Active safety coordination control system and method for intelligent vehicles under sudden emergency conditions

Abstract

The present invention belongs to the technical field of intelligent vehicles, and particularly relates to an active safety coordination control system and method for intelligent vehicles under sudden emergency conditions. The system includes a sensor, a signal processing module, a vertical force estimation module, an evaluation index acquisition module, a weight factor calculation module, and a vehicle control module. Among them, the vertical force estimation module is implemented based on a radial neural network and is used to estimate the vertical load according to the output of the signal processing module. The evaluation index acquisition module is used to calculate the yaw stability factor and the rollover stability factor based on the vertical load. The weight factor calculation module uses a fuzzy algorithm to perform fuzzy processing on the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then performs defuzzification calculation on the fuzzy rules to control the weight factor q. The control module is used to achieve safe coordination control of the vehicle according to the estimated vertical load and the weight control factor q.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent vehicles, and particularly relates to an active safety coordination control system and method for intelligent vehicles under sudden emergency conditions. Background Art

[0002] In recent years, the rapid development of intelligent auxiliary driving systems has achieved important results in ensuring the safety and comfort of vehicle driving, such as anti-lock braking systems (ABS), electronic stability control systems (ESC), adaptive cruise control systems (ACC), etc. Driving safety has always been a key concern for people. However, even in the current era of rapid technological development, some instability and collision accidents often occur. A flat tire is a common phenomenon in traffic accidents. When a vehicle has a flat tire, the mechanical properties of the vehicle tires change, which in turn causes the steering performance of the vehicle to change. Generally, drivers cannot control the vehicle to return to normal by controlling the steering wheel or braking. Therefore, designing a vehicle stability controller after a flat tire is of great significance for ensuring the safety of people and property under extreme conditions such as flat tires. The change of vehicle tire mechanical characteristic parameters and vehicle vertical load transfer are important characteristics of tires under extreme conditions such as flat tires, and are also the key points in studying vehicle stability control under extreme conditions.

[0003] The intelligent tire vertical load estimation method under flat tire conditions mainly studies the law of vertical load transfer after a vehicle has a flat tire, analyzes the main influencing factors of vehicle load transfer after a flat tire, and explores the vertical load transfer mechanism after a flat tire based on a data-theory combination method to provide a reliable vertical load change value for vehicle stability control after a flat tire. The intelligent tire vertical load estimation method under flat tire conditions mainly includes exploring the influencing factors of vertical load transfer, studying the estimation algorithm of vertical load transfer after a flat tire based on data-theory driving, and then combining with the vehicle stability controller after a flat tire to obtain a control framework for ensuring vehicle safety under flat tire conditions.

[0004] The current methods for obtaining the vertical load after a vehicle has a flat tire are mainly data-based and theory-based methods. The data-based method uses various sensors on the tire to directly obtain vertical load data through sensor signals, and this method has a high cost; the theory-based method is to calculate and summarize a set of vertical load calculation formulas through mechanical analysis. For flat tire stability control, this method has certain limitations in the application scope. Summary of the Invention

[0005] In view of this, the present invention provides an active safety coordination control system and method for intelligent vehicles under sudden emergency conditions, which can accurately estimate the vertical load and weight factor under sudden conditions, thereby ensuring the stable control of the vehicle under sudden conditions.

[0006] The technical solution of the present invention is realized as follows:

[0007] In a first aspect, the present invention provides an active safety coordination control system for intelligent vehicles under sudden emergency conditions, including a sensor, a signal processing module, a vertical force estimation module, an evaluation index acquisition module, a weight factor calculation module, and a vehicle control module; wherein,

[0008] The sensor is used to collect relevant parameters of the vehicle under sudden conditions, and the parameters include longitudinal speed, lateral speed, yaw angular velocity, and effective tire rolling radius;

[0009] The signal processing module is used to preprocess the relevant parameters collected by the sensor and remove the noise signals in the relevant parameters;

[0010] The vertical force estimation module is implemented based on a radial neural network and is used to estimate the vertical load according to the output of the signal processing module;

[0011] The evaluation index acquisition module is used to calculate the yaw stability factor and the rollover stability factor based on the vertical load;

[0012] The weight factor calculation module uses a fuzzy algorithm to perform fuzzy processing on the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then performs defuzzification calculation on the fuzzy rules to control the weight factor q;

[0013] The control module is used to realize the safety coordination control of the vehicle according to the estimated vertical load and the weight control factor q.

