阅读说明：本技术 基于反馈优化的车辆节能预测自适应巡航控制方法和装置 (Vehicle energy-saving prediction adaptive cruise control method and device based on feedback optimization ) 是由 边有钢 李崇康 胡满江 秦兆博 秦洪懋 王晓伟 秦晓辉 谢国涛 徐彪 丁荣军 于 2021-08-19 设计创作，主要内容包括：本发明公开了一种基于反馈优化的车辆节能预测自适应巡航控制方法核装置,该方法包括：步骤S1,获取自车预测状态、前车预测状态和后车预测状态；步骤S2,判断当前时刻是否为反馈增益切换时刻,若是,则进入步骤S3,反之,则沿用上一时刻的反馈增益,计算上位控制输入；步骤S3,构建经济优化问题,求解最优反馈增益,计算最优上位控制输入,并将最优控制输入序列对应的自车状态作为下一时刻自身预测状态,并返回步骤S2。本发明可在给定的反馈增益范围内,选取能耗最优的值,并通过设计反馈增益切换时间保证跟踪稳定性。(The invention discloses a vehicle energy-saving prediction adaptive cruise control method and a device based on feedback optimization, wherein the method comprises the following steps: step S1, acquiring a predicted state of the vehicle, a predicted state of the front vehicle and a predicted state of the rear vehicle; step S2, judging whether the current time is the feedback gain switching time, if yes, entering step S3, otherwise, continuing to use the feedback gain of the previous time and calculating the upper control input; and S3, constructing an economic optimization problem, solving the optimal feedback gain, calculating the optimal upper control input, taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment, and returning to the step S2. The invention can select the value with the optimal energy consumption within the given feedback gain range, and ensure the tracking stability by designing the feedback gain switching time.)

1. A vehicle energy-saving prediction adaptive cruise control method based on feedback optimization is characterized by comprising the following steps:

step S1, acquiring a predicted state of the vehicle, a predicted state of the front vehicle and a predicted state of the rear vehicle;

step S2, judging whether the current time is the feedback gain switching time, if yes, entering step S3, otherwise, continuing to use the feedback gain of the previous time and calculating the upper control input;

and S3, constructing an economic optimization problem, solving the optimal feedback gain, calculating the optimal upper control input, taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment, and returning to the step S2.

2. The feedback optimization-based vehicle energy-saving prediction adaptive cruise control method according to claim 1, wherein the constructing an economic optimization problem of step S3 comprises:

linear time-varying feedback gain control input u (t) described with function (11):

wherein K (t) is a time-varying feedback gain;is the tracking error vector of the self vehicle.

3. The feedback optimization-based vehicle energy-saving prediction adaptive cruise control method according to claim 1 or 2, characterized in that the cost function of constructing the economic optimization problem of step S3 comprises the unit displacement oil consumption l (x) per sampling time described by (18)^p(k | t)), wherein E_C(x^p(k | t)) is the corresponding x^pThe fuel consumption rate of (k | t), which is described as formula (6); p_d(x^p(k | t)) is the corresponding x^p(k | t) is described as equation (7), constructing an economic optimization problem is described as equation (13), and satisfying constraint equation (14) -equation (17);

x^p(0|t)＝x(t) (14)

in the formula, x^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t, v^p(k | t) is the speed of the own vehicle at the predicted time t + k at the current time t, and theta₀、θ₁And theta₂Is a constant coefficient, m is the mass of the vehicle, a^p(k | t) is the acceleration of the own vehicle at the predicted time t + k at the current time t,predicting the lumped air resistance coefficient of the self-vehicle at the t + k moment at the current moment t, wherein f is a rolling resistance coefficient, g is a gravity acceleration, and theta^p(k | t) is the predicted road gradient of the host vehicle at time t + k at the current time t, η_TTo mechanical system efficiency N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the predicted feedback gain for position, velocity and acceleration at time t, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t)，k_v(t)，k_a(t)]，k_p(t)、k_v(t) and k_a(t) are respectively the position, speed and acceleration feedback gains of the vehicle at the moment t,andfeedback gain and control input constraints, respectively.

