Multi-target adaptive cruise control method based on fuzzy model predictive control

Abstract

The invention relates to a multi-target adaptive cruise control method based on fuzzy model predictive control. The multi-target adaptive cruise control method comprises the following steps: collecting real-time operation data of a vehicle; inputting the real-time operation data of the vehicle into the improved vehicle longitudinal dynamic model, and predicting the future motion state of the vehicle; on the basis of the output of the vehicle longitudinal dynamics model, a feedback correction ACC system vehicle following prediction model is used for predicting and correcting the vehicle following behavior in real time, and a vehicle following control strategy is dynamically adjusted; dynamically adjusting the weight of a target function under model prediction control by adopting a fuzzy variable weight adjustment strategy; and performing model prediction control according to the target function and considering multiple constraint conditions, generating accurate acceleration and deceleration control parameters, acting on the controlled vehicle, and realizing multi-target adaptive cruise control. Compared with the prior art, safe, energy-saving and efficient self-adaptive vehicle following control can be achieved in a complex and changeable traffic environment, and an effective technical scheme is provided for improving the response speed and adaptability of an intelligent cruise control system.

Description

Technical Field

[0001] The present invention relates to the field of automatic driving assistance systems, and in particular to a multi-objective adaptive cruise control method based on fuzzy model predictive control. Background Art

[0002] With the rapid development of the global social economy and the continuous advancement of automobile industry technology, the number of cars has increased rapidly and has become the main means of transportation for people's daily travel. This phenomenon has provided great convenience for life travel and transportation, but it has also brought about frequent traffic accidents, traffic congestion and automobile exhaust emissions, which have seriously affected people's quality of life and economic development. In the context of the global advocacy of low-carbon economy and green life, traffic safety and environmental protection have become two important goals in the field of road transportation.

[0003] In recent years, many research institutions and automobile manufacturers around the world have devoted themselves to the research of advanced driver assistance systems (ADAS), aiming to improve the driver's riding comfort, reduce traffic accidents caused by fatigue and wrong operation, and thus improve the safety of vehicle systems. These systems include adaptive cruise control (ACC), forward collision warning, etc. Among them, the ACC system is the main longitudinal control technology in ADAS and is widely used in passenger cars and commercial vehicles. The ACC system is developed on the basis of traditional cruise control. It not only has a constant speed function, but also can automatically adjust the throttle and brakes through system decisions according to the driving environment and vehicle information, so as to maintain a safe distance with the vehicle in front, significantly reducing the driver's operating burden, reducing driving fatigue, improving driving safety and comfort, and effectively reducing fuel consumption and emissions.

[0004] At present, the main ACC control strategies include classic PID control, numerical table lookup method and model predictive control (MPC). Classic PID control has the characteristics of simple structure and good stability, but it is difficult to use a set of parameters to meet all working conditions in practical applications, and it is not adaptable to the scene; the numerical table lookup method can obtain the reference acceleration under the current vehicle speed, and switch the driving and braking modes according to the reference acceleration. Although this method has high reliability, it requires a large number of real vehicle tests to calibrate the corresponding relationship between vehicle speed and reference acceleration, and the economic cost is high; in contrast, model predictive control MPC can easily handle multiple objectives and constraints. Under the condition of meeting the constraints, it comprehensively coordinates the requirements of following the vehicle and comfort, and is more suitable as an ACC control algorithm.

[0005] However, it also has the following limitations: First, the current control method of the ACC system is not adaptable enough when dealing with complex and changeable working conditions. The control accuracy is easily affected by the difference between the controlled state quantity and the target state quantity. There is a lack of predictability of control errors, resulting in low accuracy and delayed response. Secondly, the current driving goals rarely consider traffic emissions and fail to effectively integrate emission reduction needs. This limitation makes it difficult for the ACC system to achieve emission reduction goals while optimizing safety. Especially in complex urban traffic environments, control methods that ignore emission reduction optimization may lead to uneconomical driving behaviors such as frequent acceleration and deceleration, stop and start, etc., thereby increasing fuel consumption and emissions, and hindering ACC's potential contribution to the green transportation transformation. Furthermore, the ACC system does not fully consider driving comfort. The control algorithm may cause unnecessary rapid acceleration or deceleration, thereby reducing the riding experience and driving stability.

[0006] Therefore, although the MPC method is more suitable for current driving needs than other traditional methods, it still needs further exploration in terms of multi-objective considerations and weight changes. Summary of the invention

[0007] The purpose of the present invention is to provide a multi-objective adaptive cruise control method based on fuzzy model predictive control to address the problem that the existing technology in the adaptive cruise control (ACC) system is difficult to fully consider the driving comfort and fuel economy and cannot adapt to the needs of complex road conditions. By dynamically adjusting the weight coefficients, setting multiple objectives, and adapting to different traffic conditions, the coordinated improvement of driving safety, comfort and emission optimization can be achieved.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] A multi-objective adaptive cruise control method based on fuzzy model predictive control comprises the following steps:

[0010] Collect real-time operation data of vehicles;

[0011] Inputting the real-time running data of the vehicle into an improved vehicle longitudinal dynamics model to predict the future motion state of the vehicle, wherein the improved vehicle longitudinal dynamics model selects the actual vehicle distance between the two vehicles, the speed of the vehicle, the relative speed of the two vehicles, the acceleration of the vehicle and the rate of change of the acceleration of the vehicle as state variables, and considers the interference of the acceleration of the preceding vehicle on the motion of the vehicle;

[0012] Based on the output of the vehicle longitudinal dynamics model, the ACC system following prediction model is used to make real-time predictions and corrections on the following behavior, and dynamically adjust the following control strategy.