[0014] Further, the yaw stability factor described in the present invention is:

[0015]

[0016] wherein, is the stable boundary coefficient of the phase plane, are the sideslip angle of the center of mass and the sideslip angular velocity of the center of mass.

[0017] Further, the rollover stability factor described in the present invention is:

[0018]

[0019] wherein, are the vertical loads on the left and right wheels of the vehicle respectively, are the positions of the axles and the number of axles respectively.

[0020] Further, in the weight factor calculation module of the present invention, the fuzzy sets of the yaw and roll risk degrees are first defined as {low, medium - low, medium, medium - high, high}, that is, {SA, LR, MR, MHR, HR}. The output fuzzy set represents the comparison degree between the two, and the resulting fuzzy set is obtained, that is, {H1, H2, H3, H4, H5}, representing the relative proportion of the roll risk to the yaw, representing small, relatively small, medium, relatively large, and large respectively, and a fuzzy rule table is formulated; the control weight factor q is set within and defuzzification is achieved through the area centroid method.

[0021] Further, the fuzzy rule table of the present invention is as follows:

[0022]

[0023] Further, the vertical force estimation module of the present invention is obtained by the following process:

[0024] Neural network structure design: The radial basis function neural network architecture is adopted, which includes three layers: the input layer, the hidden layer, and the output layer;

[0025] Neural network training: Based on the Doguff tire model and the CarSim model, a vehicle simulation model under a flat - tire condition is established. Among them, the Doguff tire model is used to simulate the change of tire stiffness, and the CarSim model is used to apply a sudden vertical force to simulate the change of the effective rolling radius of the tire; the vehicle simulation model is used to obtain the sample data for training, and the neural network is trained using the sample data.

[0026] Further, the acquisition of the vertical force estimation module of the present invention also includes an evaluation link of the neural network. The effectiveness and reliability of the obtained radial basis neural network estimation model are verified using the mean square error and generalization error.

[0027] Further, the relevant parameters of the present invention include the vehicle longitudinal acceleration , the vehicle lateral acceleration , the vehicle roll angle , the roll - angle velocity , the effective rolling radius of the tire , the longitudinal velocity , the lateral velocity and the yaw rate .

[0028] Further, the signal processing module of the present invention performs feature extraction, Kalman filtering, and moving average on the signals collected by the sensors, and eliminates the noise signals in the relevant parameters.

[0029] Second aspect, a vehicle control method based on the estimation of vertical load and weight factor under sudden working conditions according to the present invention has the following specific process:

[0030] Collect relevant parameters of the vehicle under sudden working conditions by using sensors, where the relevant parameters include longitudinal speed, lateral speed, yaw angular velocity, and effective tire rolling radius;

[0031] Preprocess the relevant parameters collected by the sensors by using a signal processing module to remove the noise signals in the relevant parameters;

[0032] Based on a vertical force estimation module, estimate the vertical load according to the data output by the signal processing module;

[0033] Calculate the yaw stability factor and rollover stability factor based on the vertical load evaluation index acquisition module;

[0034] Use a fuzzy algorithm, and the weight factor calculation module performs fuzzy processing on the yaw stability factor and rollover stability factor to obtain fuzzy rules, and then performs defuzzification calculation on the fuzzy rules to control the weight factor q;

[0035] According to the estimated vertical load and the weight control factor q, the control module realizes the safety coordinated control of the vehicle.

[0036] Since the present invention adopts the above technical solutions, it has the following advantages:

[0037] First, the present invention uses a radial basis neural network to estimate the vertical force, obtains a radial basis neural network estimation model, and cleverly combines the data-based and theory-based vertical force estimation methods to expand the adaptability range of vehicle vertical force estimation under flat tire conditions.

