CN107972667A - The man-machine harmony control method and its control system of a kind of deviation auxiliary system - Google Patents

Abstract

The invention discloses a man-machine coordination control method and a control system of a lane departure assisting system. The man-machine coordinated control method includes the following steps: after the lane departure assist system is started, according to the vehicle lateral deviation y and the target path f(t), obtain the desired steering wheel angle θ ^* required for vehicle steering; obtain the desired auxiliary rotation angle θ * according to θ ^* moment Design the driver's actual operating torque T _d and y as a double input, and the weight coefficient σ as a single output man-machine coordination controller; through σ and The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system. The man-machine coordinated control method of the present invention dynamically adjusts the assisting torque of the lane departure assisting system by outputting the assisting weight, realizes the coordinated control between the driver and the assisting system, and can effectively prevent the vehicle from deviating from the lane while reducing the driver's and Mutual interference between auxiliary systems avoids man-machine conflicts and has better man-machine coordination performance.

Description

A human-machine coordinated control method and control system for a lane departure assistance system

technical field

The invention relates to a human-machine coordinated control method and a human-machine coordinated control system in the technical field of assisted driving of intelligent vehicles, in particular to a human-machine coordinated control method and a human-machine coordinated control system of a lane departure assist system.

Background technique

Lane departure assistance system (Lane departure assistance system, LDAS) is an important part of intelligent car assisted driving technology. It can assist the driver to control the vehicle by actively intervening. Therefore, how to coordinate the relationship between the driver and the auxiliary system Control has become a hot issue in the field of intelligent vehicle assisted driving research at home and abroad.

There are two main ways to realize lane departure assist control: steering control and differential braking control. Steering control can be divided into torque control and angle control. Torque control is based on the steering system applying an additional steering force to the steering mechanism to achieve auxiliary control; steering angle control requires the steering system to control the wheels to turn to the desired angle to achieve auxiliary control. Differential braking control is to distribute the desired braking pressure to the wheels on both sides for differential braking, so that the yaw response of the vehicle can track the desired value and realize lane departure assist control.

When the electric power steering is used for lane departure assistance, the vehicle can realize lane departure assistance under various working conditions, which has strong adaptability. However, the use of steering control for lane departure assistance will cause mutual interference between the driver and the auxiliary system. If the coordination is inconsistent, it will lead to human-machine conflicts, which may increase the burden on the driver and affect the lateral safety of the vehicle. Therefore, it is of great significance to effectively coordinate the driver and the auxiliary system for lane departure assistance control to improve the performance of human-machine coordination.

Contents of the invention

Based on the technical problems existing in the background technology, the present invention proposes a human-machine coordinated control method of a lane departure assistance system and a human-machine coordinated control system thereof.

The solution of the present invention is: a human-machine coordinated control method of a lane departure assist system, which comprises the following steps:

After the lane departure assist system is started, according to the vehicle lateral deviation y and the target path f(t), the desired steering wheel angle θ ^* required for vehicle steering is obtained;

According to the desired steering wheel angle θ ^* , the desired assist torque is obtained

Design a human-machine coordination controller with two inputs and one output. The operating torque T _d and the vehicle lateral deviation y are used as the two inputs of the human-machine coordination controller. The output of the human-machine coordination controller is the weight coefficient σ;

By weight coefficient σ and desired assist torque The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system.

As a further improvement of the above scheme, the cross-lane time is used as the judgment algorithm of lane departure, and the vehicle departure judgment algorithm based on the cross-lane time predicts the vehicle trajectory through the established vehicle motion model, so as to calculate the minimum time required for the wheels to touch the edge of the lane. time.

As a further improvement of the above scheme, according to the vehicle lateral deviation y and the actual steering wheel angle θ, the expected steering wheel angle θ ^* is calculated through the driver model.

Preferably, the difference between the actual steering wheel angle θ and the desired steering wheel angle θ ^* is made, and the desired assist torque required for vehicle steering is obtained through the PID controller of the BP neural network

As a further improvement of the above scheme, the human-machine coordination controller includes a fuzzy neural network controller based on a five-layer topology, and the five-layer topology of the fuzzy neural network controller is: an input layer, a fuzzy layer, and an inference layer , the normalization layer and the output layer; the operating torque T _d and the vehicle lateral deviation y are double inputs, and the weight coefficient σ is a single output.

Preferably, the principles satisfied by the fuzzy neural network controller include:

(1) When |T _d |＞T _d ^max , the vehicle is in an emergency state, the weight coefficient σ of the actual assist torque T _a is the lowest, and the driver completely occupies the driving authority of the vehicle. Among them, Indicates the maximum value of threshold 2 set for judging the driver's operating state;

(2) When |T _d |<T _d ⁰ , the driver does not operate the steering wheel at this time, and the lane departure assist system occupies the vehicle driving authority, and the weight coefficient σ increases with the increase of the vehicle lateral deviation y, where , Indicates the minimum value of the set threshold 2;

(3) When T _d ⁰ ≤|T _d |≤T _d ^max and |y|<y ^min , the vehicle is in the center of the lane at this time, and there is no danger of deviating from the lane, so the weight coefficient of the actual assist torque T _a should be reduced σ, giving the driver as much control over the vehicle as possible, where y ^min represents the threshold three set for the vehicle to still be in the center of the lane;

(4) When T _d ⁰ ≤|T _d |≤T _d ^max and |y|≥y ^min , there are three situations for discussion: if the operation torque T _d is opposite to the actual assist torque T _a , it means that the driver At this time, it is necessary to increase the weight coefficient σ to the actual assist torque T _a to correct the vehicle trajectory; if the operating torque T _d is in the same direction as the actual assist torque T _a , it means that the driver is steering correctly.

