Anti-lock braking systems for vehicles

Abstract

An anti-lock braking system for a vehicle obtains the vehicle's speed and acceleration at multiple predetermined time points within a predetermined time period; it uses deep learning-based artificial intelligence technology to mine dynamic change characteristic information of the vehicle's speed and acceleration in the time dimension, thereby accurately identifying road conditions based on the actual state of the vehicle, including road impact, bad and washboard roads, engine drag, low-adhesion roads, etc., to improve the applicability of the anti-lock braking system in various road conditions and enhance its practicality.

Description

Technical Field

[0001] The present application relates to the field of intelligent control technology, and more particularly, to an anti-lock braking system for a vehicle. Background Art

[0002] The automotive anti-lock braking system (ABS) is a new, high-efficiency braking system developed from conventional braking systems. Currently, when a vehicle temporarily drives over bumpy surfaces, such as speed bumps or potholes, the tires can squeeze or briefly lift off the road, creating a false slip rate and causing the anti-lock braking system to falsely trigger. This false triggering of the anti-lock braking system can produce noticeable operating noise and brake pedal vibration, impacting comfort.

[0003] To address this issue, Chinese patent application publication number CN114715095A discloses an anti-lock braking system and control method to address the issue of accidental activation of the anti-lock braking function when a vehicle temporarily drives over a bumpy surface, such as a speed bump or a pothole. This method uses an onboard GPS module to acquire vehicle speed signals and an onboard inertial sensor unit to collect acceleration and yaw rate in all directions to identify the vehicle's status. While anti-lock braking control systems can integrate multiple sensors to assist in calculations, these methods are costly and have limited engineering value.

[0004] Therefore, an optimized anti-lock braking system for a vehicle is desired. Summary of the Invention

[0005] To address the aforementioned technical issues, the present application is proposed. Embodiments of the present application provide an anti-lock braking system for a vehicle that obtains the vehicle's speed and acceleration at multiple predetermined time points within a predetermined time period. Deep learning-based artificial intelligence (AI) technology is employed to mine dynamic characteristics of the vehicle's speed and acceleration over time, thereby accurately identifying road conditions based on the vehicle's actual state, including road impact, poor and washboard road conditions, engine drag, and low-adhesion road conditions. This improves the applicability of the anti-lock braking system across a variety of road conditions and enhances its practicality.

[0006] In a first aspect, an anti-lock braking system for a vehicle is provided, comprising:

[0007] Signal preprocessing module, vehicle speed signal estimation module, wheel signal and slip ratio calculation module, vehicle state estimation module, master cylinder pressure estimation module, failure processing module, vehicle control module, front and rear axle control module, master cylinder pressure correction module and hydraulic control module;

[0008] The signal preprocessing module is used to preprocess the input signal and output the preprocessing result;

[0009] The vehicle speed signal estimation module is used to estimate the vehicle speed and acceleration;

[0010] The wheel signal and slip ratio calculation module is used to calculate the slip ratio based on the vehicle speed and acceleration;

[0011] The vehicle state estimation module is used to determine the state of the vehicle during driving based on the speed and acceleration of the vehicle;

[0012] the master cylinder pressure estimation module, configured to calculate an estimated value of a master cylinder pressure signal based on the vehicle speed and acceleration, the slip ratio, and the road surface condition;

[0013] The failure processing module is used to process a failed sensor;

[0014] The vehicle control module is used to determine whether to enable the anti-lock braking function;

[0015] The front and rear axle control modules are configured to calculate target pressures for axle control and wheel control algorithms based on the vehicle speed and body acceleration;

[0016] The master cylinder pressure correction module is configured to calculate a correction value of the master cylinder pressure based on the vehicle body deceleration and the wheel deceleration; and

[0017] The hydraulic control module is used to perform wheel cylinder hydraulic pressure target control.

[0018] In the above-mentioned anti-lock braking system for a vehicle, the vehicle state estimation module includes: a vehicle state data acquisition unit, used to obtain the vehicle speed and acceleration of the vehicle at multiple predetermined time points within a predetermined time period; a parameter time series arrangement unit, used to arrange the vehicle speed and the acceleration of the vehicle at the multiple predetermined time points into a speed input vector and an acceleration input vector according to the time dimension; a speed change feature extraction unit, used to pass the speed input vector through the first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; an acceleration change feature extraction unit, used to extract the speed input vector The acceleration input vector is passed through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector; a feature fusion unit is used to fuse the speed time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector; a feature optimization unit is used to optimize the feature distribution of the vehicle state feature vector based on the speed time series feature vector and the acceleration time series feature vector to obtain an optimized vehicle state feature vector; and a vehicle state detection unit is used to pass the optimized vehicle state feature vector through a multi-label classifier to obtain a classification result, and the classification result is used to represent a label of the road surface state.

[0019] In the above-mentioned anti-lock braking system for a vehicle, the first multi-scale neighborhood feature extraction module includes a first convolution layer and a second convolution layer in parallel, and a first cascade layer connected to the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, the second convolution layer uses a one-dimensional convolution layer with a second scale, and the first scale is different from the second scale; and the second multi-scale neighborhood feature extraction module includes a third convolution layer and a fourth convolution layer in parallel, and a second cascade layer connected to the third convolution layer and the fourth convolution layer, wherein the third convolution layer uses a one-dimensional convolution kernel with a third scale, the fourth convolution layer uses a one-dimensional convolution layer with a fourth scale, and the third scale is different from the fourth scale.

[0020] In the above-mentioned anti-lock braking system for a vehicle, the speed change feature extraction unit includes: a first scale feature extraction subunit, configured to use the first convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the speed input vector using the following formula to obtain the first scale speed feature vector; wherein the formula is:

[0021]

[0022] Among them, a is the width of the first convolution kernel in the x direction, is the first convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the first convolution kernel, represents the velocity input vector;

[0023] The second scale feature extraction subunit is configured to use the second convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the velocity input vector using the following formula to obtain the second scale velocity feature vector; wherein the formula is:

[0024]

[0025] Among them, b is the width of the second convolution kernel in the x direction, is the second convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the second convolution kernel, represents the speed input vector; and a first cascading subunit, configured to cascade the first-scale speed feature vector and the second-scale speed feature vector to obtain the speed time series feature vector.

[0026] In the above-mentioned anti-lock braking system for a vehicle, the acceleration change feature extraction unit includes: a third-scale feature extraction subunit, configured to use the third convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the third-scale acceleration feature vector; wherein the formula is:

[0027]

[0028] in, is the width of the third convolution kernel in the x direction, is the third convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the third convolution kernel, represents the acceleration input vector;

[0029] a fourth-scale feature extraction subunit, configured to use the fourth convolutional layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the fourth-scale acceleration feature vector; wherein the formula is:

[0030]

[0031] in, is the width of the fourth convolution kernel in the x direction, is the fourth convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the fourth convolution kernel, represents the acceleration input vector; and a second cascading subunit, configured to cascade the third-scale acceleration feature vector and the fourth-scale acceleration feature vector to obtain the acceleration time series feature vector.

[0032] In the above-mentioned anti-lock braking system for a vehicle, the feature fusion unit is used to fuse the velocity time series feature vector and the acceleration time series feature vector using the following formula to obtain a vehicle state feature vector; wherein the formula is:

[0033]

[0034] in, is the vehicle state feature vector, is the velocity time series feature vector, is the acceleration time series eigenvector, represents positional addition, is a weighting parameter for controlling the balance between the velocity time series eigenvector and the acceleration time series eigenvector.

[0035] In the above-mentioned anti-lock braking system for a vehicle, the feature optimization unit includes: a first optimization subunit, used to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the velocity time series feature vector to obtain a first optimized vehicle state feature vector; a second optimization subunit, used to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector to obtain a second optimized vehicle state feature vector; and an optimization feature fusion subunit, used to fuse the first optimized vehicle state feature vector and the second optimized vehicle state feature vector to obtain the optimized vehicle state feature vector.

[0036] In the above-mentioned anti-lock braking system for a vehicle, the first optimization subunit is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector using the following formula to obtain a first optimized vehicle state feature vector; wherein the formula is:

[0037]

[0038] in, 、 and represent the speed time series feature vector, the vehicle state feature vector and the first optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0039] In the above-mentioned anti-lock braking system for a vehicle, the second optimization subunit is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector using the following formula to obtain a second optimized vehicle state feature vector; wherein the formula is:

[0040]

[0041] in, 、 and represent the acceleration time series feature vector, the vehicle state feature vector and the second optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0042] In the above-mentioned anti-lock braking system for a vehicle, the vehicle state detection unit includes: a fully connected encoding subunit, used to use multiple fully connected layers of the classifier to fully connect encode the optimized vehicle state feature vector to obtain an encoded classification feature vector; and a classification subunit, used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0043] Compared with the existing technology, the anti-lock braking system for vehicles provided by the present application obtains the vehicle speed and acceleration at multiple predetermined time points within a predetermined time period; it adopts artificial intelligence technology based on deep learning to mine the dynamic change characteristic information of the vehicle speed and acceleration in the time dimension, so as to accurately identify the road surface condition based on the actual state of the vehicle, including road impact, bad road surface and strong washboard road surface, engine drag, low-adhesion road surface, etc., so as to improve the applicability of the anti-lock braking system in various road conditions and enhance its practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0045] Figure 1 This is a diagram of an application scenario of an anti-lock braking system for a vehicle according to an embodiment of the present application.

[0046] Figure 2 Schematic diagram of an anti-lock braking control system according to an embodiment of the present application.

[0047] Figure 3 Schematic diagram of the structure of a hydraulic control module according to an embodiment of the present application.

[0048] Figure 4 4 is a block diagram of the vehicle state estimation module in the anti-lock braking system for a vehicle according to an embodiment of the present application.