[0038] Second, a vehicle flat tire model is formed by Carsim to train the proposed estimation method, and two evaluation indexes are proposed: mean square error and generalization error. Then, the radial basis function estimation model verification based on test data is proposed. Further improve the effectiveness and reliability of the estimation model.

[0039] Third, a method for designing the control weights of vehicle roll and yaw after flat tire based on fuzzy algorithm is proposed. The yaw stability factor is obtained by using the centroid side slip angle and the centroid side slip angular velocity phase plane, and the rollover stability factor is obtained by using the lateral load transfer ratio. Design fuzzy rules to compare the ratio between the two, and further defuzzify to obtain the weight factor, which is applied to the controller, improving the adaptability and effectiveness of the vehicle driving stability after flat tire. Description of the Drawings

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0041] Figure 1 Schematic diagram of the vehicle control system of the present invention;

[0042] Figure 2 Schematic diagram of the vertical force analysis of the tire of the present invention;

[0043] Figure 3 Graph of the change trend of tire parameters after a vehicle tire blowout;

[0044] Figure 4 Neural network architecture diagram. Detailed implementation manners

[0045] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0046] It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other; and, based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present disclosure.

[0047] It should be noted that the following describes various aspects of the embodiments within the scope of the appended claims. It should be obvious that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is illustrative only. Based on the present disclosure, those skilled in the art should understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement the device and / or practice the method. In addition, this device and / or practice this method can be implemented using other structures and / or functions in addition to one or more of the aspects described herein.

[0048] The design basis of the present invention is: analyzing the change law of tire parameters after a tire blowout, exploring the dynamic and kinematic response of the vehicle after a tire blowout, studying the vertical load reconstruction model of the vehicle after a tire blowout, and providing a reliable vertical load basis for the controller design. Further explore the coordinated control of vehicle roll and yaw stability, and propose a control weight distribution method based on the degree of risk.

[0049] The proposed coordinated control framework for vehicle roll and yaw under tire blowout conditions in the present invention can be mainly divided into two parts, namely the vertical load reconstruction method after tire blowout and the coordinated control method for driving stability. Through the "coordinated control method for vehicle roll and yaw under tire blowout conditions", the roll and yaw stabilities of the vehicle after tire blowout are improved simultaneously to achieve active safety control under sudden emergency conditions and enhance the safety performance of the vehicle. The "coordinated control method for vehicle roll and yaw under tire blowout conditions" can be "vertical load reconstruction" and "control weight design", as Figure 1 shown.

[0050] "Vertical load reconstruction" is to explore the vertical load transfer of the vehicle after tire blowout. By analyzing the influencing factors of vertical load transfer and the parameter changes of the vehicle after tire blowout, the state variables and change parameters affecting the vertical load change are obtained, and the tire vertical force is estimated based on a radial neural network. The traditional methods are mainly data-driven and theory-driven. The data-driven method directly obtains the data of the vertical force. This method has high requirements for sensors, and for sudden tire blowout situations, it has extremely high requirements for the response speed of sensors. Therefore, this method is rarely used. The theoretical vertical load estimation method summarizes a set of formulas for wheel load transfer after tire blowout through mechanical analysis, which is difficult to apply to all vehicles. For example, the formula for multi-axle vehicles will be extremely complex. The present invention proposes a data-theory-based method for estimating the vertical load of a vehicle after tire blowout, which uses a radial neural network to estimate the tire vertical force. It can be divided into "sensor signal acquisition" and "vertical load estimation".

[0051] "Control weight design" mainly studies the coordinated control method for vehicle roll and yaw stabilities after tire blowout, focusing on the analysis of the coupling relationship and coordination method between the two. The weight factors in existing studies are mostly adjusted based on experience and it is difficult to achieve self-adaptability. When the vehicle motion states are different, the required control objectives also have differences. Especially for strongly coupled systems, the research on the coordination method is the key to realizing the active safety control of a tire-blowout vehicle. The present invention proposes a control weight calculation method based on risk response, explores the risk evaluation indexes for vehicle roll and yaw after tire blowout, that is, the "roll and yaw stability evaluation indexes", and designs a fuzzy algorithm to obtain the control weights of the stability controller, that is, the "control weight design method".