4. The feedback-optimization-based vehicle energy-saving prediction adaptive cruise control method according to claim 3, characterized in that the interval pi of the feedback gain switching time of step S2 is set to equation (19):

wherein alpha and beta are constants satisfying 0< beta < alpha, and epsilon is represented by formula (20), wherein P is_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

wherein, lambda represents a matrix characteristic value, A is a system matrix, B is an input matrix, H is a headway matrix, tau is a vehicle longitudinal system time lag coefficient, and H is a headway:

5. the feedback optimization-based vehicle energy-saving prediction adaptive cruise control method according to claim 1 or 2, characterized by the electric energy consumed at each sampling instant described by the cost function (37) of step S3 for constructing an economic optimization problem, wherein x is^p(k | t) is the current time tpredit of the vehicleThe state at the time t + k is measured,the energy consumption index of the self vehicle at the moment T + k is predicted at the current moment T, the total power is taken as the index, the index is described as an expression (27), delta T is a sampling interval, the economic optimization problem is constructed and described as an expression (31), and constraint expressions (32) - (36) are met;

x^p(0|t)＝x(t) (32)

wherein the content of the first and second substances,andthe driving efficiency of the hub motor is predicted when the vehicle predicts t + k at the current moment t,predicting the brake regeneration efficiency of the hub motor at the t + k moment at the current moment t for the self-vehicle according to a motor efficiency map;andthe output power of the motor of the self-vehicle at the predicted time t + k at the current time t is obtained,predicting the motor input power at the moment t + k for the current moment t of the vehicle, N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the predicted feedback gain for position, velocity and acceleration at the current time t, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t)，k_v(t)]，k_p(t)、k_vAnd (t) is the position and speed feedback gain of the vehicle at the current time t.

6. The feedback-optimization-based vehicle energy-saving prediction adaptive cruise control method according to claim 5, characterized in that the interval pi of the feedback gain switching time of step S2 is set to equation (19):

wherein alpha and beta are constants satisfying 0< beta < alpha, and epsilon is represented by formula (39), wherein P is_σ(t) is a symmetric positive solution of the linear matrix inequality set (40);

wherein λ represents a matrix eigenvalue, H is a headway matrix, and H ═ H,0]^TH is the headway

7. A vehicle energy-saving prediction adaptive cruise control apparatus based on feedback optimization, characterized by comprising:

a state acquisition unit for acquiring a predicted state of a host vehicle, a predicted state of a preceding vehicle, and a predicted state of a following vehicle; and

a host controller having:

a switching time judgment unit for judging whether the current time is a feedback gain switching time;

the first upper control input calculation unit is used for constructing an economic optimization problem, solving an optimal feedback gain, calculating optimal upper control input and taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment under the condition that the switching moment judgment unit judges that the self-vehicle state is positive;

and a second upper control input calculation unit configured to calculate an upper control input by using the feedback gain at the previous time when the switching time determination unit determines that the switching time is negative.

8. The feedback optimization-based vehicle energy-saving prediction adaptive cruise control apparatus according to claim 7, wherein the construction economy optimization problem of said first upper control input calculation unit includes:

linear time-varying feedback gain control input u (t) described with function (11):

wherein K (t) is a time-varying feedback gain;is the tracking error vector of the self vehicle.

9. The feedback-optimization-based vehicle energy-saving prediction adaptive cruise control apparatus according to claim 8, wherein an interval pi of feedback gain switching timings of said switching timing judgment unit is set to equation (19):

wherein α and β are constants satisfying 0< β < α.

10. The feedback-optimization-based vehicle energy-saving prediction adaptive cruise control apparatus according to claim 9, wherein the value of e in equation (19) is divided into the following two cases:

the cost function for constructing the economic optimization problem in step S3 includes unit displacement oil consumption l (x) at each sampling time described in (18)^p(k | t)), wherein E_C(x^p(k | t)) is the corresponding x^pThe fuel consumption rate of (k | t), which is described as formula (6); p_d(x^p(k | t)) isx^pThe drive system output power of (k | t) is described as equation (7), ε is shown as equation (20), where P_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

step S3, constructing cost function (37) of economic optimization problem describing consumed electric energy at each sampling moment, wherein x^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t, P_b(k | T) is the energy consumption index of the self vehicle at the current time T, predicted T + k, taken as the total power, and is described as equation (27), Δ T is the sampling interval, and ε is shown as equation (39), where P is_σ(t) is a symmetric positive solution of the linear matrix inequality set (40);

wherein λ represents a matrix eigenvalue, H is a headway matrix, and H ═ H,0]^TH is the headway;

Technical Field

The invention relates to the technical field of intelligent networked automobile assistance, in particular to a method and a device for energy-saving prediction adaptive cruise control of a vehicle based on feedback optimization.