[0049] Figure 5 4 is a block diagram of the speed change feature extraction unit in the anti-lock braking system for a vehicle according to an embodiment of the present application.

[0050] Figure 6 4 is a block diagram of the acceleration change feature extraction unit in the anti-lock braking system for a vehicle according to an embodiment of the present application.

[0051] Figure 7 4 is a block diagram of the feature optimization unit in the anti-lock braking system for a vehicle according to an embodiment of the present application.

[0052] Figure 8 4 is a block diagram of the vehicle state detection unit in the anti-lock braking system for a vehicle according to an embodiment of the present application.

[0053] Figure 9 4 is a flow chart of an anti-lock braking method for a vehicle according to an embodiment of the present application.

[0054] Figure 10 Schematic diagram of a system architecture of an anti-lock braking method for a vehicle according to an embodiment of the present application.

[0055] Figure 11 2 is a schematic diagram of a vehicle speed estimation process according to an embodiment of the present application.

[0056] Figure 12 2 is a schematic diagram of an acceleration estimation process according to an embodiment of the present application.

[0057] Figure 13This is a flowchart of single-wheel impact road bump recognition according to an embodiment of the present application. DETAILED DESCRIPTION

[0058] The following will describe the technical solutions in the embodiments of this application in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0059] Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meaning as those commonly understood by those skilled in the art in the art of the present application. The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit the scope of this application.

[0060] In the description of the embodiments of this application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood in a broad sense. For example, it can be an electrical connection, or it can be the internal connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meaning of the above terms can be understood according to the specific circumstances.

[0061] It should be noted that the terms "first, second, and third" in the embodiments of the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understood that the terms "first, second, and third" can be interchanged to represent a specific order or precedence where permitted. It should be understood that the objects distinguished by "first, second, and third" can be interchanged where appropriate, such that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0062] As mentioned above, when testing for fungi, traditional medical bacterial collection detection systems can typically only detect colonies in a single culture dish. Furthermore, the detection process relies on manual observation and labeling of bacterial species. This not only requires dedicated personnel but also risks manual oversight during operation, resulting in unnecessary harm. Therefore, an optimized preventive medical testing system is desired.

[0063] As mentioned above, Chinese Patent Application Publication No. CN 114715095A discloses an anti-lock braking system and control method. While this method can address the issue of accidental activation of the anti-lock braking function when a vehicle temporarily drives over a bumpy surface, such as a speed bump or a pothole, the anti-lock braking control system in this system, while capable of integrating multiple sensors to assist in calculations, is costly and has limited engineering value. Therefore, an optimized anti-lock braking system for vehicles is desired.

[0064] Specifically, a vehicle anti-lock braking system (ABS) is proposed. This system is centered around vehicle speed estimation, vehicle state identification, master cylinder pressure estimation, and hydraulic control methods. It is adaptable to various road conditions and addresses existing issues such as poor practicality in vehicle speed estimation, complex vehicle state identification, and a lack of hydraulic control. Specifically, the ABS system includes: preprocessing an input signal and outputting a preprocessing result. Specifically, the preprocessing output includes the number of unavailable wheel speeds, sorted wheel speed values, and corresponding position numbers. Furthermore, when the wheel speeds are normal, the average of the coaxial wheel speeds is output; when the wheel speeds are unreliable, the axle speed output value is replaced with the minimum wheel speed of the entire vehicle. The system also outputs the vehicle's acceleration and maximum acceleration. The system also estimates the vehicle's speed and acceleration. The slip ratio is calculated based on the vehicle's speed and acceleration. The vehicle's driving state is determined based on the vehicle's speed and acceleration. An estimated master cylinder pressure signal is calculated based on the vehicle's speed and acceleration, the slip ratio, and the road surface condition. Address any sensor failures. Determine whether to enable the anti-lock braking system. Calculate target pressures for axle and wheel control algorithms based on the vehicle's speed and acceleration. Calculate a correction value for master cylinder pressure based on vehicle deceleration and wheel deceleration. Execute target wheel cylinder hydraulic pressure control.

[0065] Accordingly, considering that in actual vehicle anti-lock braking systems, accurately detecting the vehicle's state during driving and thereby precisely identifying the road surface state is crucial, road surface conditions include impact, poor and washboard surfaces, engine drag, and contact surfaces. Vehicle conditions can exhibit similar characteristic information across different road surfaces, making it difficult to accurately distinguish and identify these conditions. This reduces the accuracy of road surface identification and the practicality of the anti-lock braking system. In particular, considering that the vehicle's wheel speed and acceleration can reflect characteristic information about the vehicle's state, the technical solution of this application aims to detect and determine road surface conditions by analyzing the vehicle's speed and acceleration. The challenge in this process lies in accurately extracting the dynamic characteristic information of the vehicle's speed and acceleration over time, thereby accurately identifying road surface conditions based on the vehicle's actual state, including impact, poor and washboard surfaces, engine drag, and low-adhesion surfaces. This improves the applicability of the anti-lock braking system across various road conditions and enhances its practicality.

[0066] In recent years, deep learning and neural networks have been widely used in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated capabilities approaching or even surpassing those of humans in areas such as image classification, object detection, semantic segmentation, and text translation.

[0067] The development of deep learning and neural networks provides new solutions and plans for mining the dynamic change characteristic information of the vehicle speed and acceleration in the time dimension.

[0068] Specifically, in the technical solution of the present application, first, the vehicle speed and acceleration at multiple predetermined time points within a predetermined time period are obtained. Next, considering that the vehicle speed and acceleration exhibit dynamic variations in the time dimension, in order to accurately and deeply mine these dynamic variation characteristics and thereby accurately perform road condition detection, the vehicle speed and acceleration at the multiple predetermined time points need to be arranged according to the time dimension into a speed input vector and an acceleration input vector, respectively, to integrate the data distribution information of the vehicle speed and acceleration in the time dimension.

[0069] Then, considering that the vehicle speed at multiple predetermined time points within the predetermined time period exhibits fluctuation and uncertainty in the temporal dimension, the vehicle speed exhibits different dynamic change characteristics at different time period spans within the predetermined time period. Therefore, to fully extract the dynamic characteristics of the vehicle speed in time series, the speed input vector is further subjected to feature mining in a first multi-scale neighborhood feature extraction module to extract dynamic multi-scale neighborhood correlation features of the vehicle speed at different time spans within the predetermined time period, thereby obtaining a speed time series feature vector.

[0070] Furthermore, for the acceleration of the vehicle at multiple predetermined time points within the predetermined time period, the vehicle acceleration also has a dynamic change pattern in the time dimension, and it also has different patterns of dynamic change characteristics under different time period spans within the predetermined time period. Therefore, in the technical solution of the present application, in order to fully and accurately extract the dynamic correlation feature information of the vehicle's acceleration in time series, so as to accurately detect the road surface state, the acceleration input vector is further subjected to feature mining in the second multi-scale neighborhood feature extraction module to extract the dynamic multi-scale neighborhood correlation feature information of the vehicle's acceleration under different time spans, that is, the time series multi-scale dynamic change feature information of the vehicle's acceleration, thereby obtaining the acceleration time series feature vector.

[0071] Next, the speed time series feature vector and the acceleration time series feature vector are further fused to fuse the time series dynamic multi-scale change characteristics of the vehicle speed and the time series dynamic multi-scale change characteristics of the vehicle acceleration, thereby obtaining a vehicle state feature vector having time series dynamic multi-scale correlation characteristics of the coordinated state change of the vehicle speed and acceleration. Then, the vehicle state feature vector is used as a classification feature vector and passed through a multi-label classifier to obtain a classification result for a label representing the road surface state. That is, in the technical solution of the present application, the label of the classifier is the label of the road surface state, wherein the classifier determines to which classification label the classification feature vector belongs through a soft maximum function, thereby accurately identifying the road surface state based on the actual state of the vehicle, including road impact, bad road surface and strong washboard road surface, engine drag, low-adhesion road surface, etc., to improve the practicality of the anti-lock braking system in various road conditions.

[0072] In particular, in the technical solution of the present application, when the vehicle state feature vector is obtained by fusing the velocity time series feature vector and the acceleration time series feature vector, for example, by weighted point addition, it is expected to improve the responsiveness of the feature distribution of the vehicle state feature vector relative to the respective feature distributions of the velocity time series feature vector and the acceleration time series feature vector, thereby achieving better fusion in the feature vector dimension.

[0073] Furthermore, considering that the velocity time-series feature vector and the acceleration time-series feature vector each express different temporal orders (i.e., velocity corresponds to the first-order derivative of distance with respect to time, while acceleration corresponds to the first-order derivative of velocity with respect to time), the target position time-series absolute feature vector and the target position time-series change feature vector are each considered as the source feature vector, while the target position feature vector is considered as the response feature vector of the source feature vector. In this way, by integrating the features of the source feature vector with which it has a response relationship, the responsiveness of the feature distribution of the vehicle state feature vector relative to the feature distributions of the velocity time-series feature vector and the acceleration time-series feature vector can be improved.

[0074] Specifically, for the vehicle state feature vector, for example, it is recorded as , respectively calculate its and the speed time series feature vector, for example, recorded as And the acceleration time series characteristic vector, for example, is recorded as The incoherent sparse response fusion optimization of and , respectively expressed as:

[0075]

[0076]

[0077] in and represents the first and second norms of a vector, is the length of the vector, and denote vector product and vector dot product respectively, and all vectors are in row vector form.