[0052] Based on the above analysis, an active safety coordination control system for intelligent vehicles under sudden emergency conditions in an embodiment of the present application includes: sensors, a signal processing module, a vertical force estimation module, an evaluation index acquisition module, a control weight factor calculation module, and a vehicle control module; wherein,

[0053] The sensors are used to collect relevant parameters of the vehicle under sudden conditions, and the parameters include longitudinal speed, lateral speed, yaw angular velocity, and effective tire rolling radius;

[0054] The signal processing module is used to pre-process the relevant parameters collected by the sensor and eliminate the noise signals in the relevant parameters;

[0055] The vertical force estimation module is implemented based on a radial neural network and is used to estimate the vertical load according to the output of the signal processing module;

[0056] an evaluation index acquisition module, configured to calculate a yaw stability factor and a rollover stability factor based on the vertical load;

[0057] A control weight factor calculation module uses a fuzzy algorithm to fuzzify the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then defuzzifies the fuzzy rules to calculate the control weight factor q;

[0058] The control module is used to achieve safe coordinated control of the vehicle based on the estimated vertical load and weight control factor q.

[0059] Furthermore, in order to explore the factors that affect the vertical load transfer of a vehicle after a tire blowout, this embodiment arranges sensors in a targeted manner to effectively improve the utilization rate of the sensors. By analyzing the factors that affect the vertical load transfer and the changes in vehicle parameters after a tire blowout, the state quantities and change parameters that affect the vertical load change are obtained, such as Figure 3 As shown, is the tire radial stiffness, is the effective rolling radius of the tire, is the longitudinal stiffness, is the lateral stiffness, For this purpose, this embodiment summarizes a set of variation quantities applicable to the vertical load evaluation in the present invention with respect to conventional load transfer and tire blowout condition parameter variations.

[0060] First, the vertical force analysis of conventional vehicles is as follows: Figure 2 As shown, the vertical force formula of the tire can be expressed as:

[0061] (1)

[0062] in, is the vehicle mass, and are the distances from the front and rear axles to the center of mass, is the acceleration due to gravity, is the wheelbase of the front and rear axles, are the vehicle longitudinal and lateral accelerations, respectively, is the height of the vehicle's center of mass from the ground, is the wheelbase, are the roll angle, roll velocity, is the roll stiffness, is the roll damping coefficient.

[0063] From this formula, it can be concluded that the state variables related to the vertical load transfer are: vehicle longitudinal acceleration , vehicle lateral acceleration , vehicle roll angle and roll angular velocity . In addition, when a tire blowout occurs in the vehicle, the mechanical characteristics of the wheel change rapidly, mainly reflected in the effective rolling radius of the tire, longitudinal stiffness, cornering stiffness, rolling resistance coefficient, and tire radial stiffness, etc. The longitudinal stiffness, cornering stiffness, and rolling resistance coefficient do not directly cause vertical load transfer, but further affect the vertical load transfer by influencing the wheel state variables. The sudden changes in the effective rolling radius and radial stiffness of the tire will directly cause the vertical force to change rapidly. Therefore, the effective rolling radius and radial stiffness of the tire are selected as the important reasons for the influence of tire blowout on the vertical load.

[0064] In addition, although the longitudinal stiffness, cornering stiffness, and rolling resistance coefficient do not directly cause vertical load transfer, their changes lead to changes in state variables, especially longitudinal velocity, lateral velocity, and yaw angular velocity. Therefore, these three variables are taken as part of the sensor signals.

[0065] Through the above analysis, some factors affecting the change of vertical load are sorted out. However, not all the change quantities can be obtained through sensor signals. It is difficult to obtain various stiffness parameters through sensors, or the relative cost is high. Therefore, they are not considered as reference quantities for vertical force estimation.

[0066] Based on the above analysis, it is necessary to set sensors to collect the factors of the change quantities affecting the vertical load. The relevant parameters collected include: vehicle longitudinal acceleration , vehicle lateral acceleration , vehicle roll angle , roll angular velocity , effective rolling radius of the tire , longitudinal velocity , lateral velocity and yaw angular velocity .