[0013] The fuzzy variable weight adjustment strategy is used to dynamically adjust the weight of the objective function of the model predictive control according to the feedback correction of the vehicle's motion state after the ACC system's following prediction model is corrected;

[0014] Model predictive control is performed based on the objective function and taking multiple constraints into consideration to generate accurate acceleration and deceleration control parameters, which are applied to the controlled vehicle to achieve multi-objective adaptive cruise control.

[0015] The improved vehicle longitudinal dynamics model establishes the vehicle state equation based on the relative motion relationship:

[0016]

[0017] Where x(k) is the state variable of the system, x(k)=[Δd(k),Δv(k),a f (k),a j (k)] T ; u(k) is the control variable, i.e. the desired acceleration of the vehicle a des ; w(k) is the disturbance, i.e. the acceleration of the front vehicle a p ; T s is the sampling time; T L is the time constant of the inertia link; a f is the vehicle acceleration, a j is the jerk of the vehicle, Δv is the relative speed, d is the distance between the two vehicles, and Δd is the deviation between the actual vehicle distance and the ideal vehicle distance.

[0018] The objective function and multiple constraint conditions are constructed based on multiple performance indicators, and the multiple performance indicators include a vehicle following safety performance indicator, a driving speed constraint, a dynamic following performance indicator, a driving comfort performance indicator, and a fuel economy performance indicator.

[0019] For the following vehicle safety performance index, the distance between the preceding vehicle and the vehicle is set to be greater than the preset safe following distance:

[0020] d(k)≥d c

[0021] Among them, d(k) is the distance between the two vehicles at time k, d c To ensure safe following distance;

[0022] For driving speed constraints, the following constraints are imposed on the speed of the vehicle during driving:

[0023] v fmin ≤v f (k)≤v fmax

[0024] Among them, v fmin ,v fmaxare the minimum and maximum vehicle speeds, respectively, and v f (k) is the vehicle speed;

[0025] For the dynamic followability performance index, the relative speed and inter-vehicle distance deviation of the front and rear vehicles are set to converge to zero, that is, the two vehicles are in a relatively static state: Δd(k)→0

[0026] v rel (k)→0

[0027] Where Δd(k) is the inter-vehicle distance error, Δd(k) = d(k) - d des , d des is the ideal vehicle spacing, v rel (k) is the relative speed between the leading and following vehicles;

[0028] For the driving comfort performance index, select the vehicle acceleration a f The absolute value of (k) and the absolute value of the acceleration change rate jerk(k) are used as quantitative characterization parameters of comfort, and the comfort optimization objective is defined as:

[0029] min|a f (k)|

[0030] min|jerk(k)|

[0031] At the same time, the following constraints are imposed on acceleration and acceleration rate:

[0032] a fmin ≤a f (k)≤a fmax

[0033] jerk min ≤jerk(k)≤jerk max

[0034] Among them, a fmin 、a fmax are the minimum and maximum values ​​of acceleration, jerk min , jerk max are the minimum and maximum values ​​of the acceleration rate of change respectively;

[0035] For fuel economy, the driving comfort performance index is used to indirectly characterize the fuel economy level, and the expected acceleration value range is restricted:

[0036] u min ≤u(k)≤u max

[0037] Among them, u max ,u minare the maximum and minimum values ​​of the expected acceleration respectively, and u(k) is the expected acceleration of the vehicle.

[0038] The feedback correction ACC system following prediction model dynamically adjusts the acceleration and deceleration of the vehicle according to the prediction error between the real-time state and the predicted state, wherein the prediction error between the actual state and the predicted state of the ACC system at time k is defined as follows:

[0039] x p (k|k-1)=A·x(k-1)+B·u(k-1)+G·w(k-1)

[0040] Among them, x p (k|k-1) is the prediction error, x(k-1) is the system state quantity at time k-1, u(k-1) is the system control quantity at time k-1, and w(k-1) is the external interference quantity at time k-1.

[0041] The fuzzy variable weight adjustment strategy includes a fuzzy rule base, a fuzzy reasoning engine and online weight adjustment, wherein the fuzzy rule base is constructed based on multiple performance indicators and defines control strategies under different traffic scenarios; the fuzzy reasoning engine is responsible for judging the complexity of the driving environment according to the current motion state of the vehicle during real-time driving, and determining the weight adjustment plan according to the fuzzy rule base; the online weight adjustment uses dynamic fuzzy logic to update the weight coefficient of the model predictive control in real time according to actual working conditions.

