediction model, it also includes the construction and training of the machine learning prediction model.

[0062] In a specific embodiment provided by the present application, the construction and training process of the machine learning prediction model includes:

[0063] Step A: Collect the complete driving data of the sample vehicle.

[0064] It should be noted that it is necessary to collect the complete driving data of multiple sample vehicles within a preset time under different road conditions as training samples and test samples.

[0065] In some specific embodiments, the driving data of 10 test vehicles for 1000 hours under various road conditions (urban roads, mountain roads, off-road) can be collected, and the sample size n is 107.

[0066] Step B: Use the complete driving data to train the SVR model to generate a machine learning prediction model.

[0067] It should be noted that the machine learning prediction model of the present application uses grid search to optimize the SVR parameters (C = 10, γ = 0.01, ε = 0.1), and cross-validation (5-fold) to ensure generalization.

[0068] The above step B specifically includes:

[0069] Step B1: Divide the complete driving data into a training data group and a test data group.

[0070] It should be noted that in step B1, the data can be divided into a training set and a test set. In subsequent steps, the predicted value of the torsion stiffness of the stabilizer bar is obtained through test calculations, and the relative error between the predicted value and the actual value in the test set is compared to verify the effect of model training.

[0071] Specifically, 80% of the complete driving data can be divided into the training set, and the remaining 20% into the test set.

[0072] Let the input feature matrix where n is the number of samples and d is the feature dimension; the output target vector

[0073] Divided according to the test set ratio of 0.2, X and are divided into the training set and the test set. Assume the number of training set samples is n train , and the number of test set samples is n test , and n test = 0.2n, n train = 0.8n.

[0074] Then the training set feature matrix The test set feature matrix The training set target vector The test set target vector The division process can be expressed as:

[0075]

[0076] where I train and I test are the sample index sets of the training set and the test set respectively, and I train ∪I test = {1, 2,..., n},

[0077] Step B2: Standardize the training data group and the test data group.

[0078] Specifically, for each feature dimension j = 1, 2,..., d in the training set feature matrix X train , calculate its mean μ j and standard deviation σ j ,

[0079]

[0080] The elements of the standardized training set feature matrix X train are calculated as follows:

[0081]

[0082] For the test set feature matrix Xtest, use the mean μ calculated from the training set j and the standard deviation σ j to perform standardization:

[0083]

[0084] Step B3: Use the training data set to train the SVR (Support Vector Regression) model, and then test the prediction results of the SVR model through the test data set until the prediction results meet the preset requirements.

[0085] It should be noted that traditional methods, such as PID (Proportional-Integral-Derivative Controller), are linear, have poor effects in dealing with non-linear problems, and are complex in parameter tuning. While machine learning, such as SVM or neural networks, can handle non-linear relationships and automatically learn feature interactions. In addition, traditional methods rely on mathematical models, while machine learning is data-driven and does not require an accurate physical model, which has more advantages in complex systems.

[0086] Specifically, for SVR using the radial basis function (RBF) kernel, its optimization problem can be expressed as:

[0087] Minimize the objective function:

[0088]

[0089] Constraints:

[0090]

[0091] where W is the weight vector, b is the bias term, ξ i and are slack variables, C is the penalty parameter, ε is the parameter of the insensitive loss function, is the function that maps the input x to a high-dimensional feature space.

[0092] The RBF kernel function is defined as:

[0093] K(x i , x j ) = exp(-γ∥x i - x j ∥ 2 )

[0094] where γ is the kernel coefficient,

[0095] By solving the above optimization problem, the optimal weight vector W and bias term b are obtained, thereby determining the regression function f(x):

[0096]

[0097] where α i is the Lagrange multiplier.

[0098] Further, after the training is completed, each sample x in the test set test ∈X test is used to obtain the predicted value

[0099]

[0100] After obtaining the predicted value, as shown Figure 2 in the figure, it is then determined whether the predicted value meets the requirements. If it does not meet the requirements, the above step B3 is repeated until the predicted value meets the requirements, and then the construction of the machine learning prediction model is completed.