[0067] Furthermore, when the signal processing module in the embodiment of the present application processes the relevant sensor signals affecting the change of the vertical load, the Kalman filter and the moving average method are used to denoise the sensor signals, extract key information, and improve the trustworthiness and reliability of the sensor signals. The Kalman filter is an optimal recursive filtering algorithm based on linear minimum variance estimation, and the moving average method is used to eliminate accidental change factors. When applied in signal processing, it can achieve good noise reduction and so on. The present invention uses this tool to process sensor signals, but filtering and noise reduction are not the focus of the present invention, so the present invention will not be introduced in detail in this regard.

[0068] Furthermore, in the vertical force estimation module of this embodiment, the vertical force estimation is based on the relevant sensor signals obtained by the signal processing module and is based on a radial neural network for tire vertical force estimation. In the case of a flat tire, the mechanical characteristics of the wheel change. The effective rolling radius of the flat tire decreases, and its vertical force decreases significantly and is transferred to other tires. Traditional methods are mainly data-driven and theory-driven. The data-driven method directly obtains the data of the vertical force. This method has high requirements for sensors, and for sudden flat tire situations, it has extremely high requirements for the response speed of the sensors. Therefore, this method is less used. The theory-based vertical load estimation method summarizes a set of formulas for wheel load transfer after a flat tire through mechanical analysis, and it is difficult to apply to all vehicles. For example, the formulas for multi-axle vehicles will be extremely complex. In this embodiment, a neural network is used to implement the design of the vertical force estimation module. First, the neural network structure is designed. A radial neural network is used to estimate the tire vertical force. Then, based on the designed neural network structure, the neural network needs to be trained to establish a tire vertical force estimation model. Finally, to ensure the accuracy of the trained neural network, it also includes a neural network testing link, that is, setting evaluation indicators to evaluate the neural network. The specific process is as follows:

[0069] Neural network structure design: The radial basis function neural network architecture adopted includes three layers: the input layer, the hidden layer, and the output layer.

[0070] In this step, the neural network structure design mainly studies how to build the neural network architecture, analyze the working mechanism and establishment method of the radial neural network, and study the application method of the radial neural network in the vertical force estimation of vehicles under flat tire conditions. The radial basis function neural network is a feedforward neural network with the radial basis function as the activation function. It has the advantages of simple structure and fast learning speed, and theoretically, the radial basis function neural network can fit any continuous function. Therefore, it has high feasibility to apply it to the vertical force estimation of vehicles after a flat tire.

[0071] Such as Figure 4As shown, the radial basis function neural network architecture consists of three layers: the input layer, the hidden layer, and the output layer. It uses the radial basis activation function to cleverly map the input variables to the hidden layer. This process essentially realizes the transformation of the input vector from a low-dimensional and non-linearly inseparable space to a high-dimensional and linearly separable space. Subsequently, the data in the hidden layer will be processed by linear weighting and further mapped to the output layer. Through such a mechanism, the RBF neural network not only successfully constructs a non-linear mapping relationship from input to output but also ensures that the parameters in the network output have the characteristics of linear adjustability, thus greatly enhancing its ability to handle complex problems.

[0072] The activation function of the radial basis neural network can be expressed as:

[0073] (2)

[0074] Where, is the neural network input, is the center of the network hidden layer node, is the width parameter.

[0075] The output function of the radial basis neural network is:

[0076] (3)

[0077] Where, is the number of hidden layer nodes, is the th hidden layer neuron to the th output layer neuron weight, is the number of output layer neurons, is the actual output of the th output node of the network corresponding to the input sample.

[0078] Neural network training: Based on the Doguff tire model and the CarSim model, a vehicle simulation model under a flat tire condition is established. Among them, the Doguff tire model is used to reflect the change in tire stiffness, and the CarSim model is used to apply a sudden vertical force to simulate the change in the effective rolling radius of the tire; the vehicle simulation model is used to obtain the sample data for training, and the neural network is trained using the sample data.