Preferably, the universe of the input operating torque T _d is [-8,8], and the fuzzy subsets are {NB, NM, NS, Z, PS, PM, PB}, respectively representing {negative large, negative medium , negative small, zero, positive small, positive middle, positive large}; the domain of the input vehicle lateral deviation y is set to [-0.6,0.6], and the fuzzy subset is also {NB,NM,NS,Z,PS,PM, PB}, respectively represent {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}; the domain of the output weight coefficient σ is [0,1], and the fuzzy subset is {Z,S,M ,L,VL}, respectively represent {zero, small, medium, large, very large}; let the input vector X=[x ₁ ,x ₂ ] ^T (x ₁ =T _d ,x ₂ =y), the kth layer The output is represented by y ^(k) , (k=1,2,3,4,5), and the functions of each layer are: the first layer: input layer, the second layer: fuzzy layer, the third layer: reasoning layer, The fourth layer: normalization layer, the fifth layer: output layer.

More preferably, the first layer: input layer, each neuron node of the input layer corresponds to a continuous variable x _i , and the nodes of this layer directly transmit the input data to the second layer nodes, thus, the output Expressed as follows:

The second layer: the fuzzy layer, the value of the input continuous variable x _i is fuzzified according to the membership function on the three defined fuzzy subsets. Each node of this layer represents a language variable value, and the total node The number is 14, the j-th level membership degree corresponding to the i-th output of the first layer The calculation formula is expressed as:

In the formula: c _ij , σ _ij represent the center and width of the membership function respectively;

The third layer: reasoning layer, each neuron node represents a corresponding fuzzy rule, and the applicability of each fuzzy rule is calculated by matching the degree of membership obtained by the nodes in the second layer, and the total number of nodes is n, where n=49 , then the mth node of the third layer The output is:

In the formula, is the membership degree of level j corresponding to the first output of the first layer, is the membership degree of level j corresponding to the second output of the first layer;

The fourth layer: normalization layer, which performs overall normalization calculation on the network structure, the total number of nodes is n, and the mth node of the fourth layer The output is:

The fifth layer: the output layer, which clears the fuzzified variables and performs defuzzification calculations. The network output y ⁽⁵⁾ is equal to the sum of the products of the output of each node in the fourth layer and its corresponding weight:

In the formula: w _m represents the mth node and the output node of the fourth layer connection weights between.

The present invention also provides a human-machine coordinated control system of a lane departure assist system, which includes:

Desired steering wheel angle θ ^* and desired assist torque Obtaining module, after the lane departure assist system is started, according to the vehicle lateral deviation y and the target path f(t), obtain the desired steering wheel angle θ ^* and the desired assist torque required for vehicle steering

The human-machine coordinated control basis acquisition module acquires the driver's actual operating torque T _d , and uses the operating torque T _d and the vehicle lateral deviation y as the basis for man-machine coordinated control;

Human-machine coordination controller design module, design a dual-input and single-output human-machine coordination controller, the operating torque T _d and vehicle lateral deviation y are used as two inputs of the human-machine coordination controller, and the output of the human-machine coordination controller is the weight coefficient σ; and

The actual assist torque T _a optimization module, through the weight coefficient σ and the desired assist torque The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system.

The human-machine coordination control method of the lane departure assistance system of the present invention is based on the fuzzy neural network control theory, and is designed to consider the driver torque and the vehicle lateral Biased Human-Machine Coordinated Controller. The human-machine coordination controller dynamically adjusts the assist torque of the lane departure assist system by outputting the assist weight, so as to realize the coordinated control between the driver and the assist system. The invention can effectively prevent the vehicle from deviating from the lane, reduce the mutual interference between the driver and the auxiliary system, avoid man-machine conflict, and have better man-machine coordination performance.

Description of drawings

Fig. 1 is a flow chart of the man-machine coordinated control method of the lane departure assistance system of the present invention.

FIG. 2 is a schematic structural diagram of a human-machine coordinated control system adopting the human-machine coordinated control method in FIG. 1 .

Fig. 3 is a schematic diagram of a single-point preview model adopted by the driver model in Fig. 2 .

Fig. 4 is a control structure diagram of the PID controller in Fig. 2 .

Fig. 5 is a schematic diagram of the fuzzy neural network topology structure of the coordination controller in Fig. 2 .

Fig. 6 is a flow chart of the method for optimizing the actual assist torque T _a of the lane departure assist system of the present invention.

Fig. 7 is a flow chart of the hardware-in-the-loop test of the human-machine coordinated control system in Fig. 2 .

Fig. 8 is a graph of test results of the driver's input torque of the man-machine coordinated control system in Fig. 2 , that is, the driver's operating torque T _d .

FIG. 9 is a graph showing test results of the weight coefficient σ of the human-machine coordinated control system in FIG. 2 .

Fig. 10 is a graph showing the experimental results of the actual assist torque T _a of the man-machine coordinated control system in Fig. 2 .

FIG. 11 is a graph showing test results of the vehicle lateral deviation y of the human-machine coordinated control system in FIG. 2 .