[0042] The objective function is expressed as:

[0043]

[0044] Among them, N p is the prediction time domain, u is the control variable, Q and R are weight coefficients, y p is the vehicle trajectory, y ref is the target reference trajectory.

[0045] An exponential decay function is introduced into the target reference trajectory to make each component of the output vector smoothly approach zero, which can be expressed as:

[0046] y ref (k+i)=Λ i y(k)

[0047] Among them, y(k) is the target vector of the performance indicator to be optimized,

[0048] Λ=diag(λ Δd ,λ vrel ,λ af ,λ jerk )

[0049] λΔd ,λ vrel ,λ af ,λ jerk are the deviation Δd between the actual vehicle spacing and the ideal vehicle spacing, the relative speed v of the front and rear vehicles rel , vehicle acceleration a f and the reference trajectory coefficient of the ego-vehicle acceleration change rate jerk, with a value range of (0, 1).

[0050] The weight coefficient Q is expressed as:

[0051] Q = diag (q Δd ,q vrel ,q af ,q jerk )

[0052] Among them, q Δd is the weight factor of the inter-vehicle distance error, q vrel is the weight factor of the relative speed of the two vehicles, q af is the weight factor of the vehicle’s acceleration, q jerk is the weight factor of the acceleration change rate of the vehicle;

[0053] The selection of the weight coefficient R affects the expected acceleration of the model predictive control decision output.

[0054] The weight coefficient and the reference trajectory coefficient are determined as follows:

[0055] L f =q Δd =q vrel =λ Δd =λ vrwl

[0056] L c =q af =q jerk =R = λ af =λ jerk

[0057] Among them, L f is the weight coefficient of followability, L c is the weight coefficient of comfort and fuel economy.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] (1) Enhanced robustness: By introducing a feedback correction mechanism, the present invention significantly improves the robustness of the vehicle following prediction model, which can more effectively cope with external environmental interference and uncertainty in vehicle parameters, ensuring the stability and safety of the vehicle under various complex traffic conditions.

[0060] (2) Comprehensive performance optimization: The present invention comprehensively considers dynamic following, driving comfort and fuel economy under the MPC framework. By optimizing the control objectives, it can achieve smooth acceleration and deceleration while maintaining safe following, thereby improving the overall driving experience.

[0061] (3) Smooth control output: The present invention uses a smooth curve to guide the control output, which reduces sudden acceleration and deceleration, makes the vehicle following process smoother, improves driving comfort, and reduces the psychological burden of the driver. By optimizing the target reference trajectory, it ensures that each component of the output vector smoothly approaches zero, achieving more humane driving control.

[0062] (4) Improved vehicle longitudinal dynamics model: This invention significantly improves the vehicle longitudinal dynamics model by redefining the state variables and taking into account the influence of the acceleration of the preceding vehicle on the motion of the vehicle. This makes the model closer to the actual vehicle operating state and more realistically reflects the dynamic behavior of the vehicle in different traffic scenarios, thereby improving the accuracy of prediction and the effectiveness of control. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 is a flow chart of the method of the present invention;

[0064] Figure 2 is the input-output membership function of the present invention, wherein (2a) is the inter-vehicle distance error, (2b) is the relative speed, and (2c) is the followability weight coefficient;

[0065] Figure 3 It is a three-dimensional map of the followability weight coefficient of the present invention with respect to Δd and Δv;

[0066] Figure 4 The verification results of the method of the present invention under the cruise control condition, wherein (4a) and (4c) are speed changes, and (4b) and (4d) are acceleration changes;

[0067] Figure 5 The verification results of the method of the present invention under the cutting-out condition of the leading vehicle are shown in FIG. 5 , where (5a) is the speed change and (5b) is the acceleration change;

[0068] Figure 6 is the verification result of the method of the present invention under the condition of side car cutting in, where (6a) is the speed change and (6b) is the acceleration change;

[0069] Figure 7 The verification results of the method of the present invention under complex road conditions, where (7a) is the speed change, (7b) is the probability distribution histogram of the acceleration change rate, and (7c) is the probability distribution histogram of the driving comfort. DETAILED DESCRIPTION

[0070] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0071] This embodiment is aimed at optimizing the adaptive cruise control (ACC) system under complex traffic scenarios, and adopts an improved model predictive control (MPC) algorithm, focusing on solving the problem that the current ACC system does not consider driving comfort and fuel economy enough. The vehicle longitudinal dynamics model selects the actual distance between the two vehicles, the speed of the vehicle, the relative speed of the two vehicles, the acceleration of the vehicle, and the rate of change of the acceleration of the vehicle as state variables, and also considers the interference of the acceleration of the front vehicle on the movement of the vehicle, which can more truly reflect the actual operating state. Under the premise of ensuring that the vehicle dynamically follows the front vehicle, the driving comfort, fuel economy and following safety are optimized at the same time. A feedback correction ACC system following prediction model is designed, which can dynamically adjust the distance between the vehicle and the front vehicle to ensure driving safety. In terms of control strategy, the fuzzy variable weight adjustment strategy is applied to achieve the balanced optimization of comfort and fuel economy under different working conditions. By dynamically adjusting the control weight, the system can flexibly adapt to complex and changeable road and traffic conditions. The present invention not only improves the driving stability, but also effectively reduces the fuel consumption and emissions during driving.