[0079] In this step, the neural network training meets the training and learning requirements of the tire vertical force estimation model. A large amount of data on the vertical force changes caused by the changes in vehicle parameters and state variables after a tire blowout needs to be collected. However, it is relatively difficult to obtain data through tire blowout tests, and the cost is extremely high. Therefore, it is impossible to use the test method to obtain training data. This embodiment provides a method based on the CarSim high-fidelity model, designs a tire blowout condition, and obtains training data from it. This data is relatively simple to obtain and has no impact on the neural network training.

[0080] Build a vehicle tire blowout model under the tire blowout condition, and establish a vehicle simulation model under the tire blowout condition by combining the Doguff tire model and the CarSim model. The formula of the Doguff tire model is:

[0081] (4)

[0082] Where, is the lateral cornering stiffness of the tire, is the longitudinal stiffness of the tire, is the longitudinal slip ratio, is the tire side slip angle, is the road surface adhesion coefficient, is the vertical load. is the lateral force; is the correction coefficient; is the intermediate variable.

[0083] There are two important parts in building the tire blowout condition. One part is the change in tire mechanical characteristic parameters such as stiffness, which is reflected in the tire model, and the tire changes after a vehicle tire blowout are reflected through the tire model. The other part is the problem of vertical load change. By applying a sudden vertical force to CarSim, the situation where the effective rolling radius of the tire decreases after a tire blowout can be better simulated.

[0084] The mean squared error (MSE) is used as the loss function of the radial basis neural network to evaluate the quality of the neural network training. Its expression is:

[0085] (5)

[0086] Where, is the number of training samples, is the network estimation, is the true value.

[0087] After debugging and optimizing the structure parameter of the radial basis network - the diffusion coefficient spread for many times, the mean square error values of each estimation network model are minimized, thus achieving the best performance of these models on the training set. To evaluate the generalization performance of the radial basis neural network estimation model, the test error on the test set is used as an approximation of the generalization error, and the generalization error is used as an evaluation index for the generalization performance.

[0088] Neural network evaluation: Based on the mean square error and the generalization error, the effectiveness and reliability of the obtained radial basis neural network estimation model are verified.

[0089] The above test is based on the training results of the theoretical model. Whether the final model can meet the actual application requirements still requires further analysis of the results. Therefore, it is necessary to further verify this model, that is, to verify the effectiveness and reliability of the obtained radial basis neural network estimation model through experimental data. The mean square error and the generalization error of the estimation network model are also used as evaluation indexes. If these indexes are small, the obtained radial basis neural network estimation model is considered effective.

[0090] Furthermore, the evaluation index calculation module of this embodiment calculates the yaw stability factor and the rollover stability factor; specifically:

[0091] The "evaluation index of roll and yaw stability" mainly studies the parameters characterizing roll and yaw stability, and analyzes the yaw instability risk of the vehicle after a tire blowout. Currently, there are many methods for judging whether the vehicle state is unstable. Generally, the yaw rate and the sideslip angle of the center of mass are used as the best state variables for judging vehicle yaw instability. The phase plane method is a commonly used means for analyzing yaw stability. By drawing the phase plane and dividing the stable region, the degree of vehicle instability can be further obtained. The sideslip angle of the center of mass - the angular velocity of the sideslip of the center of mass is selected as the evaluation index for the yaw risk.

[0092] (6)

[0093] Among them, is the yaw stability factor, is the stable boundary coefficient of the phase plane, are the sideslip angle of the center of mass and the angular velocity of the sideslip of the center of mass. In addition, for different yaw instability risks, different values are selected, and the yaw instability risk area can be divided.

[0094] In addition, for the evaluation of the vehicle rollover risk, the lateral load transfer ratio (LTR) is the most widely used at present and can effectively reflect the rollover index of the vehicle under the current dangerous working conditions.

[0095] The calculation of the lateral load transfer ratio mainly depends on estimating the vertical loads of the inner and outer wheels of the vehicle, that is, the provided vertical load data.