[0078] Here, the incoherent sparse response fusion is to respectively transform the velocity time series feature vector and the acceleration time series characteristic vector As the ground-truth distribution of the response between feature domains, the fuzzy bit distribution responsiveness of the vector difference represented by a norm and the real differential embedding responsiveness based on the modulus constraint of the differential vector are used to obtain the incoherent sparse fusion representation between vectors to improve the descriptiveness of the fusion probability distribution of the feature vectors with a response relationship, thereby improving the feature fusion of the response feature vector with the source feature vector with a response relationship. In this way, the optimized feature vector is then fused. and By using the vehicle state feature vector to obtain the vehicle state feature vector, the vehicle state feature vector can be integrated with the velocity time series feature vector and the acceleration time series feature vector, thereby improving the accuracy of the classification results obtained by the vehicle state feature vector through the classifier. In this way, the road surface condition can be accurately identified based on the actual state of the vehicle, thereby improving the applicability of the anti-lock braking system in various road conditions and enhancing its practicality.

[0079] Figure 1 FIG is an application scenario diagram of an anti-lock braking system for a vehicle according to an embodiment of the present application. Figure 1 As shown, in this application scenario, first, the vehicle speed of the vehicle at multiple predetermined time points within a predetermined time period is obtained (for example, Figure 1 C1 as shown in the figure) and acceleration (e.g. Figure 1 Then, the obtained vehicle speed and acceleration are input to a server (eg, Figure 1 In S) as shown in , the server is capable of processing the vehicle speed and the acceleration based on an anti-lock braking algorithm for the vehicle to generate a classification result of a label representing a road surface state.

[0080] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0081] In the vehicle anti-lock braking control system, wheel speed sensors collect the wheel speeds of each vehicle wheel, input the information into a control unit for reference vehicle speed estimation, calculate wheel acceleration and reference slip ratio, set different wheel acceleration and reference slip ratio thresholds based on different road surfaces, and output solenoid valve control signals to a hydraulic control unit via the control unit to adjust the pressure of each brake wheel cylinder and control the vehicle wheels.

[0082] The anti-lock braking control system software runs in 5ms and 20ms cycles respectively. In one embodiment of the present application, Figure 2 FIG. 1 is a general schematic diagram of an anti-lock braking control system according to an embodiment of the present application, as shown in FIG. Figure 2 As shown, the anti-lock braking control system 100 mainly includes 10 control modules: a signal preprocessing module 1, a vehicle speed signal estimation module 2, a wheel signal and slip ratio calculation module 3, a vehicle state estimation module 4, a master cylinder pressure estimation module 5, a failure processing module 6, a vehicle control module 7, a front and rear axle control module 8, a master cylinder pressure correction module 9, and a hydraulic control module 10; the signal preprocessing module 1 is used to preprocess the input signal and output the preprocessing result; the vehicle speed signal estimation module 2 is used to estimate the vehicle speed and acceleration; the wheel signal and slip ratio calculation module 3 is used to calculate the slip ratio based on the vehicle speed and acceleration; the vehicle state estimation module 4 is used to calculate the slip ratio based on the vehicle state. The vehicle speed and acceleration are used to determine the state of the vehicle during driving; the master cylinder pressure estimation module 5 is used to calculate the estimated value of the master cylinder pressure signal based on the vehicle speed and acceleration, the slip rate and the road surface state; the failure processing module 6 is used to handle the faulty sensor; the vehicle control module 7 is used to determine whether to turn on the anti-lock braking function; the front and rear axle control modules 8 are used to calculate the target pressure of the axle control and wheel control algorithms based on the vehicle speed and body acceleration; the master cylinder pressure correction module 9 is used to calculate the correction value of the master cylinder pressure based on the body deceleration and wheel deceleration; and the hydraulic control module 10 is used to perform wheel cylinder hydraulic pressure target control.

[0083] Specifically, the signal preprocessing module 1 is the first module in the ABS algorithm. It preprocesses the signals input from other modules and outputs them as relevant signals required for calculations by subsequent modules. It primarily includes: 1) converting the external target braking force into the vehicle's target brake fluid pressure required by the ABS; 2) determining the target states of the ABS and EBD functions by diagnosing the target state and wheel speed sensor signal quality; and 3) preprocessing the wheel speed and wheel acceleration signals sent by the wheel speed sensor signal processing module, converting them into the wheel speed, axle speed, and acceleration signals required for ABS control, facilitating vehicle speed and acceleration calculations by subsequent modules. The signal preprocessing module 1 outputs: the highest wheel speed, the second highest wheel speed, the second lowest wheel speed, the lowest wheel speed, the wheel numbers corresponding to these four wheel speeds, the number of faulty wheel speed sensors, the axle speeds of the front and rear axles, the average axle deceleration, the average deceleration of all four wheels, the highest wheel deceleration, and the highest, second highest, second lowest, and lowest wheel speeds after four-wheel geometry compensation, as well as their wheel numbers.

[0084] The vehicle speed signal estimation module 2 primarily estimates vehicle speed and acceleration information. While ABS systems may utilize multiple sensors, wheel speed signals are essential for all vehicle types and operating conditions, depending on vehicle configuration and sensor failure. Therefore, ABS vehicle speed estimation primarily relies on wheel speed sensors. A crucial parameter in speed estimation is the gradient. An appropriate reference wheel must be selected based on wheel dynamics, and the vehicle body acceleration is estimated using the reference wheel acceleration.

[0085] The wheel signal and slip ratio calculation module 3 primarily calculates slip ratio based on vehicle speed and includes the following functions: 1) four-wheel slip ratio calculation; 2) wheel deceleration average calculation; 3) wheel deceleration average filtered value calculation; 4) rear axle slip ratio calculation; and 5) front axle slip ratio calculation. This module provides input to other corresponding modules.

[0086] The vehicle state estimation module 4 mainly determines the state of the vehicle during driving through vehicle speed and body acceleration. Its main functions include: 1) road impact recognition; 2) bad road surface recognition; 3) high-adhesion road surface recognition; 4) low-adhesion road surface recognition; 4) separation road surface state monitoring; 5) low-adhesion EBD monitoring; 6) coasting monitoring.

[0087] The master cylinder pressure estimation module 5 calculates an estimated master cylinder pressure signal based on vehicle speed, acceleration, slip ratio, and road conditions. This estimated master cylinder pressure signal is sent to the wheel control and axle control state machines for vehicle state control. The calculated signal is also sent to the hydraulic control module to support hydraulic control calculation of wheel cylinder pressures. Key functions include: 1) master cylinder pressure calculation module; 2) wheel control monitoring enablement; 3) external control monitoring enablement; and 4) wheel cylinder pressure control monitoring.

[0088] The failure processing module 6 is a failure processing algorithm. When a sensor fails, the ABS needs to shut down related functions and provide a target brake pressure through this algorithm. This module includes: 1) a failure monitoring and function arbitration module; 2) a calculation of the target brake pressure after failure.

[0089] The vehicle control module 7 is used to implement the ABS control algorithm module, which determines the opening and closing of the ABS function, as well as the setting of the axle control command and the calculation of the axle control target pressure. It mainly includes the following functions: 1) setting the ABS entry and exit speed thresholds; 2) vehicle control, including the setting of the axle control command and the calculation of the axle control target pressure.

[0090] The front and rear axle control module 8 calculates the target pressure of the axle control and wheel control algorithms based on the vehicle speed and body acceleration to achieve independent control of the front and rear axles and four wheels. The output control signals include the axle control target state, the wheel control target state and the target pressure of each wheel, and are sent to the hydraulic control module for execution through the hydraulic control module interface.

[0091] The master cylinder pressure correction module 9 calculates the correction value of the master cylinder pressure based on the vehicle body deceleration and wheel deceleration, and feeds it back to all pressure-related algorithms for compensation.

[0092] The hydraulic control module 10 is mainly responsible for executing the wheel cylinder hydraulic pressure target control of the upper control module, and its main functions include: wheel cylinder pressure control and motor control.

[0093] Figure 3 This is a schematic diagram of the structure of the hydraulic control module according to an embodiment of the present application. The structure of the hydraulic control module 10 is as follows: Figure 3 As shown, the wheel cylinder pressure control described in (1) estimates the model pressure of the wheel cylinder based on the EV control current and AV control time; secondly, the EV control current and AV control time are calculated based on the target wheel cylinder pressure to achieve the purpose of wheel cylinder pressure control and model pressure estimation.

[0094] (2) The motor target speed calculation is used to calculate the motor target speed based on the accumulator fluid volume, ensuring that the accumulator is not fully filled during ABS operation and providing sufficient boost capacity to provide a target for motor control.

[0095] (3) The motor pump load calculation described above is used to process calculations related to the motor pump load, including accumulator fluid volume estimation, pump load calculation, etc.

[0096] (4) The motor speed control described above controls the motor speed based on the target motor speed output by the motor target speed calculation module and the motor pump load calculation and control module, as well as the actual motor load. First, the load, speed, and other information are converted into equivalent voltage quantities to unify the calculation basis. Then, based on the motor dynamics equation and back electromotive force sampling, the Kalman filter method is used to estimate the actual motor speed. Then, based on the speed target, a feedforward plus feedback speed closed-loop controller is established to calculate the duty cycle of the actual drive motor. Finally, based on the actual driving situation, an interrupt drive instruction is appropriately issued to perform back electromotive force sampling.