Background

Driving assistance systems such as Adaptive Cruise Control (ACC) provide technical support for reducing energy consumption. Through V2X (Vehicle to event) communication, the Vehicle can adjust the self motion state to realize the optimal control of a specific target by using the prediction information of the front Vehicle.

The existing ACC system energy-saving control method comprehensively considers multiple targets such as track tracking, fuel consumption, riding comfort and the like, and reduces economic optimality. In addition, most of the existing methods can only ensure bounded stability of tracking errors, and cannot realize more accurate tracking effect. Therefore, an ACC method with optimal energy consumption and high tracking accuracy is urgently needed.

Disclosure of Invention

It is an object of the present invention to provide a feedback-optimization-based vehicle energy-saving prediction adaptive cruise control method and apparatus that overcomes or at least mitigates at least one of the above-identified deficiencies of the prior art.

In order to achieve the above object, the present invention provides a feedback optimization-based adaptive cruise control method for vehicle energy conservation prediction, comprising:

step S1, acquiring a predicted state of the vehicle, a predicted state of the front vehicle and a predicted state of the rear vehicle;

step S2, judging whether the current time is the feedback gain switching time, if yes, entering step S3, otherwise, continuing to use the feedback gain of the previous time and calculating the upper control input;

and S3, constructing an economic optimization problem, solving the optimal feedback gain, calculating the optimal upper control input, taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment, and returning to the step S2.

Further, the building of the economic optimization problem of step S3 includes:

linear time-varying feedback gain control input u (t) described with function (11):

wherein K (t) is a time-varying feedback gain;is the tracking error vector of the self vehicle.

Further, the cost function of the step S3 for constructing the economic optimization problem includes the unit displacement oil consumption l (x) at each sampling time described in (18)^p(k | t)), wherein E_C(x^p(k | t)) is the corresponding x^pThe fuel consumption rate of (k | t), which is described as formula (6); p_d(x^p(k | t)) is the corresponding x^p(k | t) is described as equation (7), constructing an economic optimization problem is described as equation (13), and satisfying constraint equation (14) -equation (17);

x^p(0|t)＝x(t) (14)

in the formula, x^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t, v^p(k | t) is the speed of the own vehicle at the predicted time t + k at the current time t, and theta₀、θ₁And theta₂Is a constant coefficient, m is the mass of the vehicle, a^p(k | t) is the acceleration of the own vehicle at the predicted time t + k at the current time t,predicting the lumped air resistance coefficient of the self-vehicle at the t + k moment at the current moment t, wherein f is a rolling resistance coefficient, g is a gravity acceleration, and theta^p(k | t) is the predicted road gradient of the host vehicle at time t + k at the current time t, η_TTo mechanical system efficiency N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the position and velocity at time tPredictive feedback gain of degree and acceleration, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t)，k_v(t)，k_a(t)]，k_p(t)、k_v(t) and k_a(t) are respectively the position, speed and acceleration feedback gains of the vehicle at the moment t,andfeedback gain and control input constraints, respectively.

Further, the interval pi of the feedback gain switching timing of step S2 is set to equation (19):

wherein α and β are 0<β<A constant of α, ε is represented by formula (20), wherein P_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

wherein, lambda represents a matrix characteristic value, A is a system matrix, B is an input matrix, H is a headway matrix, tau is a vehicle longitudinal system time lag coefficient, and H is a headway:

further, the cost function (37) of step S3 for constructing the economic optimization problem describes the consumed electric energy at each sampling moment, wherein x is^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t,the energy consumption index of the self vehicle at the moment T + k is predicted at the current moment T, the total power is taken as the index, the index is described as an expression (27), delta T is a sampling interval, the economic optimization problem is constructed and described as an expression (31), and constraint expressions (32) - (36) are met;

x^p(0|t)＝x(t) (32)

wherein the content of the first and second substances,andthe driving efficiency of the hub motor is predicted when the vehicle predicts t + k at the current moment t,predicting the brake regeneration efficiency of the hub motor at the t + k moment at the current moment t for the self-vehicle according to a motor efficiency map;andthe output power of the motor of the self-vehicle at the predicted time t + k at the current time t is obtained,predicting the motor input power at the moment t + k for the current moment t of the vehicle, N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the predicted feedback gain for position, velocity and acceleration at the current time t, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t),k_v(t)]，k_p(t)、k_vAnd (t) is the position and speed feedback gain of the vehicle at the current time t.