[0096] (7)

[0097] Wherein, are the vertical loads on the left and right wheels of the vehicle respectively, are the position of the axle and the number of axles respectively. When the vehicle is on a good road surface, LTR is 0. Therefore, the absolute value of LTR is between When LTR is 1, one side of the wheel is completely suspended, and the rollover risk reaches the maximum at this time. Therefore, the rollover stability factor can represent the rollover risk of the vehicle through LTR:

[0098] (8)

[0099] Furthermore, the embodiment of the present application proposes a control weight design control weight factor based on a fuzzy algorithm. First, define the fuzzy sets of yaw and rollover risk degree as {low, medium-low, medium, medium-high, high}, that is, {SA, LR, MR, MHR, HR}. The output fuzzy set represents the comparison degree between the two. When the roll stability factor is small, yaw dominates at this time. Therefore, the yaw stability controller requires a higher weight. By analogy, the result fuzzy set is obtained, that is, {H1, H2, H3, H4, H5}, which represents the relative proportion of rollover risk to yaw, and represents small (H1), relatively small (H2), medium (H3), relatively large (H4), and large (H5) respectively. Therefore, formulate a fuzzy rule table:

[0100]

[0101] In addition, an anti-fuzzy rule needs to be designed, that is, to obtain the control weight. The fuzzification result obtained by the fuzzy rule only shows the comparison between rollover and yaw risks. Designing an anti-fuzzy rule is the key to obtaining the control weight factor. However, it is difficult to obtain the corresponding relationship between the weight factor and the fuzzy output through theoretical analysis, because the control weight factor q is between By using the area centroid method to realize defuzzification, the core is to calculate the center of the area covered by the membership function and the abscissa, and use the calculation result as the representative value in the set.

[0102] Furthermore, the control module of the embodiment of the present application controls the vehicle's emergency conditions according to the estimated vertical load and the control factor q.

[0103] The vehicle driving stability coordination method provided by the present invention mainly aims at the limitation of the current active safety technology after a tire blowout, which is limited to yaw stability control. On the one hand, the research on the vertical load reconstruction method mainly includes sensor collection and theoretical calculation. The former has high cost and great difficulty in acquisition; the latter has great limitations and poor adaptability to multi-axle vehicles. Therefore, a data-theory-based vertical load estimation method is proposed. The present invention provides two aspects: sensor setting and vertical load estimation. Among them, through the analysis of the factors affecting the change of the vertical load of the vehicle after a tire blowout and the limitations of the sensors, a sensor setting strategy is proposed at a reasonable cost, providing a data basis for subsequent vertical load estimation. The vertical load estimation uses a radial basis neural network for estimation, designs the structure of the radial basis neural network, sets the neural network training method and framework, and proposes an evaluation index based on the radial basis neural network estimation model. On the other hand, regarding the research on the coordination method of vehicle roll and yaw stability after a tire blowout, the present invention proposes a fuzzy-based control weight design. The yaw stability factor is obtained through the centroid side slip angle and the centroid side slip angular velocity phase plane, and the roll stability factor is obtained through the lateral load transfer ratio. Further, fuzzification and defuzzification are designed to obtain the control weight design between the two, and finally applied to the coordination controller. The related control technologies are relatively mature, so the present invention does not take control as the key point of the invention.

[0104] Second, the active safety coordination control method for an intelligent vehicle under sudden emergency conditions in this embodiment is as follows:

[0105] Use sensors to collect relevant parameters of the vehicle under sudden conditions, and the relevant parameters include longitudinal speed, lateral speed, yaw angular velocity, and effective tire rolling radius;

[0106] Use the signal processing module to preprocess the relevant parameters collected by the sensors to remove the noise signals in the relevant parameters;

[0107] Based on the vertical force estimation module, realize the estimation of the vertical load according to the data output by the signal processing module;

[0108] Calculate the yaw stability factor and the rollover stability factor based on the vertical load evaluation index acquisition module;

[0109] Use the fuzzy algorithm, and the weight factor calculation module fuzzifies the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then defuzzifies the fuzzy rules to calculate the control weight factor q;

[0110] According to the estimated vertical load and the weight control factor q, the control module realizes the safety coordination control of the vehicle.