[0101] Further, after the construction of the machine learning prediction model is completed, in working condition three of step S3, the real-time data can be input into the machine learning prediction model to adjust the stabilizer bar according to the optimal torsional stiffness value obtained by prediction.

[0102] In a second aspect, the present application provides an automobile stabilizer bar control system, which includes: a data acquisition module, a road condition classification module, and a stiffness prediction module; wherein,

[0103] The data acquisition module is used to collect the driving state data and suspension stroke change data of the target vehicle in real time; the road condition classification module is used to judge whether the target vehicle is in a turning condition according to the driving state data and suspension stroke change data; the stiffness prediction module is used to, when the target vehicle is in a turning condition, input the driving state data and suspension stroke change data into the pre-constructed machine learning prediction model to output the optimal torsional stiffness value of the stabilizer bar.

[0104] Specifically, the data acquisition module includes: a strain sensor and a controller area network; wherein,

[0105] The strain sensor is used to be installed at the connection between the suspension spring and the shock absorber; the controller area network is used to read the lateral acceleration, yaw rate, steering wheel angle and rotational speed of the target vehicle body.

[0106] In a third aspect, an embodiment of the present application provides an automobile stabilizer bar control device. The automobile stabilizer bar control device can be a device with data processing functions such as a personal computer (PC), a laptop computer, a server, etc.

[0107] Referring to Figure 3 Figure 3 ​This is a schematic diagram of the hardware structure of the vehicle stabilizer bar control device involved in the solution of the embodiment of the present application. In the embodiment of the present application, the vehicle stabilizer bar control device may include a processor, a memory, a communication interface, and a communication bus.

[0108] Among them, the communication bus can be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0109] The communication interface includes interfaces such as input / output (I / O) interfaces, physical interfaces, and logical interfaces for interconnecting components inside the vehicle stabilizer bar control device, as well as interfaces for interconnecting the vehicle stabilizer bar control device with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display screen (Display), a keyboard (Keyboard), etc.

[0110] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical memory, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0111] The processor can be a general-purpose processor, and the general-purpose processor can call the vehicle stabilizer bar control program stored in the memory and execute the vehicle stabilizer bar control method provided by the embodiment of the present application. For example, the general-purpose processor can be a central processing unit (CPU). Among them, the method executed when the vehicle stabilizer bar control program is called can refer to the various embodiments of the vehicle stabilizer bar control method of the present application, which will not be elaborated here.

[0112] Those skilled in the art can understand that Figure 3 the hardware structure shown in does not constitute a limitation to the present application, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0113] Fourthly, the embodiment of the present application further provides a readable storage medium.

[0114] An automobile stabilizer bar control program is stored on a readable storage medium of the present application. When the automobile stabilizer bar control program is executed by a processor, the steps of the above-mentioned automobile stabilizer bar control method are implemented.

[0115] Among them, the method implemented when the automobile stabilizer bar control program is executed can refer to each embodiment of the automobile stabilizer bar control method of the present application, and will not be elaborated here.

[0116] In summary, the present application uses a machine learning model to automatically capture in real time the non-linear coupling influence relationship of body roll affected by multiple factors such as vehicle speed, load, suspension state, and road conditions, so as to output the optimal stabilizer bar stiffness parameters, improve the control accuracy of the stabilizer bar, and reduce the safety risk of the whole vehicle during driving.

[0117] It should be noted that the serial numbers of the above embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0118] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, in essence, or the part that makes a contribution to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium as described above (such as ROM / RAM, magnetic disk, optical disc), and includes several instructions to enable a terminal device to execute the methods described in each embodiment of the present application.

[0119] The terms "including" and "having" and any variations thereof in the specification, claims and drawings of the present application are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices. The descriptions of terms such as "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit that "first", "second" and "third" are of different types.