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

The traditional lane departure assist system will be activated when it is judged that the vehicle is about to deviate from the lane and the driver does not operate the steering wheel. Once the driver intervenes, the assist system will stop working. The system provides lane departure assistance through the Electric Power Steering System (EPS). For example, the motor driving the EPS applies torque to the steering column to change the front wheel angle _δf of the car, and the change of the front wheel angle _δf will cause the adjustment of the vehicle state and position, which is reflected in the vehicle’s movement on the road relative to the centerline of the lane The adjustment of the vehicle lateral deviation y.

The man-machine coordination control method of the lane departure assistance system of the present invention is used to cooperate with the driver to complete the steering when the vehicle is about to deviate from the lane. The system can effectively coordinate the driver and the lane departure assistance system, and timely perform lane departure assistance control to improve human-machine coordination performance. Therefore, the present invention can reduce the mutual interference between the driver and the lane departure assistance system while effectively preventing the vehicle from departing from the lane, avoiding human-machine conflicts, and having better human-machine coordination performance.

Example 1

Please refer to FIG. 1 and FIG. 2 , the human-machine coordinated control method of the lane departure assistance system of the present invention includes the following steps.

Step S11: Obtain the yaw rate ω, vehicle speed v and the vehicle lateral deviation y of the vehicle on the road relative to the centerline of the lane during the driving process of the vehicle, and use the yaw rate ω, vehicle speed v and vehicle lateral deviation y as the lane departure Judgments based.

Step S12, taking the minimum time required for the predicted wheel to touch the edge of the lane as the crossing time, comparing the crossing time with the set threshold one, and judging that the vehicle is about to Deviate from the exit lane.

In this embodiment, the crossing time is used as the judging algorithm for lane departure. Comparing the calculated cross-lane time with the set threshold one, and then judging whether the vehicle is about to deviate from the lane.

The vehicle departure judgment algorithm based on cross-lane time predicts the vehicle trajectory through the established vehicle motion model, so as to calculate the minimum time required for the wheels to touch the edge of the lane, that is, the cross-lane time. The specific expression for calculating the crossing time TLC is:

In the formula, d _lane represents the width of the lane, d _b represents the wheel base, θ is the vehicle heading angle (that is, the actual steering wheel angle), which can be obtained from the integral of the yaw rate ω, L represents the wheelbase, and ω, v, y are all obtained from step S11 Yaw rate ω, vehicle speed v, vehicle lateral deviation y.

In step S13, it is determined whether to activate the lane departure assistance system according to the judgment result.

When it is judged that the vehicle is about to deviate from the lane, the lane departure assistance system is activated. If in step S12, the calculated cross-lane time is less than the set threshold one, it means that the vehicle is about to deviate from the lane, and then step S13 activates the lane departure assistance system. If the calculated cross-lane time is greater than or equal to the set threshold one, it means that the vehicle will not deviate from the lane, and the lane departure assistance system will not be activated.

Step S14, according to the vehicle lateral deviation y and the actual steering wheel angle θ, obtain the desired steering wheel angle θ ^* and the desired assisting torque required for vehicle steering

In this embodiment, according to state parameters such as the vehicle lateral deviation y and the actual steering wheel angle θ, the expected steering wheel angle θ ^* and the expected assisting torque required for vehicle steering are respectively obtained through the driver model and the PID algorithm of the neural network First calculate the expected steering wheel angle θ ^* through the driver model, make the difference between the actual steering wheel angle θ and the expected steering wheel angle θ ^* , and obtain the desired assist torque required for vehicle steering through the PID controller of the BP neural network

The driver model is a single-point preview model as shown in Figure 3: f(t) is the target trajectory of the vehicle, y(t) is the lateral coordinates of the current position of the vehicle, and T is the preview time.

Assuming that the preview distance is d, the relationship between the preview time T and the preview distance d is:

According to the lateral speed of the vehicle, that is, the vehicle speed v and the lateral acceleration of the vehicle, the lateral coordinate y(t+T) of the vehicle position at time t+T can be predicted. At this time, an ideal steering angle is selected to make the vehicle generate lateral acceleration At time t+T, the lateral coordinate y(t+T) of the vehicle position is equal to the lateral coordinate f(t+T) of the target trajectory, then:

f(t+T)=y(t+T)

The optimal lateral acceleration can be obtained by combining the two equations

According to the vehicle kinematics relationship, the actual lateral acceleration can be obtained The relationship with the actual steering wheel angle θ:

In the formula, R is the steering radius of the vehicle, and i _sw is the transmission ratio of the steering system.

Finally, the optimal steering wheel angle required to track the target trajectory is the desired steering wheel angle θ ^* :

The PID controller of the BP neural network is shown in Figure 4, that is, the PID control structure of the neural network is mainly composed of two parts: the classic PID controller and the neural network. Classical PID controller: directly perform closed-loop control on the controlled object, and the three parameters of the controller are online tuning. Neural network: The output state of the neurons in the output layer corresponds to the three adjustable parameters of the PID controller. Through the self-learning of the neural network and the adjustment of the weighting coefficient, the output of the neural network corresponds to the PID control under a certain optimal control law parameter.

The neural network adopts a three-layer feed-forward network with a 3-5-3 structure. The number of neurons in the input layer is 3, which are the expected value, actual value and deviation of the yaw rate; the number of neurons in the hidden layer is 5; the number of neurons in the output layer is 3, which are the PID control parameters.