[0072] Specifically, this embodiment provides a multi-objective adaptive cruise control method based on fuzzy model predictive control, such as Figure 1 As shown, the following steps are included:

[0073] S1, collects real-time operation data of vehicles;

[0074] S2, inputting the real-time running data of the vehicle into an improved vehicle longitudinal dynamics model to predict the future motion state of the vehicle, wherein the improved vehicle longitudinal dynamics model selects the actual vehicle distance between the two vehicles, the speed of the vehicle, the relative speed of the two vehicles, the acceleration of the vehicle, and the rate of change of the acceleration of the vehicle as state variables, and considers the interference of the acceleration of the preceding vehicle on the motion of the vehicle, so as to more truly reflect the actual running state;

[0075] S3, based on the output of the vehicle longitudinal dynamics model, the ACC system following prediction model is used to make real-time predictions and corrections on the following behavior, and dynamically adjust the following control strategy;

[0076] S4, using a fuzzy variable weight adjustment strategy to dynamically adjust the weight of the objective function of the model predictive control according to the motion state of the vehicle corrected by the ACC system following vehicle prediction model based on feedback correction;

[0077] S5, based on the objective function and taking into account multiple constraints, model predictive control is performed to generate accurate acceleration and deceleration control parameters, which act on the controlled vehicle to achieve multi-objective adaptive cruise control.

[0078] The improved vehicle longitudinal dynamics model establishes the vehicle state equation based on the relative motion relationship, and selects a variety of state variables to more realistically reflect the vehicle's operating status. The model includes factors such as the desired acceleration of the vehicle, the rate of change of the vehicle's acceleration, the acceleration of the front vehicle, the relative speed of the two vehicles, and the deviation between the actual and ideal vehicle spacing. By comprehensively considering the disturbance effect of the front vehicle's acceleration, the adaptability to dynamic traffic environments is further improved. This setting is closer to the actual vehicle operating state, which helps to improve the control accuracy and response speed of the ACC system under complex traffic conditions. The established vehicle state equation is as follows:

[0079]

[0080] Where x(k) is the state variable of the system, x(k)=[Δd(k),Δv(k),a f (k),a j (k)] T ; u(k) is the control variable, i.e. the desired acceleration of the vehicle a des ; w(k) is the disturbance, i.e. the acceleration of the front vehicle a p ; T s is the sampling time; T L is the time constant of the inertia link; a f is the vehicle acceleration, a j is the jerk of the vehicle, Δv is the relative speed, d is the distance between the two vehicles, and Δd is the deviation between the actual vehicle distance and the ideal vehicle distance.

[0081] The objective function and multiple constraints are constructed based on multiple performance indicators. By setting multiple performance indicators and multiple constraints, a balance can be achieved among multiple conflicting optimization objectives. The control goal of the ACC system designed by the present invention is to ensure that the vehicle dynamically follows the preceding vehicle while optimizing driving comfort, fuel economy and following safety. Based on the longitudinal kinematic model of the vehicle, each control objective is analyzed, and the system optimization objective is converted into a constrained quadratic programming problem to achieve multi-objective control of the ACC system. These include following safety performance indicators, driving speed constraints, dynamic followability performance indicators, driving comfort performance indicators, and fuel economy performance indicators.

[0082] (1) Following vehicle safety performance indicators

[0083] Safety is the most basic and important indicator that the ACC system needs to ensure. To ensure safe driving and avoid rear-end collisions, the distance between the vehicle in front and the vehicle should always be greater than a preset safe following distance:

[0084] d(k)≥d c

[0085] Among them, d(k) is the distance between the two vehicles at time k, d c For safe following distance.

[0086] (2) Driving speed constraints

[0087] Considering the provisions of traffic laws and the limitations of the vehicle's own physical capabilities, it is necessary to restrict the speed of the vehicle during driving:

[0088] v fmin ≤v f (k)≤v fmax

[0089] Among them, v fmin ,v fmax are the minimum and maximum vehicle speeds, respectively, and v f (k) is the vehicle speed.

[0090] (3) Dynamic followability performance index

[0091] The following performance of the adaptive cruise control system should meet the requirement that the vehicle stably follows the vehicle in front. If the actual vehicle spacing is too large, it will not only reduce the road utilization rate, but also easily cause vehicles in the adjacent lane to frequently enter the lane. A too small vehicle spacing increases the possibility of rear-end collisions, causing psychological tension in the driver. The following performance is mainly reflected in the expected vehicle spacing and the speed of the front vehicle. The relative speed and vehicle spacing error of the front and rear vehicles should converge to zero, that is, the two vehicles are in a relatively static state Δ. d(k)→0

[0092] v rel (k)→0

[0093] Where Δd(k) is the inter-vehicle distance error, Δd(k) = d(k) - d des , d des is the ideal vehicle spacing, v rel (k) is the relative speed between the front and rear vehicles.