[0097] Specifically, the vehicle anti-lock braking system (ABS) is a vehicle anti-lock braking system and control method, centered around vehicle speed estimation, vehicle state identification, master cylinder pressure estimation, and hydraulic control methods. It is adaptable to various road conditions and addresses existing issues such as poor practicality of vehicle speed estimation, complex vehicle state identification, and lack of hydraulic control. Specifically, the ABS system includes: preprocessing an input signal and outputting a preprocessing result. Specifically, the preprocessing output includes the number of unavailable wheel speeds, sorted wheel speed values, and corresponding position numbers. Furthermore, when the wheel speeds are normal, the average of the coaxial wheel speeds is output; when the wheel speeds are unreliable, the axle speed output value is replaced with the minimum wheel speed of the entire vehicle. The system also outputs the vehicle acceleration and maximum acceleration. The system also estimates the vehicle speed and acceleration. The slip ratio is calculated based on the vehicle speed and acceleration. The vehicle's driving state is determined based on the vehicle speed and acceleration. An estimated master cylinder pressure signal is calculated based on the vehicle speed and acceleration, the slip ratio, and the road surface condition. Address any sensor failures. Determine whether to enable the anti-lock braking system. Calculate target pressures for axle and wheel control algorithms based on the vehicle's speed and acceleration. Calculate a correction value for master cylinder pressure based on vehicle deceleration and wheel deceleration. Execute target wheel cylinder hydraulic pressure control.

[0098] It should be understood that the present invention has the following advantages:

[0099] 1. The anti-lock braking control system and method of the present invention only uses wheel speed sensors and does not require other sensors for auxiliary implementation, so it has low cost and good stability.

[0100] 2. The anti-lock braking control system and method of the present invention uses wheel speed to estimate vehicle speed. The vehicle speed estimation method is divided into three cases: when the wheel speed sensor is damaged, when the vehicle is not under ABS control, and when the vehicle is under ABS control. Different methods are used to estimate the vehicle speed in each case, meeting various working conditions, with strong robustness and practicality.

[0101] 3. The anti-lock braking control system and method of the present invention identifies vehicle status, including road impact identification, bad road surface and strong washboard road surface identification, engine drag identification, low-adhesion road surface identification, etc., and is applicable to various road conditions.

[0102] 4. The anti-lock brake control system and method of the present invention controls the hydraulic actuator unit, including estimating master cylinder pressure and wheel cylinder pressure and controlling the solenoid valve. Compared with existing anti-lock brake control systems and methods, this method fills a gap in the existing method.

[0103] Figure 4 FIG. 1 is a block diagram of the vehicle state estimation module in the anti-lock braking system for a vehicle according to an embodiment of the present application. Figure 4 As shown, the vehicle state estimation module 4 includes: a vehicle state data acquisition unit 110, for acquiring the vehicle speed and acceleration of the vehicle at multiple predetermined time points within a predetermined time period; a parameter time series arrangement unit 120, for arranging the vehicle speed and the acceleration of the vehicle at the multiple predetermined time points into a speed input vector and an acceleration input vector according to the time dimension; a speed change feature extraction unit 130, for passing the speed input vector through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; an acceleration change feature extraction unit 140, for arranging the acceleration input vector into a speed time series feature vector. The input vector passes through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector; a feature fusion unit 150 is used to fuse the speed time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector; a feature optimization unit 160 is used to optimize the feature distribution of the vehicle state feature vector based on the speed time series feature vector and the acceleration time series feature vector to obtain an optimized vehicle state feature vector; and a vehicle state detection unit 170 is used to pass the optimized vehicle state feature vector through a multi-label classifier to obtain a classification result, and the classification result is used to represent a label of the road surface state.

[0104] Specifically, in an embodiment of the present application, the vehicle state data acquisition unit 110 is used to obtain the vehicle speed and acceleration of the vehicle at multiple predetermined time points within a predetermined time period. Considering that in an actual vehicle's anti-lock braking system, how to accurately detect the vehicle's state during driving in order to accurately identify the road surface state is key, however, since the road surface state includes road impact, bad road surface and strong washboard road surface, engine drag, and docking road surface, and the vehicle state has similar characteristic information on different road surfaces, it is difficult to accurately distinguish and identify them, thereby reducing the accuracy of road surface identification and the practicality of the anti-lock braking system. In particular, considering that the vehicle's wheel speed and acceleration can reflect the vehicle's state characteristic information.

[0105] Therefore, the technical solution of this application aims to detect and determine road conditions by analyzing the vehicle's speed and acceleration. The challenge in this process lies in accurately extracting the dynamic characteristics of the vehicle's speed and acceleration over time, thereby accurately identifying road conditions based on the vehicle's actual condition, including road impact, poor and washboard surfaces, engine drag, and low-adhesion surfaces. This improves the applicability of the anti-lock braking system across various road conditions and enhances its practicality.

[0106] In recent years, deep learning and neural networks have been widely used in fields such as computer vision, natural language processing, and text signal processing. Furthermore, deep learning and neural networks have demonstrated capabilities approaching or even surpassing those of humans in areas such as image classification, object detection, semantic segmentation, and text translation.

[0107] The development of deep learning and neural networks provides new solutions and plans for mining the dynamic change characteristic information of the vehicle speed and acceleration in the time dimension.

[0108] Specifically, in the technical solution of the present application, first, the speed and acceleration of the vehicle at multiple predetermined time points within a predetermined time period are obtained.

[0109] Specifically, in the embodiment of the present application, the parameter time series arrangement unit 120 is configured to arrange the vehicle speeds and accelerations at the plurality of predetermined time points into a speed input vector and an acceleration input vector according to the time dimension. Next, considering that both the vehicle speed and acceleration exhibit dynamic variations in the time dimension, in order to accurately and deeply exploit these dynamic variation characteristics and thereby accurately perform road condition detection, it is necessary to arrange the vehicle speeds and accelerations at the plurality of predetermined time points into a speed input vector and an acceleration input vector according to the time dimension, thereby integrating the data distribution information of the vehicle speed and acceleration in the time dimension.

[0110] Specifically, in an embodiment of the present application, the speed change feature extraction unit 130 and the acceleration change feature extraction unit 140 are used to pass the speed input vector through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; and to pass the acceleration input vector through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector.

[0111] Then, considering that the vehicle speed at multiple predetermined time points within the predetermined time period exhibits fluctuation and uncertainty in the temporal dimension, the vehicle speed exhibits different dynamic change characteristics at different time period spans within the predetermined time period. Therefore, to fully extract the dynamic characteristics of the vehicle speed in time series, the speed input vector is further subjected to feature mining in a first multi-scale neighborhood feature extraction module to extract dynamic multi-scale neighborhood correlation features of the vehicle speed at different time spans within the predetermined time period, thereby obtaining a speed time series feature vector.

[0112] Furthermore, for the acceleration of the vehicle at multiple predetermined time points within the predetermined time period, the vehicle acceleration also has a dynamic change pattern in the time dimension, and it also has different patterns of dynamic change characteristics under different time period spans within the predetermined time period. Therefore, in the technical solution of the present application, in order to fully and accurately extract the dynamic correlation feature information of the vehicle's acceleration in time series, so as to accurately detect the road surface state, the acceleration input vector is further subjected to feature mining in the second multi-scale neighborhood feature extraction module to extract the dynamic multi-scale neighborhood correlation feature information of the vehicle's acceleration under different time spans, that is, the time series multi-scale dynamic change feature information of the vehicle's acceleration, thereby obtaining the acceleration time series feature vector.

[0113] Among them, the first multi-scale neighborhood feature extraction module 130 includes a first convolution layer and a second convolution layer in parallel, and a first cascade layer connected to the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, the second convolution layer uses a one-dimensional convolution layer with a second scale, and the first scale is different from the second scale; and the second multi-scale neighborhood feature extraction module 134 includes a third convolution layer and a fourth convolution layer in parallel, and a second cascade layer connected to the third convolution layer and the fourth convolution layer, wherein the third convolution layer uses a one-dimensional convolution kernel with a third scale, the fourth convolution layer uses a one-dimensional convolution layer with a fourth scale, and the third scale is different from the fourth scale.

[0114] In this application, Figure 5 FIG. 1 is a block diagram of the speed change feature extraction unit in the anti-lock braking system for a vehicle according to an embodiment of the present application, as shown in FIG. Figure 5 As shown, the speed change feature extraction unit 130 includes: a first scale feature extraction subunit 131, configured to use the first convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the speed input vector using the following formula to obtain the first scale speed feature vector; wherein the formula is:

[0115]

[0116] Among them, a is the width of the first convolution kernel in the x direction, is the first convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the first convolution kernel, represents the speed input vector; and a second-scale feature extraction subunit 132, configured to use the second convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the speed input vector using the following formula to obtain the second-scale speed feature vector; wherein the formula is:

[0117]

[0118] Among them, b is the width of the second convolution kernel in the x direction, is the second convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the second convolution kernel, represents the velocity input vector; and a first cascading subunit 133, configured to cascade the first-scale velocity feature vector and the second-scale velocity feature vector to obtain the velocity time series feature vector.

[0119] Furthermore, Figure 6 FIG. 1 is a block diagram of the acceleration change feature extraction unit in the anti-lock braking system for a vehicle according to an embodiment of the present application, as shown in FIG. Figure 6 As shown, the acceleration change feature extraction unit 140 includes: a third-scale feature extraction subunit 141, configured to use the third convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the third-scale acceleration feature vector; wherein the formula is:

[0120]

[0121] in, is the width of the third convolution kernel in the x direction, is the third convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the third convolution kernel, represents the acceleration input vector; and a fourth-scale feature extraction subunit 142, configured to use the fourth convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the fourth-scale acceleration feature vector; wherein the formula is:

[0122]

[0123] in, is the width of the fourth convolution kernel in the x direction, is the fourth convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the fourth convolution kernel, represents the acceleration input vector; and a second cascading subunit 143, configured to cascade the third-scale acceleration feature vector and the fourth-scale acceleration feature vector to obtain the acceleration time series feature vector.

[0124] It is worth mentioning that compared with traditional feature engineering, the multi-scale neighborhood feature extraction module is essentially a deep neural network model based on deep learning, which can fit any function through a predetermined training strategy and has higher feature extraction generalization ability.