Let the input vector X=[x ₁ (n), x ₂ (n), x ₃ (n)] ^T , x ₁ (n), x ₂ (n), x ₃ (n) represent ω ^* (n) respectively , ω(n) and its deviation e(n); the output of the kth layer is represented by y ^(k) (n), (k=1,2,3); the activation function of the hidden layer neurons is positive and negative symmetric The sigmoid function:

The outputs of the output layer are

Since these three parameters cannot be negative, the activation function of the output layer is

Therefore, the control law of the BP neural network PID controller is

Define the performance index function as

As shown in Figure 5, the BP learning algorithm is used to iteratively correct the weighting coefficients of the network, that is, search and adjust the negative gradient direction of the weighting coefficients according to ε(n), and add a momentum item to make the search quickly converge to the global minimum

In the formula, η is the learning rate; α is the momentum factor; w _li is the weighting coefficient of the hidden layer and the output layer.

In step S15, the driver's actual operating torque T _d is obtained, and the operating torque T _d and the vehicle lateral deviation y are used as the basis for man-machine coordinated control.

Step S16. Design a human-machine coordination controller with two inputs and one output. The operating torque T _d and the vehicle lateral deviation y are used as the two inputs of the human-machine coordination controller. The output of the human-machine coordination controller is the weight coefficient σ. That is, according to the operating torque T _d and the vehicle lateral deviation y, a human-machine coordination controller with two inputs and one output is designed.

The man-machine coordination controller comprises a fuzzy neural network controller based on a five-layer topology, and the five-layer topology of the fuzzy neural network controller is: an input layer, a fuzzy layer, an inference layer, a normalization layer and an output layer layer; take the operating torque T _d and the vehicle lateral deviation y as the double input, and the weight coefficient σ as the single output. Therefore, a dual-input and single-output man-machine coordination controller is designed based on the fuzzy neural network theory of five-layer topology.

The human-machine coordination controller is designed based on fuzzy neural network theory and fully considering the driver's operating torque T _d and the vehicle lateral deviation y.

The design of the fuzzy neural network controller for human-machine coordination needs to meet the specific principles.

(1) When the driver torque |T _d |>T _d ^max , the vehicle is in an emergency state, the weight coefficient of the actual assist torque T _a is the lowest, and the driver completely occupies the sovereignty of the vehicle.

(2) When |T _d |<T _d ⁰ , the driver does not operate the steering wheel at this time, and the lane departure assist system occupies the driving authority of the vehicle, and the weight coefficient σ increases with the increase of the lateral deviation y of the lateral vehicle . in, Indicates the maximum and minimum values of threshold 2 set for judging the driver's operating state.

(3) When T _d ⁰ ≤|T _d |≤T _d ^max and |y|<y ^min , the vehicle is in the center of the lane at this time, and there is no danger of deviating from the lane, so the weight coefficient of the actual assist torque T _a should be reduced σ, giving the driver as much autonomy as possible. Among them, y ^min represents the threshold three set for the vehicle is still in the center of the lane.

(4) When T _d ⁰ ≤|T _d |≤T _d ^max and |y|≥y ^min , there are three situations to discuss: if the driver torque is the direction of the operating torque T _d and the actual assist torque T _a On the contrary, it indicates that the driver misoperated. At this time, it is necessary to give the actual assist torque T _{a a larger} weight coefficient _σ to correct the vehicle trajectory; correct. The greater the driver torque, the smaller the weight coefficient σ of the actual assist torque T _a to reduce the intervention of the auxiliary system on the driver; if the lateral deviation y is large, the weight coefficient σ of the actual assist torque T _a is also larger, and vice versa.

The fuzzy neural network of the human-machine coordination controller adopts a five-layer topological structure of double input/single output, namely input layer, fuzzy layer, reasoning layer, normalization layer and output layer. Taking the operating torque T _d and the vehicle lateral deviation y as input, the weight coefficient σ is the output.

Assume that the domain of the input operating torque T _d is [-8,8], and the fuzzy subsets are {NB, NM, NS, Z, PS, PM, PB}, respectively representing {negative large, negative medium, negative small , zero, positive small, central, positive large}; the universe of vehicle lateral deviation y is set to [-0.6,0.6], and the fuzzy subset is also {NB,NM,NS,Z,PS,PM,PB}, respectively Indicates {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}; the domain of the output weight coefficient σ is [0,1], and the fuzzy subset is {Z, S, M, L, VL }, representing {zero, small, medium, large, very large} respectively. Let the input vector X=[x ₁ ,x ₂ ] ^T (x ₁ =T _d ,x ₂ =y), the output of the kth layer is y ^(k) , (k=1,2,3,4,5) Indicates that the functions of each layer are as follows:

The first layer: the input layer. Each neuron node in the input layer corresponds to a continuous variable x _i , and the nodes in this layer directly pass the input data to the second layer nodes, thus, the output Expressed as follows:

The second layer: fuzzy layer. The value of the input continuous variable x _i is fuzzified according to the membership function on the defined fuzzy subset. Each node of this layer represents a language variable value, and the total number of nodes is 14. The j-th level membership degree corresponding to the i-th output of the first layer The calculation formula can be expressed as:

In the formula: c _ij , σ _ij represent the center and width of the membership function respectively.