[0094] (4) Driving comfort performance index

[0095] To meet the requirements of driving comfort, the vehicle should first be kept at a constant speed, so that the acceleration is within a certain range and avoids large changes. Secondly, the acceleration change rate of the vehicle should be as small as possible. f The absolute value of jerk(k) and the absolute value of acceleration change rate jerk(k) are used as quantitative characterization parameters of ACC system comfort, and the comfort optimization objective is defined as:

[0096] min|a f (k)|

[0097] min|jerk(k)|

[0098] At the same time, the following constraints are imposed on acceleration and acceleration rate:

[0099] a fmin ≤a f (k)≤a fmax

[0100] jerk min ≤jerk(k)≤jerk max

[0101] Among them, a fmin 、a fmax are the minimum and maximum values ​​of acceleration, jerk min , jerk max are the minimum and maximum values ​​of the acceleration rate of change, respectively.

[0102] (5) Fuel economy performance indicators

[0103] Fuel economy reflects the energy consumption of ACC vehicles. Fuel consumption is mainly related to the magnitude of vehicle acceleration. When the average driving speed and driving distance are the same, fuel consumption increases with the increase of the absolute value of acceleration. Moreover, the better the driving smoothness of the vehicle, the higher the fuel economy. The evaluation index of driving comfort is used to indirectly characterize the fuel economy level. In addition to the above constraints, the range of expected acceleration values ​​output by the upper-level decision algorithm of the ACC system is also restricted:

[0104] u min ≤u(k)≤u max

[0105] Among them, u max ,u min are the maximum and minimum values ​​of the expected acceleration respectively, and u(k) is the expected acceleration of the vehicle.

[0106] Affected by external environmental interference and vehicle parameter uncertainty, there is inevitably a deviation between the predicted value of the prediction model and the actual state. In order to enhance the robustness of the following prediction model, the present invention introduces a feedback correction ACC system following prediction model. The feedback correction ACC system following prediction model adopts a closed-loop control structure, and realizes accurate following in complex traffic environments through real-time feedback adjustment. First, based on the acceleration, relative speed and current vehicle distance of the self-vehicle and the front vehicle, the model predictive control (MPC) algorithm is used to predict the relative position and vehicle distance changes in multiple time steps in the future, thereby ensuring the rationality and foresight of the prediction. Then, the deviation between the actual vehicle distance and the ideal vehicle distance is measured in real time to provide an adjustment basis for feedback control. In the feedback control link, the fuzzy variable weight strategy is used to dynamically adjust the acceleration or deceleration of the vehicle according to the error signal to balance the following performance and driving comfort, avoiding frequent rapid acceleration or deceleration. In addition, the optimal control command is generated in combination with the prediction result after feedback correction, and the acceleration and braking of the vehicle are dynamically adjusted, so that the self-vehicle can quickly adapt to the sudden changes in traffic flow and vehicle distance. Overall, the feedback correction model not only demonstrates high precision and response speed in dynamic following control, but also improves driving stability and comfort, reduces unnecessary acceleration and deceleration operations, effectively reduces fuel consumption and exhaust emissions, and provides technical support for the realization of green transportation.

[0107] In this embodiment, the ACC system following prediction model is feedback-corrected to dynamically adjust the acceleration and deceleration of the vehicle according to the prediction error between the real-time state and the predicted state. The prediction error between the actual state and the predicted state of the ACC system at time k is defined as follows:

[0108] x p (k|k-1)=A·x(k-1)+B·u(k-1)+G·w(k-1)

[0109] Among them, x p (k|k-1) is the prediction error, x(k-1) is the system state quantity at time k-1, u(k-1) is the system control quantity at time k-1, and w(k-1) is the external interference quantity at time k-1.

[0110] The feedback correction ACC system following prediction model defines the prediction error between the actual state and the predicted state of the ACC system at time k, taking into account the system state, control quantity and external disturbance at the previous time k-1. According to the basic principle of MPC (model predictive control), the current actual measurement value is used as the initial condition, and the future state is predicted based on the established vehicle longitudinal kinematic model. In the controller design, it is assumed that the control quantity remains unchanged outside the control domain, and the prediction domain N is set. p and control time domain N c. At the same time, the acceleration of the leading vehicle remains unchanged throughout the prediction time domain, thus establishing a disturbance vector. Based on these assumptions, the matrix form of the ACC system following prediction model is derived. Through the analysis of the system performance objectives, the model aims to enable the controlled output to effectively track the reference output while taking into account driving comfort, fuel economy and following safety. The control objectives include limiting the acceleration changes of the ego vehicle to achieve a smooth following effect. Under the MPC framework, an objective function is determined to optimize the deviation between the control quantity and the actual output by comprehensively considering dynamic following, driving comfort and economy. In order to meet the control requirements such as comfort and economy, the control output of the ACC system does not directly track the set value, but gradually reaches a new steady-state value along a specified smooth curve. In addition, by introducing an exponential decay function as the target reference trajectory for optimizing the performance index, it is ensured that the components of the output vector smoothly tend to zero to achieve a smooth following effect.