[0125] The multi-scale neighborhood feature extraction module includes multiple parallel one-dimensional convolution layers. During the feature extraction process of the multi-scale neighborhood feature extraction module, the multiple parallel one-dimensional convolution layers respectively perform one-dimensional convolution encoding on the input data with one-dimensional convolution kernels of different scales to capture local implicit features of the sequence.

[0126] Specifically, in this embodiment of the present application, the feature fusion unit 160 is configured to fuse the velocity time-series feature vector and the acceleration time-series feature vector to obtain a vehicle state feature vector. Subsequently, the velocity time-series feature vector and the acceleration time-series feature vector are further fused to combine the temporal dynamic multi-scale variation characteristics of the vehicle velocity and the temporal dynamic multi-scale variation characteristics of the vehicle acceleration, thereby obtaining a vehicle state feature vector having temporal dynamic multi-scale correlation characteristics of the coordinated state variation of the vehicle velocity and acceleration.

[0127] The feature fusion unit 150 is configured to fuse the velocity time series feature vector and the acceleration time series feature vector using the following formula to obtain a vehicle state feature vector; wherein the formula is:

[0128]

[0129] in, is the vehicle state feature vector, is the velocity time series eigenvector, is the acceleration time series eigenvector, represents positional addition, is a weighting parameter for controlling the balance between the velocity time series eigenvector and the acceleration time series eigenvector.

[0130] Specifically, in the embodiment of the present application, the feature optimization unit 160 is configured to perform feature distribution optimization on the vehicle state feature vector based on the speed time series feature vector and the acceleration time series feature vector to obtain an optimized vehicle state feature vector.

[0131] In particular, in the technical solution of the present application, when the vehicle state feature vector is obtained by fusing the velocity time series feature vector and the acceleration time series feature vector, for example, by weighted point addition, it is expected to improve the responsiveness of the feature distribution of the vehicle state feature vector relative to the respective feature distributions of the velocity time series feature vector and the acceleration time series feature vector, thereby achieving better fusion in the feature vector dimension.

[0132] Furthermore, considering that the velocity time-series feature vector and the acceleration time-series feature vector each express different temporal orders (i.e., velocity corresponds to the first-order derivative of distance with respect to time, while acceleration corresponds to the first-order derivative of velocity with respect to time), the target position time-series absolute feature vector and the target position time-series change feature vector are each considered as the source feature vector, while the target position feature vector is considered as the response feature vector of the source feature vector. In this way, by integrating the features of the source feature vector with which it has a response relationship, the responsiveness of the feature distribution of the vehicle state feature vector relative to the feature distributions of the velocity time-series feature vector and the acceleration time-series feature vector can be improved.

[0133] Figure 7 FIG. 1 is a block diagram of the feature optimization unit in the anti-lock braking system for a vehicle according to an embodiment of the present application, as shown in FIG. Figure 7 As shown, the feature optimization unit 160 includes: a first optimization subunit 161, used to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector to obtain a first optimized vehicle state feature vector; a second optimization subunit 162, used to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector to obtain a second optimized vehicle state feature vector; and an optimization feature fusion subunit 163, used to fuse the first optimized vehicle state feature vector and the second optimized vehicle state feature vector to obtain the optimized vehicle state feature vector.

[0134] Among them, for the vehicle state feature vector, for example, it is recorded as , respectively calculate its and the speed time series feature vector, for example, recorded as And the acceleration time series characteristic vector, for example, is recorded as The incoherent sparse response fusion optimization of and The first optimization subunit 161 is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector using the following formula to obtain a first optimized vehicle state feature vector; wherein the formula is:

[0135]

[0136] in, 、 and represent the speed time series feature vector, the vehicle state feature vector and the first optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0137] The second optimization subunit 162 is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector using the following formula to obtain a second optimized vehicle state feature vector; wherein the formula is:

[0138]

[0139] in, 、 and represent the acceleration time series feature vector, the vehicle state feature vector and the second optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0140] Here, the incoherent sparse response fusion is to respectively transform the velocity time series feature vector and the acceleration time series characteristic vector As the ground-truth distribution of the response between feature domains, the fuzzy bit distribution responsiveness of the vector difference represented by a norm and the real differential embedding responsiveness based on the modulus constraint of the differential vector are used to obtain the incoherent sparse fusion representation between vectors to improve the descriptiveness of the fusion probability distribution of the feature vectors with a response relationship, thereby improving the feature fusion of the response feature vector with the source feature vector with a response relationship. In this way, the optimized feature vector is then fused. and By using the vehicle state feature vector to obtain the vehicle state feature vector, the vehicle state feature vector can be integrated with the velocity time series feature vector and the acceleration time series feature vector, thereby improving the accuracy of the classification results obtained by the vehicle state feature vector through the classifier. In this way, the road surface condition can be accurately identified based on the actual state of the vehicle, thereby improving the applicability of the anti-lock braking system in various road conditions and enhancing its practicality.

[0141] Specifically, in this embodiment of the present application, the vehicle state detection unit 170 is configured to pass the optimized vehicle state feature vector through a multi-label classifier to obtain a classification result, wherein the classification result is used to represent a label of the road surface state. The vehicle state feature vector is then used as a classification feature vector and passed through the multi-label classifier to obtain a classification result representing a label of the road surface state.

[0142] That is, in the technical solution of the present application, the label of the classifier is the label of the road surface state, wherein the classifier determines to which classification label the classification feature vector belongs through a soft maximum function, so as to accurately identify the road surface state based on the actual state of the vehicle, including road impact, bad road surface and strong washboard road surface, engine drag, low-adhesion road surface, etc., so as to improve the practicality of the anti-lock braking system in various road conditions.

[0143] Figure 8 FIG. 1 is a block diagram of the vehicle state detection unit in the anti-lock braking system for a vehicle according to an embodiment of the present application, as shown in FIG. Figure 8 As shown, the vehicle state detection unit 170 includes: a fully connected encoding subunit 171, which is used to use multiple fully connected layers of the classifier to fully connect encode the optimized vehicle state feature vector to obtain an encoded classification feature vector; and a classification subunit 172, which is used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0144] In a specific example of the present application, the classifier is used to process the optimized vehicle state feature vector using the following formula to generate a classification result, wherein the formula is: , where X represents the optimized vehicle state feature vector, to is the weight matrix, to Represents the bias vector.

[0145] In summary, an anti-lock braking system 100 for a vehicle based on an embodiment of the present application is explained, which obtains the vehicle speed and acceleration at multiple predetermined time points within a predetermined time period; adopts artificial intelligence technology based on deep learning to mine the dynamic change characteristic information of the vehicle speed and acceleration in the time dimension, so as to accurately identify the road surface condition based on the actual state of the vehicle, including road impact, bad road surface and strong washboard road surface, engine drag, low-adhesion road surface, etc., so as to improve the applicability of the anti-lock braking system in various road conditions and enhance its practicality.

[0146] As described above, the vehicle anti-lock braking system 100 according to an embodiment of the present application can be implemented in various terminal devices, such as a vehicle anti-lock braking server. In one example, the vehicle anti-lock braking system 100 according to an embodiment of the present application can be integrated into the terminal device as a software module and / or hardware module. For example, the vehicle anti-lock braking system 100 can be a software module in the terminal device's operating system, or can be an application developed specifically for the terminal device; of course, the vehicle anti-lock braking system 100 can also be one of the terminal device's many hardware modules.

[0147] Alternatively, in another example, the anti-lock braking system 100 for a vehicle and the terminal device may be separate devices, and the anti-lock braking system 100 for a vehicle may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0148] In one embodiment of the present application, Figure 9 FIG. 1 is a flow chart of an anti-lock braking method for a vehicle according to an embodiment of the present application. Figure 9 As shown, according to an embodiment of the present application, the anti-lock braking method for a vehicle includes: 210, obtaining the vehicle speed and acceleration of a plurality of predetermined time points within a predetermined time period; 220, arranging the vehicle speed at the plurality of predetermined time points and the acceleration of the vehicle at the plurality of predetermined time points into a speed input vector and an acceleration input vector according to the time dimension; 230, passing the speed input vector through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; 240, passing the acceleration input vector through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector; 250, fusing the speed time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector; 260, based on the speed time series feature vector and the acceleration time series feature vector, performing feature distribution optimization on the vehicle state feature vector to obtain an optimized vehicle state feature vector; and, 270, passing the optimized vehicle state feature vector through a multi-label classifier to obtain a classification result, and the classification result is used to represent a label of a road state.

[0149] Figure 10 FIG. 1 is a schematic diagram of a system architecture of an anti-lock braking method for a vehicle according to an embodiment of the present application. Figure 10As shown, in the system architecture of the anti-lock braking method for a vehicle, first, the vehicle speed and acceleration of the vehicle at multiple predetermined time points within a predetermined time period are obtained; then, the vehicle speed of the multiple predetermined time points and the acceleration of the vehicle at the multiple predetermined time points are arranged according to the time dimension as a speed input vector and an acceleration input vector; then, the speed input vector is passed through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; then, the acceleration input vector is passed through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector; then, the speed time series feature vector and the acceleration time series feature vector are fused to obtain a vehicle state feature vector; then, based on the speed time series feature vector and the acceleration time series feature vector, feature distribution optimization is performed on the vehicle state feature vector to obtain an optimized vehicle state feature vector; and finally, the optimized vehicle state feature vector is passed through a multi-label classifier to obtain a classification result, and the classification result is used to represent a label of a road surface state.