The third layer: reasoning layer. Each neuron node represents a corresponding fuzzy rule, and the applicability of each rule is calculated by matching the membership degree obtained in the second layer. The total number of nodes is n (n=49), then the mth node The output is:

In the formula, is the membership degree of level j corresponding to the first output of the first layer, is the membership degree of level j corresponding to the second output of the first layer. Simply put, it is the output of the second layer when i is 1 and 2 respectively.

The fourth layer: normalization layer. Perform an overall normalized calculation on the network structure, the total number of nodes is n, and the mth node of the fourth layer The output is:

The fifth layer: the output layer. Clear the fuzzified variables and perform defuzzification calculations. The network output y ⁽⁵⁾ is equal to the sum of the products of the output of each node in the fourth layer and its corresponding weight.

In the formula: w _m represents the mth node and the output node of the fourth layer connection weights between.

Step S17, through the weight coefficient σ and the desired assist torque The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system.

The human-machine coordination controller generates a weight coefficient σ in real time according to the value of the operating torque T _d and the vehicle lateral deviation y, and dynamically adjusts the size of the actual auxiliary torque T _a through this weight coefficient σ, and coordinates while ensuring safety. controls between the driver and assistance systems;

The designed human-machine coordination controller generates a dynamic weight coefficient σ in real time according to the driver's operating torque T _d and the value of the vehicle lateral deviation y, and through this weight coefficient σ and the desired auxiliary torque required by the vehicle steering Doing the product to adjust the size of the actual auxiliary torque T _a in real time can not only ensure that the vehicle does not deviate from the lane, but also realize the coordinated control between the driver and the auxiliary system.

_The actual assist torque T _a obtained through the above steps acts on the _steering system together with the driver’s operating torque T _d . At this time, it is necessary to give the actual assist torque T _a a larger weight coefficient σ to correct the vehicle trajectory. Lane departure assistance can be performed independently through the EPS system, such as changing the front wheel angle _δf of the car, the change of the car's front wheel angle _δf will cause the adjustment of the vehicle state, and finally change the vehicle lateral deviation y.

If the operating torque T _d is in the same direction as the actual assist torque T _a , it means that the driver is steering correctly. There is no need for lane departure assist via the EPS mechanism. The larger the operating torque T _d is, the smaller the weight coefficient σ of the actual assisting torque T _a is to reduce the intervention of the auxiliary system on the driver. At this time, the driver's operation and the assisting torque provided by the auxiliary system are coordinated The vehicle turns. If the vehicle lateral deviation y is larger, the weight coefficient σ of the actual assist torque T _a is also larger, and vice versa.

In other embodiments, the human-machine coordinated control method of the lane departure assistance system of the present invention may include the following simplified steps:

Taking the minimum time required for the predicted wheel to touch the edge of the lane as the crossing time, comparing the crossing time with a set threshold one, and starting the lane departure assist system when the crossing time is less than the set threshold one;

According to the vehicle lateral deviation y and the target path f(t), the desired steering wheel angle θ ^* required for vehicle steering is obtained;

According to the desired steering wheel angle θ ^* , the desired assist torque is obtained

Design the man-machine coordination controller with the driver's actual operating torque T _d and the vehicle lateral deviation y as the double input and the weight coefficient σ as the single output;

By weight coefficient σ and desired assist torque The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system.

The method proposed in this embodiment aims to provide a human-machine coordination control method of the lane departure assistance system. This method aims at the human-machine coordination problem between the driver and the lane departure assistance system in the process of lane departure assistance, and uses fuzzy neural Network control theory, design a man-machine coordination controller that considers the driver's operating torque T _d and the vehicle lateral deviation y, and dynamically adjusts the actual assist torque T a of the lane departure assist system by outputting the assist weight coefficient _σ to realize the driver's Coordinated control with auxiliary systems. The invention can effectively prevent the vehicle from deviating from the lane, reduce the mutual interference between the driver and the auxiliary system, avoid man-machine conflict, have better man-machine coordination performance, and can be further popularized.

Example 2

Please refer to FIG. 2 again. FIG. 2 shows a schematic structural diagram of a human-machine coordinated control system adopting the human-machine coordinated control method of Embodiment 1. The human-machine coordinated control system of the present invention includes an EPS mechanism and an optimization system for the actual auxiliary torque T _a .

The EPS mechanism includes a lane departure judging basis acquisition module, a lane departure judging module, and a departure assist control system starting module.

The lane departure judgment basis acquisition module acquires the yaw rate ω, the vehicle speed v, and the vehicle lateral deviation y of the vehicle on the road surface relative to the centerline of the lane during the driving process of the vehicle, and calculates the yaw rate ω, the vehicle speed v and the vehicle lateral deviation y is used as the basis for judging lane departure by the lane departure judging module.

The lane departure judging module takes the minimum time required for the predicted wheel to touch the edge of the lane as the crossing time, and compares the crossing time with a set threshold value 1, and when the crossing time is less than the set threshold value It is judged that the vehicle is about to deviate from the lane.

The starting module of the departure assistance control system decides whether to activate the lane departure assistance system according to the judgment result of the lane departure judgment module.

The optimization system for the actual assist torque T _a includes the desired steering wheel angle θ ^* and the desired assist torque Acquisition module, man-machine coordination control basis acquisition module, man-machine coordination controller design module, actual auxiliary torque T _a optimization module.