[0120] In the description of the embodiments of the present application, terms such as "exemplary", "for example" or "for instance" are used to indicate examples, illustrations or explanations. Any embodiment or design solution described as "exemplary", "for example" or "for instance" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of terms such as "exemplary", "for example" or "for instance" is intended to present relevant concepts in a specific manner.

[0121] In the description of the embodiments of the present application, unless otherwise specified, " / " means "or". For example, A / B may mean A or B. The "and / or" in the text is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B may mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of the present application, "a plurality of" means two or more than two.

[0122] In some of the processes described in the embodiments of the present application, there are multiple operations or steps that appear in a specific order. However, it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or may be executed in parallel. The serial numbers of the operations are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0123] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structural or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.

Claims

1. A vehicle stabilizer bar control method based on machine learning, characterized in that: The automobile stabilizer bar control method comprises: Collect the target vehicle's driving status data and suspension travel change data in real time; Determine whether the target vehicle is in a steering condition based on the driving state data and the suspension travel change data; If the target vehicle is in a turning condition, the driving state data and suspension travel change data are input into a pre-built machine learning prediction model to output the optimal torsional stiffness value of the stabilizer bar, and the stabilizer bar stiffness is adjusted to the optimal torsional stiffness value.

2. The vehicle stabilizer bar control method based on machine learning as claimed in claim 1, characterized in that: The step of inputting the driving state data and the suspension travel change data into the pre-built machine learning prediction model also includes: Collect complete driving data of sample vehicles; The SVR model is trained using the complete driving data to generate a machine learning prediction model.

3. The vehicle stabilizer bar control method based on machine learning as claimed in claim 2, characterized in that: The method of training the SVR model using the complete driving data includes: Divide the complete driving data into a training data group and a test data group; The SVR model is trained using the training data set, and then the prediction results of the SVR model are tested using the test data set until the prediction results meet the preset requirements.

4. The vehicle stabilizer bar control method based on machine learning as claimed in claim 3, characterized in that: After the complete driving data is divided into a training data group and a test data group, the method further includes: performing standardization processing on the training data group and the test data group.

5. The vehicle stabilizer bar control method based on machine learning as claimed in claim 2, characterized in that: The collecting of complete driving data of sample vehicles includes: collecting complete driving data of multiple sample vehicles within a preset time under different road conditions.

6. The vehicle stabilizer bar control method based on machine learning as claimed in claim 1, characterized in that: The real-time collection of the target vehicle's driving state data and suspension travel change data includes: The suspension travel change data is monitored in real time through the strain sensor, and the body lateral acceleration, yaw angular velocity, steering wheel angle and speed are obtained through the target vehicle's controller area network.

7. The vehicle stabilizer bar control method based on machine learning as claimed in claim 1, characterized in that: After the real-time collection of the target vehicle's driving state data and suspension travel change data, the method further includes: The driving state data and suspension travel change data are preprocessed to filter out high-frequency noise.

8. The vehicle stabilizer bar control method based on machine learning as claimed in claim 1, characterized in that: When judging whether the target vehicle is in a turning condition according to the driving state data and the suspension travel change data, the method further includes: When the target vehicle is on a flat road, the stabilizer bar is driven to reduce its stiffness to a preset threshold; When the target vehicle is in a driving condition on a bumpy road, the stabilizer bar connection is disconnected.

9. A vehicle stabilizer bar control system based on machine learning, characterized in that: include: A data acquisition module, which is used to collect the driving status data and suspension travel change data of the target vehicle in real time; A road condition classification module, which is used to determine whether the target vehicle is in a turning condition based on the driving state data and the suspension travel change data; The stiffness prediction module is used to input the driving state data and suspension travel change data into a pre-built machine learning prediction model when the target vehicle is in a steering condition, so as to output the optimal torsional stiffness value of the stabilizer bar.

10. The vehicle stabilizer bar control system based on machine learning as claimed in claim 9, characterized in that: The data acquisition module comprises: A strain sensor is used to be installed at the connection between the suspension spring and the shock absorber; The controller area network is used to read the lateral acceleration, yaw rate, steering wheel angle and rotation speed of the target vehicle.

Images