[0111] In the design of the ACC system controller, the current actual measurement value is used as the initial condition, and the future state of the vehicle is predicted based on the ACC system workshop longitudinal kinematics model formula. By comparing the predicted value with the expected reference value, the optimal control sequence of the system can be obtained. This invention assumes that the state of the system can be measured and N is set. p is the prediction time domain, N c is the control time domain. In order to derive the prediction equation, it is assumed that the control quantity remains unchanged outside the control time domain, that is:

[0112] u(k+N c -1)=u(k+N c )=……=u(k+N p -1)

[0113] Definition N c The control input vector of the system within the step is:

[0114]

[0115] It is also necessary to assume that the acceleration of the preceding vehicle, which is the disturbance of the system, remains consistent with the measured value at the current moment in the entire prediction time domain, that is:

[0116]

[0117] The disturbance vector in the system prediction time domain is defined as:

[0118]

[0119] By introducing the feedback correction mechanism, it is derived that p is the prediction time domain, with N c The matrix form of the ACC system following vehicle prediction model in the control time domain is:

[0120] X=A p ·x(k)+B p ·U+G p ·Φ+W p ·e(k)

[0121] Y=C p ·x(k)+F p ·U+S p ·Φ+L p e(k)-D p

[0122] N p The predicted state quantity X and predicted output quantity Y of the internal system are defined as follows:

[0123]

[0124]

[0125] Among them, each coefficient matrix satisfies:

[0126]

[0127]

[0128]

[0129]

[0130] The performance requirement of the ACC system is to make the controlled output y k (p) can track the reference output y ref (k), and considering the needs of improving driving comfort and fuel economy, the change of the control quantity, that is, the acceleration of the vehicle, should be limited. Under the MPC framework, dynamic following, driving comfort, fuel economy and following safety are comprehensively considered, taking into account various control objectives, and optimizing the system control quantity and the deviation between the actual output and the reference quantity to ensure that the ACC vehicle can quickly and smoothly track the front vehicle, and then determine the objective function as follows:

[0131]

[0132] Among them, N p is the prediction time domain, u is the control variable, Q and R are weight coefficients, y p is the vehicle trajectory, y ref is the target reference trajectory.

[0133] The weight coefficient Q is expressed as:

[0134]

[0135] Among them, q Δd is the weight factor of the inter-vehicle distance error. The larger its value is, the closer the actual inter-vehicle distance between the two vehicles is to the expected inter-vehicle distance calculated by the safe inter-vehicle distance model. is the weight factor of the relative speed of the two vehicles, is the weight factor of the vehicle’s acceleration, q jerk is the weight factor of the vehicle's acceleration rate of change. Similarly, The larger the value, the smaller the relative speed between the two vehicles, the acceleration of the vehicle, and the rate of change of acceleration.

[0136] The selection of the weight coefficient R affects the expected acceleration of the model predictive control decision output. The smaller its value, the greater the amplitude of the control amount.

[0137] Since the present invention comprehensively considers followability, comfort and fuel economy, in order to improve the adaptability to different traffic scenes, different weights are selected, so that Among them, L f is the weight coefficient of followability, L c is the weight coefficient of comfort and fuel economy, and the two have relative meanings.

[0138] In order to meet the control objectives of comfort and economy and achieve a smooth and smooth following effect, the control output of the ACC system does not directly track the set value, but gradually reaches a new steady-state value along a specified smooth curve. To this end, an exponential decay function is introduced as the target reference trajectory for optimizing the performance index, so that each component of the output vector tends to zero smoothly. The target reference trajectory is as follows:

[0139] y ref (k+i)=Λ i y(k)

[0140] Among them, y(k) is the target vector of the performance indicator to be optimized,

[0141]

[0142] λ jerk are the deviation Δd between the actual vehicle spacing and the ideal vehicle spacing, the relative speed v of the front and rear vehicles rel , vehicle acceleration a f and the reference trajectory coefficient of the ego-vehicle acceleration change rate jerk, with a value range of (0, 1).

[0143] In order to improve the tracking accuracy and stability of MPC control and improve the adaptive ability, the present invention designs a fuzzy control online optimization MPC control to realize adaptive variable weight control. The fuzzy variable weight adjustment strategy includes a fuzzy rule base, a fuzzy reasoning engine, and an online weight adjustment, wherein the fuzzy rule base is constructed based on multiple performance indicators and defines the control strategies under different traffic scenarios; the fuzzy reasoning engine is responsible for judging the complexity of the driving environment (such as the urgency of following the vehicle) according to the current motion state of the vehicle (such as acceleration, relative vehicle distance, speed difference) during real-time driving, and determining the weight adjustment scheme according to the fuzzy rule base; the online weight adjustment uses dynamic fuzzy logic to update the weight coefficient of the model predictive control in real time according to the actual working conditions.

[0144] The fuzzy variable weight adjustment strategy is used to optimize MPC control online to achieve adaptive variable weight control. The vehicle following performance and ride comfort required under different driving conditions are different, and the fixed weight MPC control algorithm cannot meet the performance requirements of the actual vehicle driving. The control algorithm should adaptively adjust the performance indicators according to the relative speed of the two vehicles and the distance error between the vehicles. The greater the degree of danger, the more important the corresponding following performance is, and the comfort and economy are relatively smaller. Fuzzy control is used to adjust each weight coefficient in real time, so that the vehicle can smoothly follow the vehicle in front.