Desired steering wheel angle θ ^* and desired assist torque The acquisition module, according to the vehicle lateral deviation y and the target path f(t), obtains the desired steering wheel angle θ ^* and the desired assist torque required for vehicle steering

The human-machine coordinated control basis acquisition module acquires the driver's actual operating torque T _d , and uses the operating torque T _d and the vehicle lateral deviation y as the basis for man-machine coordinated control.

The human-machine coordination controller design module designs a double-input and single-output human-machine coordination controller. The operating torque T _d and the vehicle lateral deviation y are used as the two inputs of the human-machine coordination controller. The output of the human-machine coordination controller is the weight coefficient σ.

The actual assist torque T _a optimization module passes the weight coefficient σ and the expected assist torque The product is used to dynamically adjust the magnitude of the actual assist torque T _a of the lane departure assist system.

The details of the man-machine coordinated control system have been described in the man-machine coordinated control method of Embodiment 1, and will not be repeated here.

Example 3

Please refer to Fig. 2 and Fig. 6. Embodiment 3 shows the optimization method of the actual assist torque T _a of the lane departure assist system of the present invention, and the optimization method includes the following steps.

Step S21 , according to the vehicle lateral deviation y and the target path f(t) during the driving process of the vehicle, the desired steering wheel angle θ ^* required for vehicle steering is obtained.

According to the vehicle lateral deviation y and the target path f(t), the desired steering wheel angle θ ^* is calculated through the driver model, and the calculation method of the desired steering wheel angle θ ^* is as described in step S14 in Embodiment 1, and will not be repeated here. introduce.

Step S22, according to the actual steering wheel angle θ and the expected steering wheel angle θ ^* , the expected assist torque required for vehicle steering is obtained

Make the difference between the actual steering wheel angle θ and the desired steering wheel angle θ ^* , and obtain the desired auxiliary torque required for vehicle steering through the PID controller of the BP neural network desired assist torque The calculation method of is as described in step S14 in Embodiment 1, and will not be repeated here.

Step S23, design a human-machine coordination controller with two inputs and one output. The operating torque T _d and the vehicle lateral deviation y during vehicle running are used as the two inputs of the human-machine coordination controller, and the output of the human-machine coordination controller is the weight Coefficient σ.

The calculation method of the weight coefficient σ is as described in step S16 in Embodiment 1, and will not be repeated here.

Step S24, through the weight coefficient σ and the desired assist torque The product is used to dynamically optimize the magnitude of the actual assist torque T _a of the lane departure assist system.

If the direction of the driver’s torque, that is, the operating torque _Td , is opposite to the actual assisting torque _Ta , it means that the driver misoperated. At this time, it is necessary to give the actual assisting torque _{Ta a} larger weight coefficient σ to correct the vehicle trajectory. Lane departure assistance can be performed separately through the EPS system, such as changing the front wheel angle _δf of the car, the change of the car's front wheel angle _δf will cause the adjustment of the vehicle road model, and finally change the vehicle lateral deviation y.

If the operating torque T _d is in the same direction as the actual assist torque T _a , it means that the driver is steering correctly. There is no need for lane departure assist via the EPS mechanism. The larger the operating torque T _d is, the smaller the weight coefficient σ of the actual assisting torque T _a is to reduce the intervention of the assisting system on the driver. At this time, the driver's operation and the lane departure assistance of the EPS mechanism can be carried out synchronously . If the vehicle lateral deviation y is larger, the weight coefficient σ of the actual assist torque T _a is also larger, and vice versa.

Example 4

Please refer to FIG. 2 again. FIG. 2 also shows a schematic structural diagram of an optimization system for actual assist torque T _a using the method for optimizing actual assist torque T _a in Embodiment 3. Referring to FIG. The optimization system of the actual assist torque T _a of the present invention includes a desired steering wheel angle θ ^* acquisition module, the desired assist torque Acquisition module, man-machine coordination controller design module, actual auxiliary torque T _a optimization module.

The expected steering wheel angle θ ^* acquisition module obtains the expected steering wheel angle θ ^* required for vehicle steering according to the vehicle lateral deviation y and the target path f(t) during vehicle driving.

desired assist torque The acquisition module obtains the expected assist torque required for vehicle steering according to the actual steering wheel angle θ and the expected steering wheel angle θ ^*

The human-machine coordination controller design module designs a dual-input and single-output human-machine coordination controller. The operating torque T _d and the vehicle lateral deviation y during vehicle running are used as the two inputs of the human-machine coordination controller. The human-machine coordination controller The output of is the weight coefficient σ.

The actual assist torque T _a optimization module passes the weight coefficient σ and the expected assist torque The product is used to dynamically optimize the magnitude of the actual assist torque T _a of the lane departure assist system.

The details of the optimization system for the actual assist torque T _a have been described in the method for optimizing the actual assist torque T _a in Embodiment 3, and will not be repeated here.

Example 5

In order to verify the effectiveness and feasibility of the man-machine coordinated control method in Embodiment 1, the man-machine coordinated control method will be verified in detail below.