[0111] In summary, the above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An active safety coordination control system for intelligent vehicles under sudden emergency conditions, characterized in that It includes sensors, signal processing modules, vertical force estimation modules, evaluation index acquisition modules, weight factor calculation modules and vehicle control modules; among them, Sensors for collecting relevant parameters of the vehicle under sudden operating conditions, including longitudinal speed, lateral speed, yaw rate, and effective tire rolling radius; The signal processing module is used to pre-process the relevant parameters collected by the sensor and eliminate noise signals in the relevant parameters; The vertical force estimation module is implemented based on a radial neural network and is used to estimate the vertical load according to the output of the signal processing module; an evaluation index acquisition module, configured to calculate a yaw stability factor and a rollover stability factor based on the vertical load; A weight factor calculation module uses a fuzzy algorithm to perform fuzzy processing on the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then defuzzifies the fuzzy rules to calculate a control weight factor q; The control module is used to achieve safe and coordinated control of the vehicle based on the estimated vertical load and the weight control factor q.

2. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 1, characterized in that, The yaw stability factor is as follows: Among them, is the stability boundary coefficient of the phase plane, is the centroid side slip angle and the centroid side slip angular velocity.

3. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 2, characterized in that, The rollover stability factor is as follows: wherein, are respectively the vertical loads on the left and right wheels of the vehicle, are respectively the position of the axle and the number of axles.

4. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 3, characterized in that, The weight factor calculation module first defines a fuzzy set of yaw and rollover risk levels as {low, medium-low, medium, medium-high, high}, i.e., {SA, LR, MR, MHR, HR}, and outputs a fuzzy set representing the degree of comparison between the two, obtaining a result fuzzy set, i.e., {H1, H2, H3, H4, H5}, representing the relative weight of rollover risk to yaw, representing small, small, medium, large, and large, respectively, and formulating a fuzzy rule table; Set the control weight factor q within and realize defuzzification through the area centroid method.

5. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 4, characterized in that, The fuzzy rule table is:

6. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 3, wherein The vertical force estimation module is obtained by the following process: Neural network structure design: The radial basis function neural network architecture consists of three layers: input layer, hidden layer and output layer; Neural network training: A vehicle simulation model under tire blowout conditions is established based on the Doguff tire model and the CarSim model. The Doguff tire model is used to simulate changes in tire stiffness, and the CarSim model is used to apply sudden vertical forces to simulate changes in the tire's effective rolling radius. The vehicle simulation model is used to obtain sample data for training, and the sample data is used to train the neural network.

7. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 6, characterized in that, The acquisition of the vertical force estimation module also includes a neural network evaluation link, which uses mean square error and generalization error to verify the effectiveness and reliability of the obtained radial basis function neural network estimation model.

8. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 1, characterized in that, The relevant parameters include the longitudinal acceleration of the vehicle , the lateral acceleration of the vehicle , the roll angle of the vehicle , the roll angular velocity , the effective rolling radius of the tire , the longitudinal velocity , the lateral velocity and the yaw angular velocity .

9. The active safety coordination control system for intelligent vehicles under sudden emergency conditions according to claim 1, wherein The signal processing module performs feature extraction, Kalman filtering and sliding average on the signals collected by the sensor to eliminate noise signals in related parameters.

10. An active safety coordination control method for intelligent vehicles under sudden emergency conditions, characterized in that, The specific process is: Using sensors to collect relevant parameters of the vehicle under sudden working conditions, the relevant parameters including longitudinal speed, lateral speed, yaw angular velocity and tire effective rolling radius; Use the signal processing module to pre-process the relevant parameters collected by the sensor and eliminate the noise signals in the relevant parameters; Based on the vertical force estimation module, the vertical load is estimated according to the output data of the signal processing module; Calculate the yaw stability factor and rollover stability factor based on the vertical load evaluation index acquisition module; Using the fuzzy algorithm, the weight factor calculation module performs fuzzy processing on the yaw stability factor and the rollover stability factor to obtain fuzzy rules, and then performs defuzzification calculation on the fuzzy rules to control the weight factor q; According to the estimated vertical load and the weight control factor q, the control module realizes the safety coordinated control of the vehicle.

Images