[0145] In the embodiment of the present invention, the weight coefficients affecting the MPC tracking performance are: In order to improve the adaptability of the ACC system, different weight coefficients are selected and defined. is the follow-up weight coefficient, L c =λ af =λ jerk is the comfort economy weight coefficient.

[0146] That is, the weight coefficient and the reference trajectory coefficient are determined as follows:

[0147]

[0148]

[0149] Among them, L f is the weight coefficient of followability, L c is the weight coefficient of comfort and fuel economy.

[0150] Due to L f , L c With relativity, let L c = 1. The inter-vehicle distance error and relative speed are fuzzy classified into five levels: PB (maximum), PS (large), ZO (small), NS (small), and NB (minimum), with domains of [-60, 60] m and [-20, 20] m / s respectively; the followability weight coefficient L fThe fuzzy is divided into five levels: B (large), MB (medium-large), ZO (medium), MS (medium-small), S (small), and the domain is [0.1, 2.5]. The input-output membership function is as follows: Figure 2 shown.

[0151] When the following target and the ego vehicle are in the PB domain, the two vehicles are in a relatively safe driving state, and more attention should be paid to comfort and economic performance. On the contrary, when the following target and the ego vehicle are in the NB domain, the two vehicles are in a relatively dangerous driving state, and more attention should be paid to the safety performance of the following vehicle. Design fuzzy rules, and then use the centroid method to defuzzify. The three-dimensional map of the following weight coefficient with respect to Δd and Δv is as follows Figure 3 shown.

[0152] The embodiment of the present invention applies the multi-objective adaptive cruise control strategy method based on fuzzy MPC to simulation to verify its improvement effect on vehicle driving optimization. Carsim software is used to establish a front-wheel drive B-class sedan with an engine power of 125kW, a 5-speed automatic transmission, and an ABS function in the braking system, and a Carsim and MATLAB / Simulink joint simulation platform is built. The Simulink input variables are longitudinal speed, acceleration, engine crankshaft speed, torque converter output torque, transmission ratio and relative distance between the two vehicles, and the output is throttle opening and brake master cylinder pressure. To comprehensively evaluate the effect of the new control strategy, the embodiment of the present invention compares the multi-objective adaptive cruise control strategy method based on fuzzy MPC proposed by the method of the present invention with the traditional model predictive control (MPC) method.

[0153] The simulation verification includes the cruise control condition, the front vehicle cutting out condition, the adjacent vehicle cutting in condition, and the complex road condition. Figure 4 , Figure 5 , Figure 6 , Figure 7 In the simulation verification, the multi-objective adaptive cruise control strategy method based on fuzzy MPC of the present invention is marked as ImprovedMPC and compared with the traditional model predictive control MPC method.

[0154] Depend on Figure 4 It can be seen that under the cruise control condition, the adaptive cruise control algorithm can effectively track the preset target speed and the control effect is good. Figure 5 It can be seen that when the front vehicle cuts out, the adaptive cruise control algorithm can effectively regulate the vehicle speed change and smoothly transition from the following state to the constant speed cruise state. Figure 6 It can be seen that when the vehicle next to you cuts in, the control algorithm can make the acceleration change smoother, the vehicle can decelerate in advance, and the braking intensity can be reduced. Figure 7It can be seen that under complex road conditions, the control algorithm can help reduce 9.25% of trip emissions, bring better driving comfort, and will not cause significant travel time delays. It can be seen that the system of the present invention has significant advantages in adaptive following control under complex traffic conditions, and provides a new solution for improving the safety, comfort and fuel economy of the ACC system.

[0155] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A multi-objective adaptive cruise control method based on fuzzy model predictive control, characterized in that: The following steps are involved: Collect real-time operation data of vehicles; Inputting the real-time running data of the vehicle into an improved vehicle longitudinal dynamics model to predict the future motion state of the vehicle, wherein the improved vehicle longitudinal dynamics model selects the actual vehicle distance between the two vehicles, the speed of the vehicle, the relative speed of the two vehicles, the acceleration of the vehicle and the rate of change of the acceleration of the vehicle as state variables, and considers the interference of the acceleration of the preceding vehicle on the motion of the vehicle; Based on the output of the vehicle longitudinal dynamics model, the ACC system following prediction model is used to make real-time predictions and corrections on the following behavior, and dynamically adjust the following control strategy. The fuzzy variable weight adjustment strategy is used to dynamically adjust the weight of the objective function of the model predictive control according to the feedback correction of the vehicle's motion state after the ACC system's following prediction model is corrected; Model predictive control is performed based on the objective function and taking multiple constraints into consideration to generate accurate acceleration and deceleration control parameters, which are applied to the controlled vehicle to achieve multi-objective adaptive cruise control.

2. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 1, characterized in that: The improved vehicle longitudinal dynamics model establishes the vehicle state equation based on the relative motion relationship: Where x(k) is the state variable of the system, x(k)=[Δd(k),Δv(k),a f (k),a j (k)] T ; u(k) is the control variable, i.e. the desired acceleration of the vehicle a des ; w(k) is the disturbance, i.e. the acceleration of the front vehicle a p ; T s is the sampling time; T L is the time constant of the inertia link; a f is the vehicle acceleration, a j is the jerk of the vehicle, Δv is the relative speed, d is the distance between the two vehicles, and Δd is the deviation between the actual vehicle distance and the ideal vehicle distance.

3. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 1, characterized in that: The objective function and multiple constraint conditions are constructed based on multiple performance indicators, and the multiple performance indicators include a vehicle following safety performance indicator, a driving speed constraint, a dynamic following performance indicator, a driving comfort performance indicator, and a fuel economy performance indicator.

4. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 3, characterized in that: For the following vehicle safety performance index, the distance between the preceding vehicle and the vehicle is set to be greater than the preset safe following distance: d(k)≥d c Among them, d(k) is the distance between the two vehicles at time k, d c To ensure safe following distance; For driving speed constraints, the following constraints are imposed on the speed of the vehicle during driving: v fmin ≤v f (k)≤v fmax Among them, v fmin ,v fmax are the minimum and maximum vehicle speeds, respectively, and v f (k) is the vehicle speed; For the dynamic followability performance index, the relative speed and inter-vehicle distance deviation of the front and rear vehicles are set to converge to zero, that is, the two vehicles are in a relatively static state: Δd(k)→0 v rel (k)→0 Where Δd(k) is the inter-vehicle distance error, Δd(k) = d(k) - d des , d des is the ideal vehicle spacing, v rel (k) is the relative speed between the leading and following vehicles; For the driving comfort performance index, select the vehicle acceleration a f The absolute value of (k) and the absolute value of the acceleration change rate jerk(k) are used as quantitative characterization parameters of comfort, and the comfort optimization objective is defined as: my|a f (k)| min|jerk(k)| At the same time, the following constraints are imposed on acceleration and acceleration rate: a fmin ≤a f (k)≤a fmax jerk min ≤jerk(k)≤jerk max Among them, a fmin 、a fmax are the minimum and maximum values ​​of acceleration, jerk min , jerk max are the minimum and maximum values ​​of the acceleration rate of change respectively; For fuel economy, the driving comfort performance index is used to indirectly characterize the fuel economy level, and the expected acceleration value range is restricted: in min ≤u(k)≤u max Among them, u max ,u min are the maximum and minimum values ​​of the expected acceleration respectively, and u(k) is the expected acceleration of the vehicle.

5. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 2, characterized in that: The feedback correction ACC system following prediction model dynamically adjusts the acceleration and deceleration of the vehicle according to the prediction error between the real-time state and the predicted state, wherein the prediction error between the actual state and the predicted state of the ACC system at time k is defined as follows: x p (k|k-1)=A·x(k-1)+B·u(k-1)+G·w(k-1) Among them, x p (k|k-1) is the prediction error, x(k-1) is the system state quantity at time k-1, u(k-1) is the system control quantity at time k-1, and w(k-1) is the external interference quantity at time k-1.

6. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 1, characterized in that: The fuzzy variable weight adjustment strategy includes a fuzzy rule base, a fuzzy reasoning engine and online weight adjustment, wherein the fuzzy rule base is constructed based on multiple performance indicators and defines control strategies under different traffic scenarios; the fuzzy reasoning engine is responsible for judging the complexity of the driving environment according to the current motion state of the vehicle during real-time driving, and determining the weight adjustment plan according to the fuzzy rule base; the online weight adjustment uses dynamic fuzzy logic to update the weight coefficient of the model predictive control in real time according to actual working conditions.

7. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 1, characterized in that: The objective function is expressed as: Among them, N p is the prediction time domain, u is the control variable, Q and R are weight coefficients, y p is the vehicle trajectory, y ref is the target reference trajectory.

8. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 7, characterized in that: An exponential decay function is introduced into the target reference trajectory to make each component of the output vector smoothly approach zero, which can be expressed as: y ref (k+i)=Λ i ·y(k) Among them, y(k) is the target vector of the performance indicator to be optimized, Λ=diag(λ Δd ,l vrel ,l af ,l jerk ) λ Δd ,λ vrel ,λ af ,λ jerk are the deviation Δd between the actual vehicle spacing and the ideal vehicle spacing, the relative speed v of the front and rear vehicles rel , vehicle acceleration a f and the reference trajectory coefficient of the ego-vehicle acceleration change rate jerk, with a value range of (0, 1).

9. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 8, characterized in that: The weight coefficient Q is expressed as: Among them, q Δd is the weight factor of the inter-vehicle distance error, is the weight factor of the relative speed of the two vehicles, is the weight factor of the vehicle’s acceleration, q jerk is the weight factor of the acceleration change rate of the vehicle; The selection of the weight coefficient R affects the expected acceleration of the model predictive control decision output.

10. The multi-objective adaptive cruise control method based on fuzzy model predictive control according to claim 9, characterized in that: The weight coefficient and the reference trajectory coefficient are determined as follows: Among them, L f is the weight coefficient of followability, L c is the weight coefficient of comfort and fuel economy.

Images