Using the simulation environment based on the CarSim vehicle model, combined with LabVIEW to conduct hardware-in-the-loop test research. The test platform and test block diagram are shown in Figure 7. The test bench built by the present invention is mainly composed of upper computer, lower computer, interface system and steering system. Establish the CarSim vehicle dynamics model and virtual road according to the vehicle parameters in the upper computer, and combine CarSim/LabVIEW to write the LabVIEW lane departure assistance control program; the lower computer is the PXI system of NI, which runs the program established by the upper computer in real time; the interface system is The torque and other signals collected by the sensor are transmitted to the PXI system, and the control signal is output to the controller of the actuator (such as the EPS motor controller that controls the auxiliary torque and the servo motor that generates the steering sense).

Choose a straight road as the simulated road, the road width is 3.75m, the vehicle speed is constant at 80km/h, a torque of 10N m is applied in 1s-1.5s to make the vehicle deviate from the center of the lane, and two representative driver operation modes are selected for man-machine The experimental verification of the coordinated control strategy, that is, when the vehicle deviates from the lane, the driver reacts, performs misoperation and correct operation.

Fig. 8-Fig. 11 are the test results of the man-machine coordinated control strategy, wherein Fig. 8 is the test result curve diagram of the driver's input torque, that is, the driver's operating torque T _d , and Fig. 9 is the test result curve diagram of the weight coefficient σ, Fig. 10 is a graph of the test results of the actual assist torque T _a , and FIG. 11 is a graph of the test results of the vehicle lateral deviation y.

When the driver steers correctly, the output weight coefficient σ of the human-machine coordination controller drops significantly, and the actual assist torque T _a is relatively small, thus giving the driver more sovereignty and reducing the impact of the assist system on the driver. interference. When the driver misoperates the steering wheel, the output weight remains at a large value, and the auxiliary controller, ie, the EPS mechanism, outputs a larger actual assist torque T _a to compensate for the wrong operation torque T _d applied by the driver. It can be seen from Figure 10 that no matter what operation the driver performs when the vehicle deviates, LDAS (Lane Departure Assist System) can still ensure that the vehicle does not deviate from the lane.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.

Claims (10)

1. A man-machine coordination control method of a lane departure auxiliary system is characterized by comprising the following steps:

after the lane departure auxiliary system is started, an expected steering wheel rotation angle theta required by vehicle steering is obtained according to the vehicle transverse deviation y and the target path f (t)^*；

According to desired steering wheel angle theta^*Deriving a desired assistance torque；

Designing a human-machine coordination controller with double inputs and single output, and operating a torque T_dAnd the vehicle transverse deviation y is used as two inputs of a man-machine coordination controller, and the output of the man-machine coordination controller is a weight coefficient sigma;

by the weight coefficient sigma and the desired assist torqueMultiplying to dynamically adjust an actual assist torque T of the lane departure assist system_aThe size of (2).

2. The method of claim 1, wherein a minimum time required for the wheels to contact the lane edge is predicted as a lane crossing time, the lane crossing time is compared with a first set threshold, and the lane departure support system is activated when the lane crossing time is less than the first set threshold.

3. The human-computer coordination control method of a lane departure assist system according to claim 2, wherein a lane crossing time is used as a determination algorithm of a lane departure, and a vehicle departure determination algorithm based on the lane crossing time predicts a vehicle travel track through the established vehicle motion model, thereby calculating a minimum time required for a wheel to contact a lane edge.

4. The human-machine coordination control method of the lane departure assistance system according to claim 1, wherein the desired steering wheel angle θ is calculated by a driver model based on the vehicle lateral deviation y and the target path f (t)^*。

5. The method of claim 4, wherein the actual steering wheel angle θ and the desired steering wheel angle θ are determined by a human-machine coordination control method^*Making a difference, and obtaining the expectation needed by the vehicle steering through a PID controller of a BP neural networkAssistance torque。

6. The human-machine coordination control method of the lane departure assistance system according to claim 1, wherein said human-machine coordination controller comprises a five-layer topology based fuzzy neural network controller, said five-layer topology of said fuzzy neural network controller being: the system comprises an input layer, a fuzzy layer, an inference layer, a normalization layer and an output layer; with operating torque T_dAnd the vehicle transverse deviation y is double input, and the weight coefficient sigma is single output.

7. The human-machine coordination control method of the lane departure assistance system according to claim 6, wherein the fuzzy neural network controller satisfies the criteria comprising:

(1) when | T_d|＞T_d ^maxAt this time, the vehicle is in an emergency state, and the actual assist torque T_aHas the lowest weight coefficient sigma, the driver fully occupies the vehicle driving ownership, wherein,a maximum value of the second threshold value set for judging the operation state of the driver;

(2) when | T_d|＜T_d ⁰When the driver does not operate the steering wheel, the lane departure assist system occupies the vehicle driving master, and the weight coefficient sigma is increased as the vehicle lateral deviation y is increased, wherein,a minimum value representing the set threshold two;

(3) when T is_d ⁰≤|T_d|≤T_d ^maxAnd y < y^minAt this time, since the vehicle is in the center of the lane and there is no danger of deviating from the lane, the actual assist torque T is reduced_aThe weighting factor sigma of (a) gives the driver as much ownership as possible of the vehicle, wherein y^minA third threshold value set to indicate that the vehicle is still considered to be in the center of the lane;

(4) when T is_d ⁰≤|T_d|≤T_d ^maxAnd y | ≧ y^minAt this point, three cases are discussed: if the operating torque T_dAnd the actual assist torque T_aThe direction is opposite, which indicates that the driver operates by mistake, and the actual auxiliary torque T needs to be given at the moment_aIncreasing the weight coefficient sigma to correct the vehicle running track; if the operating torque T_dAnd the actual assist torque T_aThe direction is the same, which indicates that the driver turns correctly.