[0150] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, the first multi-scale neighborhood feature extraction module includes a first convolution layer and a second convolution layer in parallel, and a first cascade layer connected to the first convolution layer and the second convolution layer, wherein the first convolution layer uses a one-dimensional convolution kernel with a first scale, the second convolution layer uses a one-dimensional convolution layer with a second scale, and the first scale is different from the second scale; and the second multi-scale neighborhood feature extraction module includes a third convolution layer and a fourth convolution layer in parallel, and a second cascade layer connected to the third convolution layer and the fourth convolution layer, wherein the third convolution layer uses a one-dimensional convolution kernel with a third scale, the fourth convolution layer uses a one-dimensional convolution layer with a fourth scale, and the third scale is different from the fourth scale.

[0151] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, the speed input vector is passed through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector, including: using the first convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the speed input vector using the following formula to obtain the first-scale speed feature vector; wherein the formula is:

[0152]

[0153] Among them, a is the width of the first convolution kernel in the x direction, is the first convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the first convolution kernel, represents the velocity input vector;

[0154] The second convolution layer of the first multi-scale neighborhood feature extraction module performs one-dimensional convolution encoding on the velocity input vector using the following formula to obtain the second-scale velocity feature vector; wherein the formula is:

[0155]

[0156] Among them, b is the width of the second convolution kernel in the x direction, is the second convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the second convolution kernel, represents the speed input vector; and concatenating the first-scale speed feature vector and the second-scale speed feature vector to obtain the speed time series feature vector.

[0157] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, the acceleration input vector is passed through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector, including: using the third convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the third-scale acceleration feature vector; wherein the formula is:

[0158]

[0159] in, is the width of the third convolution kernel in the x direction, is the third convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the third convolution kernel, represents the acceleration input vector;

[0160] The fourth convolution layer of the second multi-scale neighborhood feature extraction module is used to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the fourth-scale acceleration feature vector; wherein the formula is:

[0161]

[0162] in, is the width of the fourth convolution kernel in the x direction, is the fourth convolution kernel parameter vector, is the local vector matrix operated with the convolution kernel function, is the size of the fourth convolution kernel, represents the acceleration input vector; and cascading the third-scale acceleration feature vector and the fourth-scale acceleration feature vector to obtain the acceleration time series feature vector.

[0163] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, fusing the velocity time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector includes: fusing the velocity time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector using the following formula; wherein the formula is:

[0164]

[0165] in, is the vehicle state feature vector, is the velocity time series feature vector, is the acceleration time series eigenvector, represents positional addition, is a weighting parameter for controlling the balance between the velocity time series eigenvector and the acceleration time series eigenvector.

[0166] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, based on the speed time series feature vector and the acceleration time series feature vector, the vehicle state feature vector is subjected to feature distribution optimization to obtain an optimized vehicle state feature vector, including: calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector to obtain a first optimized vehicle state feature vector; calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector to obtain a second optimized vehicle state feature vector; and, fusing the first optimized vehicle state feature vector and the second optimized vehicle state feature vector to obtain the optimized vehicle state feature vector.

[0167] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector to obtain a first optimized vehicle state feature vector further includes: calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector using the following formula to obtain the first optimized vehicle state feature vector; wherein the formula is:

[0168]

[0169] in, 、 and represent the speed time series feature vector, the vehicle state feature vector and the first optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0170] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector to obtain a second optimized vehicle state feature vector further includes: calculating the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector using the following formula to obtain the second optimized vehicle state feature vector; wherein the formula is:

[0171]

[0172] in, 、 and represent the acceleration time series feature vector, the vehicle state feature vector and the second optimized vehicle state feature vector respectively, and They represent the first norm and second norm of the vector respectively, is the length of the vector, and Represent vector product and vector dot product respectively, and all vectors are in row vector form, is the transpose of a vector.

[0173] In a specific example, in the above-mentioned anti-lock braking method for a vehicle, the optimized vehicle state feature vector is passed through a multi-label classifier to obtain a classification result, and the classification result is used to represent the label of the road state, including: using multiple fully connected layers of the classifier to fully connect encode the optimized vehicle state feature vector to obtain an encoded classification feature vector; and, passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0174] Those skilled in the art will appreciate that the specific operations of each step in the above-mentioned anti-lock braking method for a vehicle have been described in detail above. Figures 1 to 8 The anti-lock braking system for a vehicle has been described in detail, and therefore, its repeated description will be omitted.

[0175] Specifically, in this application, the present invention provides a method for automobile anti-lock braking control, wherein the method mainly includes: a vehicle speed estimation method, a vehicle acceleration estimation method, a vehicle state identification method, a slip ratio calculation method, a master cylinder pressure estimation method, an ABS control method, and a hydraulic control method.

[0176] Wherein, the vehicle speed estimation method includes a vehicle signal preprocessing method and a vehicle speed estimation method.

[0177] (1) The signal preprocessing method provides available wheel speeds and rankings for vehicle speed estimation. It includes:

[0178] 1) Sort the four wheel speeds, output the number of unavailable wheel speeds, and output the sorted wheel speed values ​​and corresponding position numbers.

[0179] 2) Calculate the front and rear axle speeds. When the wheel speeds are normal, use the average of the coaxial wheel speeds. When the wheel speeds are unreliable, use the minimum wheel speed of the vehicle instead of the axle speed.

[0180] 3) Calculate the front and rear axle accelerations, their acceleration difference, the vehicle acceleration, and the maximum acceleration, all within a ±15 m / s² range. The front axle acceleration is: aFA = 0.5 * (aFL + aFR), where aFL is the left front wheel acceleration and aFR is the right front wheel acceleration. The rear axle acceleration is: aRA = 0.5 * (aRL + aRR), where aRL is the left rear wheel acceleration and aRR is the right rear wheel acceleration. The front axle acceleration difference is: aDiff_FA = 0.5 * (aFL - aFR); the rear axle acceleration is: aDiff_RA = 0.5 * (aRL - aRR). The vehicle acceleration is: aVeh = 0.25 * (aFL + aFR + aRL + aRR). The maximum acceleration is: aMax = max(max(max(aFL, aFR), aRL), aRR).

[0181] (2) The vehicle speed estimation method is primarily based on wheel speed sensors. If a wheel slips and no reliable wheel is detected, the wheel speed is not used for calculation. The vehicle speed estimation method is divided into three cases: when the wheel speed sensor is damaged, when the vehicle is not under ABS control, and when the vehicle is under ABS control. Different methods are used to estimate the vehicle speed in each case. Figure 11 The process flow of the vehicle speed estimation according to the embodiment of the present application is as follows: Figure 11 shown.

[0182] 1) Based on the obtained wheel speed signals, calculate the wheel decelerations corresponding to the highest wheel speed vMax1, the second highest wheel speed vMax2, the lowest wheel speed vMin1, and the second lowest wheel speed vMin2, respectively.

[0183] 2) At the same time, a maximum wheel speed counter is set. If the highest wheel speed is always a certain wheel and the vehicle speed is less than this wheel speed, the timer is incremented. If the time exceeds a certain value, this wheel speed is unreliable and is not selected as the negative vehicle speed vNeg.

[0184] 3) When two or more wheel speed sensors are damaged, only the highest wheel speed vMax1 and the lowest wheel speed vMin1 are used as the judgment basis, where the positive vehicle speed vPos = vMin1 and the negative vehicle speed vNeg = vMax1.

[0185] 4) When not in ABS control, the shaft speed is used as the reference speed. The positive vehicle speed vPos is estimated using the lowest shaft speed to prevent drive shaft acceleration slip. The negative vehicle speed vNeg is estimated using the highest shaft speed to prevent underestimation of vehicle speed due to slip at the rear wheels during deceleration, such as when Electronic Brake-force Distribution (EBD) occurs. Engine drag is also taken into account. When engine drag occurs, the positive vehicle speed vPos is estimated using the second highest wheel speed vMax2. When the lowest wheel speed is less than 0.5 m / s, the positive vehicle speed vPos is estimated using the second lowest wheel speed vMin2.

[0186] The negative vehicle speed, vNeg, is subject to upper and lower limits, which vary depending on the operating conditions. When there are >1 damaged wheel speed sensors, when the road is rough, or when braking around a corner, the upper limit is the second lowest wheel speed, vMin2, and the lower limit is the second highest wheel speed, vMax2. This is because, in certain operating conditions, such as braking around a corner, it is important to prevent the wheel speed on the inside of the curve from being too low, potentially affecting the axle speed.

[0187] 5) During ABS control, the second-highest wheel speed vMax2 is selected as the reference speed by default, but the first-highest wheel speed vMax1 is closely monitored and adopted only when it remains constant for a specified period (130ms). vMax1 is not used directly because the maximum wheel speed is not always reliable. If the vehicle is turning or a small spare tire is installed, the maximum wheel speed may be significantly higher than the actual vehicle speed.

[0188] The vehicle acceleration estimation method is based on a reference speed and a filtering method. The vehicle acceleration is the differential of a reference speed correction value. The method mainly includes reference speed selection, reference speed correction, and acceleration filtering. Figure 12 The acceleration estimation process flow diagram according to the embodiment of the present application is as follows: Figure 12 shown.

[0189] 1) The reference speed selection is as follows: when ABS is not engaged, the reference speed is the estimated vehicle speed; when ABS is engaged, the wheel speeds are unstable, and the reference speed is the fastest wheel, vMax1; when the front axle or rear axle slips, vMin2 is selected; further, when the front axle slips and vMax1 and vMax2 are the two rear wheels, the right rear wheel is selected; when the rear axle slips and vMax1 and vMax2 are the two front wheels, the right front wheel is selected; the final reference speed is whichever is greater than the vehicle speed.

[0190] 2) The vehicle reference speed correction value (Vref) is obtained by correcting the input reference speed. First, calculate the original vehicle acceleration value: aFzg_init = front-to-rear speed difference / 0.02. Compensate and follow the input reference speed value based on aFzg_init. During rapid compensation, the input value is used. During slip, the larger of the input and the calculated value is used. If the value is less than the input, follow the input according to the forward gradient limit. If the reference input is unreliable, use aFzg as the gradient following input.