8. The method of claim 6, wherein the input operation torque T is set_dHas a discourse field of [ -8,8]The fuzzy subset is { NB, NM, NS, Z, PS, PM, PB }, NB, NM, NS, Z, PS, PM, PB is the operating torque T_dFuzzy linguistic variables after fuzzification respectively represent { negative big, negative middle, negative small, zero, positive small, positive middle and positive big }; the input universe of vehicle lateral deviation y is set to [ -0.6,0.6 [ -0.6 []The fuzzy subset is also { NB, NM, NS, Z, PS, PM, PB }, which respectively represents { big negative, middle negative, small negative, zero, small positive, middle positive, big positive }; the output weight coefficient sigma has a domain of [0,1 ]]The fuzzy subset is { Z, S, M, L, VL }, which respectively represents { zero, small, medium, large }; let input vector X be [ X ]₁,x₂]^T(x₁＝T_d,x₂Y), output of k-th layer using y^(k)And (k ═ 1,2,3,4,5), each layer functions as: a first layer: input layer, second layer: blurring layer, third layer: inference layer, fourth layer: normalization layer, fifth layer: and (5) outputting the layer.

9. The human-machine-coordinated-control method of the lane departure assistance system according to claim 8, characterized in that the first layer: an input layer, each neuron node of the input layer corresponds to a continuous variable x_iThe nodes of this layer directly transfer the input data to the second layerLayer nodes, thus, outputsIs represented as follows:

<mrow> <msubsup> <mi>y</mi> <mi>i</mi> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <msub> <mi>x</mi> <mi>i</mi> </msub> <mo>,</mo> <mrow> <mo>(</mo> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow>

a second layer: fuzzification layer, which is the continuous variable x of input_iThe fuzzy processing is carried out according to membership function on three defined fuzzy subsets, each node of the layer represents a language variable value, the total node number is 14, the ith output of the first layer corresponds to the jth membershipThe calculation formula is expressed as:

<mrow> <msubsup> <mi>y</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <mfrac> <msup> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>i</mi> </msub> <mo>-</mo> <msub> <mi>c</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <msubsup> <mi>&amp;sigma;</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mn>2</mn> </msubsup> </mfrac> <mo>)</mo> </mrow> </mrow>

in the formula: c. C_ij,σ_ijRespectively representing the center and the width of the membership function;

and a third layer: and in the inference layer, each neuron node represents a corresponding fuzzy rule, the applicability of each fuzzy rule is calculated by matching the membership obtained by the second layer of nodes, the total number of the nodes is n, wherein n is 49, and the mth node in the third layer isThe output of (c) is:

<mrow> <msubsup> <mi>y</mi> <mi>m</mi> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <msubsup> <mi>y</mi> <mrow> <mn>1</mn> <mi>j</mi> </mrow> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </msubsup> <msubsup> <mi>y</mi> <mrow> <mn>2</mn> <mi>j</mi> </mrow> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </msubsup> <mo>,</mo> <mrow> <mo>(</mo> <mi>m</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>n</mi> <mo>)</mo> </mrow> </mrow>

in the formula,the j-th degree of membership corresponding to the 1 st output of the first layer,the j-th level membership degree corresponding to the 2 nd output of the first layer;

a fourth layer: a normalization layer for performing overall normalization calculation on the network structure, the total node number is n, and the mth node of the fourth layerThe output of (c) is:

<mrow> <msubsup> <mi>y</mi> <mi>m</mi> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <mfrac> <msubsup> <mi>y</mi> <mi>m</mi> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </msubsup> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>m</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <msubsup> <mi>y</mi> <mi>m</mi> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </msubsup> </mrow> </mfrac> </mrow>

and a fifth layer: the output layer is used for clarifying the fuzzified variable, performing anti-fuzzy calculation and outputting y through a network⁽⁵⁾Equal to the sum of products of the outputs of the nodes of the 4 th layer and the corresponding weights:

<mrow> <msup> <mi>y</mi> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </msup> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>m</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <msub> <mi>w</mi> <mi>m</mi> </msub> <msubsup> <mi>y</mi> <mi>m</mi> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </msubsup> </mrow>

in the formula: w is a_mRepresenting the m-th node and the output node of the 4 th layerThe connection weight between them.

10. A human-machine-coordinated control system of a lane departure assistance system, characterized by comprising:

desired steering wheel angle theta^*And a desired assist torque T_a ^*The acquisition module is used for obtaining an expected steering wheel corner theta required by vehicle steering according to the vehicle transverse deviation y and the target path f (t) after the lane departure auxiliary system is started^*And desired assist torque；

The man-machine coordination control obtains the actual operation torque T of the driver according to an obtaining module_dWill operate a torque T_dAnd the vehicle transverse deviation y is used as the basis of man-machine coordination control;

a design module of the human-computer coordination controller, a design module of the human-computer coordination controller with double input and single output, and an operation torque T_dAnd the vehicle transverse deviation y is used as two inputs of a man-machine coordination controller, and the output of the man-machine coordination controller is a weight coefficient sigma; and

actual assist torque T_aAn optimization module for optimizing the desired assist torque by a weighting factor sigmaMultiplying to dynamically adjust an actual assist torque T of the lane departure assist system_aThe size of (2).