[0191] 3) The filtering method described above uses a second-order filter to filter acceleration, with the filter coefficient determined by whether ABS is activated. The filtered acceleration is calculated as follows: aFzg = a1 * aFzg + b0 * aFzg_raw, where aFzg is the final output vehicle acceleration and aFzg_raw is the newly calculated vehicle acceleration. aFzg follows the value of aFzg_raw. A larger b0 indicates faster tracking, and vice versa. The filter coefficient b0 is selected based on different operating conditions, as shown in Table 1 below:

[0192] Table 1 Filter coefficient b0 under different working conditions

[0193] Serial number Working conditions b0 value 1 Default gradient 0.01 2 Low Attachment 0.005 3 Default gradient used when ABS is not activated 0.03 4 Fast Follow 0.06 5 Underbraking 0.02 6 When the ABS is not activated, the vehicle is accelerating 0.05 7 If the acceleration is underestimated when ABS is not activated, a quick filter may be used. 0.20 8 When ABS is not activated, the acceleration is less than -8m / s^2 0.12 9 Low attachment pavement 0.12

[0194] The vehicle state recognition method mainly includes road impact recognition, bad road surface and strong washboard road surface recognition, engine drag recognition, and docking road surface recognition.

[0195] (1) The road bump recognition described above detects the four wheels separately, calculates and outputs the road bump flag based on the wheel speed, slip rate and acceleration. Figure 13 This is a flowchart of single wheel impact road bump recognition according to an embodiment of the present application. The single wheel judgment state is as follows: Figure 13 shown.

[0196] In state 0, if the wheel acceleration is less than [-25, -100] m / s2 and the wheel slip ratio is less than 0.15, the detection timer is reset and the vehicle enters state 1.

[0197] In state 1, if the wheel acceleration is less than [-50, -130] m / s2, it enters state 2; if the wheel acceleration is greater than 0, it resets to state 0.

[0198] In state 2, if the wheel acceleration is greater than 0, it means that the wheel has returned to positive acceleration, and then enters state 3.

[0199] In state 3, if the wheel acceleration is greater than [70, 100] m / s2, enter state 4.

[0200] In state 4, if the wheel speed recovers to vehicle speed + 0.7 m / s or more, the system enters state 5. During states 3 and 4, if the wheel acceleration is found to be less than -15 m / s2, the system resets to state 0.

[0201] In state 5, if the wheel acceleration is less than 0 and the timer timing is less than 100ms, enter state 6.

[0202] State 6: As long as it is in this state, the Bump flag is set.

[0203] (2) Identification of bad road surfaces and strong washboard roads. First, the insensitivity factor is calculated based on the vehicle speed table, not exceeding 2. The four wheels are calculated separately, and the wheel acceleration threshold corresponding to the bad road surface is calculated based on whether the ABS is working, and combined with the insensitivity factor calculation. If the change in wheel acceleration is one positive and one negative, and the ABS state is not in high or low adhesion, the factor is automatically increased by 1. Based on the value of the factor, the following judgment is made: if the factor is greater than 4, it may be a bad road surface; if the factor is greater than 6, the bad road flag is set.

[0204] The front wheel is used to determine whether the road surface is a strong washboard road. The acceleration threshold is calculated based on the vehicle speed table. Monitoring is turned on when the acceleration is less than 0. When the detection is turned on, the wheel speed is greater than the vehicle speed - am / s and less than the vehicle speed + bm / s, where a and b are empirical values, the timer is set to 0.12s and the counter is accumulated. When the timer is less than 0 and the counter is greater than 0, it is decremented. If the single wheel counter is greater than 7 or both wheels are greater than 4, the flag is set.

[0205] (3) The engine drag identification and engine drag flag are output. Detect whether the engine is dragging when the ABS is activated: 1) The current axle control mode is ABS; 2) The other axle control mode is not ABS; 3) The front wheel slip ratio exceeds 0.2; 4) The front wheel unstable time count exceeds 0.2s; 5) No brake pedal state (ABS activation request) is detected

[0206] (4) The aforementioned road surface identification monitors whether the road adhesion coefficient jumps from a high-adhesion road surface to a low-adhesion road surface. Its identification conditions are: 1) the front axle is in ABS mode; 2) the vehicle speed is greater than 10m / s; 3) the slip ratio of both front wheels is greater than 0.5; 4) the sum of the instability time of both front wheels is greater than 160ms.

[0207] The reference vehicle speed for slip ratio calculation is the correction value Vref of the reference vehicle speed. Slip ratio = 1.2*wheel speed difference / (Vref+1.2), where the maximum limit of the denominator is 7.

[0208] The master cylinder pressure estimation method uses an estimated master cylinder pressure as the master cylinder pressure when a master cylinder pressure signal sensor is not installed. The estimated master cylinder pressure signal is calculated based on vehicle speed, acceleration, slip ratio, and road conditions. This estimated master cylinder pressure signal is then sent to the wheel control and axle control state machines for vehicle state control. The calculated signal is also sent to the hydraulic control module to support wheel cylinder pressure calculation.

[0209] When the driver applies the brake pedal, the pedal pushes the brake fluid from the reservoir into the brake circuit, causing the master cylinder to output PHz. Once the brake fluid enters the wheel cylinder, it outputs Prad. The braking force is calculated as: Frad = Prad*CPxA, where CP is the brake efficiency factor, a calibrated quantity related to the cylinder diameter D, the effective braking radius R, and the friction coefficient μ of the friction pad. The braking force, combined with frictional resistance from the ground, decelerates the vehicle. Based on Newton's first law, F = ma, PHz can be estimated from the vehicle deceleration aVeh and the wheel deceleration awhl.

[0210] The ABS control method determines whether the ABS function is turned on or off.

[0211] (1) The ABS entry / exit speed threshold calculation method is used to calculate the ABS entry and exit speed thresholds. Generally, the entry speed threshold is 1.0 m / s; the exit speed threshold is 0.5 m / s. At the same time, according to different working conditions, for example, the speed thresholds for the entry and exit of the front and rear axles are determined based on the off-road sign position, and the entry speed threshold is 1.1 m / s; if encountering a hydroplaning road or emergency braking, the axis control input pressure is kept unchanged, and the axis control command is set to immediate shutdown; if the vehicle speed is too low or the driver has no braking intention, the axis control command ABS part is set to preemptive shutdown; if the ABS has not yet reached the start condition, the axis control command ABS part is set to coordinated shutdown; if EBD is not available and one wheel on the current axle is in ABS mode, the EBD mode is preemptive shutdown.

[0212] The hydraulic control method mainly includes: a wheel cylinder pressure control method, a pressure reducing valve AV control method, a pressure increasing valve EV control method, and a wheel cylinder pressure estimation method.

[0213] (1) The wheel cylinder pressure control method described above determines whether to perform a boost or depressurization action based on the wheel cylinder hydraulic pressure control target and the actual wheel cylinder model hydraulic pressure, thereby determining whether the boost valve EV and the depressurization valve AV are operating. By default, EV is normally open and AV is normally closed, meaning that boost is allowed but depressurization is not. When the target pressure is greater than the actual pressure, boosting is required; when the target pressure is less than the actual pressure, depressurization is required.

[0214] (2) The pressure reducing valve AV control method described above calculates the driving time based on the actual model pressure and the target wheel cylinder pressure, and estimates the wheel cylinder pressure change through the AV in the current cycle based on the actual driving time. Since the AV is an on-off control valve, the actual AV driving time needs to be calculated based on the target pressure and the actual model pressure.

[0215] According to the physical characteristics of the switch valve, the control time calculation formula is: TAV = (pModel – pAcc) / pGrad + TResponse. Where ΔPAV = pModel – pAcc, pAcc is the accumulator fluid pressure, and pGrad is the AV decompression rate of the current cycle. This is calculated based on ΔPAV using the Bernoulli equation: pGrad = cWheel * qAV, where qAV = Cd*A* (2 / ρ*ΔPAV )1 / 2, where cWheel is the wheel cylinder stiffness, Cd is the flow coefficient, A is the AV valve orifice area, and ρ is the brake fluid density.

[0216] In addition, the response time Tresponse consists of two parts: the opening response time and the closing response time:

[0217] TResponse=TOpenResponse - TCloseResponse

[0218] The opening response time TOpenResponse is related to the current wheel cylinder pressure and AV drive voltage:

[0219] TOpenResponse=copen + kOpenU*Uvr + kOpenP*pModle + kOpenUP*Uvr*PModel

[0220] Where Uvr is the AV driving voltage, and kOpenU, kOpenP, and kOpenUP are coefficient constants related to the driving voltage.

[0221] The closing response time TCloseResponse is only related to the wheel cylinder pressure at the end of driving. Here, the target wheel cylinder pressure pTgt is used to predict the closing response time: TCloseResponse = cClose + kCloseP*pTgt, where cClose and kCloseP are constants and calibrated values.

[0222] At the same time, the decompression amount pDecAV and the average flow rate qAvMean are calculated based on the actual AV control time and actual decompression rate of the current cycle:

[0223] pDecAV = pGrad*min(0.005,TAV)

[0224] qAvMean = qAV*min(0.005,TAV ) / 0.005

[0225] (3) In the boost valve EV control method, the EV calculates the target flow through the EV based on the target wheel cylinder hydraulic pressure and the actual wheel cylinder model pressure, and then calculates the EV control current based on the actual pressure difference of the EV and the flow characteristics of the EV, thereby achieving the purpose of valve control. The calculation of the control current of the EV is different depending on the control mode of the EV. For the open mode, the EV maintains the default normally open state and the control current is always 0; for the closed mode, the EV remains closed and its control current is a maximum of 1.5A. When the valve control strategy is pressure correction control, the EV needs to achieve linear control of the wheel cylinder pressure. To this end, the target pressure needs to be converted into a target flow: qTgt = k / 0.005*(pTgt - pModel) / cWheel, where the coefficient k is used to adjust the EV response speed, generally k=2, representing the typical response time of the EV. According to the physical characteristics of the solenoid valve, when the actual pressure difference ΔPEV and the target flow qTgt are given, the control current of the solenoid valve can be determined. Therefore, the control current of the EV is calculated by lookup table based on the above information: IEV = LookupTable(qTgt, ΔPEV).

[0226] Based on the estimated current value IEV and the actual EV pressure difference, the current EV flow rate can be estimated: qEV = LookupTable(IEV, ΔPEV).

[0227] The pressure change during the current EV control cycle can be estimated by combining the wheel cylinder PV characteristics: pIncEV = q_Ev*cWheel*0.005.

[0228] (4) The wheel cylinder pressure estimation method described above estimates the wheel cylinder model pressure in combination with the previous AV and EV pressure calculations, that is, pDecAV and pIncEV are accumulated in each cycle. The initial value of the wheel cylinder pressure when the system is just powered on is the master cylinder pressure sampling value. The wheel cylinder pressure estimation formula is: pModel = pModel_old + pIncEV - pDecAV.

[0229] The present application also provides a computer program product, which includes instructions. When the instructions are executed, the device performs operations corresponding to the above method.

[0230] In one embodiment of the present application, a computer-readable storage medium is further provided, wherein the computer-readable storage medium stores a computer program for executing the above-mentioned method.

[0231] It should be understood that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, they can be implemented in the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware. Furthermore, they can be implemented in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0232] The methods, systems, and computer program products of the embodiments of the present application are described by flowcharts and / or block diagrams. It should be understood that each process and / or box in the flowcharts and / or block diagrams, as well as the combination of processes and / or boxes in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0233] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0234] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0235] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0236] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0237] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0238] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0239] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0240] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. An anti-lock braking system for a vehicle, characterized in that: include: Signal preprocessing module, vehicle speed signal estimation module, wheel signal and slip ratio calculation module, vehicle state estimation module, master cylinder pressure estimation module, failure processing module, vehicle control module, front and rear axle control module, master cylinder pressure correction module and hydraulic control module; The signal preprocessing module is used to preprocess the input signal and output the preprocessing result; The vehicle speed signal estimation module is used to estimate the vehicle speed and acceleration; The wheel signal and slip ratio calculation module is used to calculate the slip ratio based on the vehicle speed and acceleration; The vehicle state estimation module is used to determine the state of the vehicle during driving based on the speed and acceleration of the vehicle; the master cylinder pressure estimation module, configured to calculate an estimated value of a master cylinder pressure signal based on the vehicle speed and acceleration, the slip ratio, and a road surface condition; The failure processing module is used to process a failed sensor; The vehicle control module is used to determine whether to enable the anti-lock braking function; The front and rear axle control modules are configured to calculate target pressures for axle control and wheel control algorithms based on the vehicle speed and body acceleration; The master cylinder pressure correction module is used to calculate a correction value of the master cylinder pressure based on the vehicle body deceleration and the wheel deceleration; as well as The hydraulic control module is used to perform wheel cylinder hydraulic pressure target control; The vehicle state estimation module includes: A vehicle status data acquisition unit, configured to acquire the speed and acceleration of the vehicle at a plurality of predetermined time points within a predetermined time period; a parameter time series arrangement unit, configured to arrange the vehicle speeds and accelerations of the vehicle at the plurality of predetermined time points into a speed input vector and an acceleration input vector according to a time dimension; a speed change feature extraction unit, configured to pass the speed input vector through a first multi-scale neighborhood feature extraction module to obtain a speed time series feature vector; an acceleration change feature extraction unit, configured to pass the acceleration input vector through a second multi-scale neighborhood feature extraction module to obtain an acceleration time series feature vector; a feature fusion unit, configured to fuse the velocity time series feature vector and the acceleration time series feature vector to obtain a vehicle state feature vector; a feature optimization unit, configured to perform feature distribution optimization on the vehicle state feature vector based on the velocity time series feature vector and the acceleration time series feature vector to obtain an optimized vehicle state feature vector; and The vehicle state detection unit is used to pass the optimized vehicle state feature vector through a multi-label classifier to obtain a classification result, and the classification result is used to represent a label of the road state.

2. The anti-lock braking system for a vehicle according to claim 1, characterized in that The first multi-scale neighborhood feature extraction module includes a first convolutional layer and a second convolutional layer in parallel, and a first cascade layer connected to the first convolutional layer and the second convolutional layer, wherein the first convolutional layer uses a one-dimensional convolution kernel with a first scale, and the second convolutional layer uses a one-dimensional convolution kernel with a second scale, and the first scale is different from the second scale; The second multi-scale neighborhood feature extraction module includes a third convolution layer and a fourth convolution layer in parallel, and a second cascade layer connected to the third convolution layer and the fourth convolution layer, wherein the third convolution layer uses a one-dimensional convolution kernel with a third scale, and the fourth convolution layer uses a one-dimensional convolution layer with a fourth scale, and the third scale is different from the fourth scale.

3. The anti-lock braking system for a vehicle according to claim 2, characterized in that: The speed change feature extraction unit includes: A first scale feature extraction subunit is configured to use the first convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the velocity input vector using the following formula to obtain the first scale velocity feature vector; Wherein, the formula is: Where a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(xa) is the local vector matrix operated with the convolution kernel function, w is the size of the first convolution kernel, and X represents the velocity input vector; A second scale feature extraction subunit is configured to use the second convolution layer of the first multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the velocity input vector using the following formula to obtain a second-scale velocity feature vector; Wherein, the formula is: Wherein, b is the width of the second convolution kernel in the x direction, F(b) is the second convolution kernel parameter vector, G(xb) is the local vector matrix operated with the convolution kernel function, m is the size of the second convolution kernel, and X represents the velocity input vector; and The first cascading subunit is configured to cascade the first-scale velocity feature vector and the second-scale velocity feature vector to obtain the velocity time series feature vector.

4. The anti-lock braking system for a vehicle according to claim 3, characterized in that: The acceleration change feature extraction unit includes: a third-scale feature extraction subunit, configured to use the third convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the third-scale acceleration feature vector; Wherein, the formula is: Where c is the width of the third convolution kernel in the x direction, F(c) is the third convolution kernel parameter vector, G(xc) is the local vector matrix operated with the convolution kernel function, n is the size of the third convolution kernel, and Y represents the acceleration input vector; a fourth-scale feature extraction subunit, configured to use the fourth convolution layer of the second multi-scale neighborhood feature extraction module to perform one-dimensional convolution encoding on the acceleration input vector using the following formula to obtain the fourth-scale acceleration feature vector; Wherein, the formula is: Wherein, d is the width of the fourth convolution kernel in the x direction, F(d) is the fourth convolution kernel parameter vector, G(xd) is the local vector matrix operated with the convolution kernel function, m is the size of the fourth convolution kernel, and Y represents the acceleration input vector; and The second cascading subunit is configured to cascade the third-scale acceleration feature vector and the fourth-scale acceleration feature vector to obtain the acceleration time series feature vector.

5. The anti-lock braking system for a vehicle according to claim 4, characterized in that: The feature fusion unit is configured to fuse the velocity time series feature vector and the acceleration time series feature vector using the following formula to obtain a vehicle state feature vector; Wherein, the formula is: Wherein, V is the vehicle state characteristic vector, V1 is the velocity time series characteristic vector, V2 is the acceleration time series characteristic vector, represents position addition, and λ and β are weighting parameters for controlling the balance between the velocity time series feature vector and the acceleration time series feature vector.

6. The anti-lock braking system for a vehicle according to claim 5, characterized in that: The feature optimization unit includes: A first optimization subunit is configured to calculate an incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector to obtain a first optimized vehicle state feature vector; a second optimization subunit, configured to calculate an incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector to obtain a second optimized vehicle state feature vector; and The optimized feature fusion subunit is used to fuse the first optimized vehicle state feature vector and the second optimized vehicle state feature vector to obtain the optimized vehicle state feature vector.

7. The anti-lock braking system for a vehicle according to claim 6, characterized in that: The first optimization subunit is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the speed time series feature vector using the following formula to obtain a first optimized vehicle state feature vector; Wherein, the formula is: Among them, V1, V3 and V3 ' denote the speed time series feature vector, the vehicle state feature vector, and the first optimized vehicle state feature vector, respectively; ||·||1 and ||·||2 denote the first norm and the second norm of the vector, respectively; L denotes the length of the vector; and ⊙ represent vector product and vector dot product respectively, and all vectors are in row vector form, (·) T is the transpose of a vector.

8. The anti-lock braking system for a vehicle according to claim 7, characterized in that: The second optimization subunit is further configured to calculate the incoherent sparse response fusion feature of the vehicle state feature vector and the acceleration time series feature vector using the following formula to obtain a second optimized vehicle state feature vector; Wherein, the formula is: Wherein, V2, V3 and V3″ respectively represent the acceleration time series feature vector, the vehicle state feature vector and the second optimized vehicle state feature vector, ||·||1 and ||·||2 respectively represent the first norm and the second norm of the vector, L is the length of the vector, and ⊙ represent vector product and vector dot product respectively, and all vectors are in row vector form, (·) T is the transpose of a vector.

9. The anti-lock braking system for a vehicle according to claim 8, characterized in that: The vehicle state detection unit includes: a fully connected encoding subunit, configured to perform fully connected encoding on the optimized vehicle state feature vector using multiple fully connected layers of the classifier to obtain an encoded classification feature vector; and The classification subunit is used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

Images