Further, the interval pi of the feedback gain switching timing of step S2 is set to equation (19):

wherein α and β are 0<β<A constant of α, ε is represented by formula (39), wherein P_σ(t) is a symmetric positive solution of the linear matrix inequality set (40);

wherein λ represents a matrix eigenvalue, H is a headway matrix, and H ═ H,0]^TH is the headway

The invention also provides a feedback optimization-based vehicle energy-saving prediction adaptive cruise control device, which comprises:

a state acquisition unit for acquiring a predicted state of a host vehicle, a predicted state of a preceding vehicle, and a predicted state of a following vehicle; and

a host controller having:

a switching time judgment unit for judging whether the current time is a feedback gain switching time;

the first upper control input calculation unit is used for constructing an economic optimization problem, solving an optimal feedback gain, calculating optimal upper control input and taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment under the condition that the switching moment judgment unit judges that the self-vehicle state is positive;

and a second upper control input calculation unit configured to calculate an upper control input by using the feedback gain at the previous time when the switching time determination unit determines that the switching time is negative.

Further, the construction economic optimization problem of the first upper control input computing unit includes:

linear time-varying feedback gain control input u (t) described with function (11):

wherein K (t) is a time-varying feedback gain;is the tracking error vector of the self vehicle.

Further, the interval pi of the feedback gain switching time of the switching time judgment unit is set to equation (19):

wherein α and β are constants satisfying 0< β < α.

Further, the value of epsilon in the formula (19) is divided into the following two cases:

the cost function for constructing the economic optimization problem in step S3 includes unit displacement oil consumption l (x) at each sampling time described in (18)^p(k | t)), wherein E_C(x^p(k | t)) is the corresponding x^pThe fuel consumption rate of (k | t), which is described as formula (6); p_d(x^p(k | t)) is the corresponding x^pThe drive system output power of (k | t) is described as equation (7), ε is shown as equation (20), where P_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

step S3, constructing cost function (37) of economic optimization problem describing consumed electric energy at each sampling moment, wherein x^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t, P_b(k | T) is the energy consumption index of the self vehicle at the current time T, predicted T + k, taken as the total power, and is described as equation (27), Δ T is the sampling interval, and ε is shown as equation (39), where P is_σ(t) is a symmetric positive solution of the linear matrix inequality set (40);

wherein λ represents a matrix eigenvalue, H is a headway matrix, and H ═ H,0]^TH is the headway;

due to the adoption of the technical scheme, the invention has the following advantages: by receiving the state information of the front vehicle and the rear vehicle, the optimal control input is solved by taking economy as a target, and higher tracking precision is ensured. The invention aims to overcome the reduction of specific performance caused by multi-objective optimization in the existing scheme, and provides a vehicle self-adaptive cruise longitudinal speed control method for optimizing linear feedback gain based on economic model predictive control. The invention can select the value with the optimal energy consumption within the given feedback gain range, and ensure the tracking stability by designing the feedback gain switching time.

Drawings

Fig. 1 is a schematic diagram of a vehicle energy-saving prediction adaptive cruise control method based on feedback optimization according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of the asymptotic convergence of the tracking error according to the embodiment of the present invention.

Detailed Description

The invention is described in detail below with reference to the figures and examples.

As shown in fig. 1 and fig. 2, the vehicle energy-saving prediction adaptive cruise control method based on feedback optimization provided by the embodiment of the invention comprises the following steps:

in step S1, the predicted state of the own vehicle, the predicted state of the preceding vehicle, and the predicted state of the following vehicle are acquired. The predicted state of the self vehicle, the predicted state of the front vehicle and the predicted state of the rear vehicle are obtained through a vehicle-mounted sensor and a V2X communication unit, and the predicted states mainly comprise the windward area A of the front vehicle_v，lAnd predicting the assumed state of the current time t + k of the front vehicleAssumed position of rear vehicle at current time t and predicted time t + kAnd the frontal area A of the rear vehicle_v，rWherein, in the step (A),respectively predicting an assumed position, an assumed speed and an assumed acceleration at the time t + k for the current time t of the preceding vehicle, N_pis the prediction time domain.

Of course, it should be noted that when the V2X communication unit fails, the above-mentioned assumed states of the preceding vehicle and the following vehicle can be obtained by the predictive equations.

The predicted state of the own vehicle, the predicted state of the preceding vehicle, and the predicted state of the following vehicle obtained in step S1 are input to the upper controller. The upper controller may set the initialization feedback gain to zero, use the current state of the host vehicle as an initial value of the predicted state, and predict the state of the vehicle in the zero-input predicted time domain based on the linear longitudinal kinematics model. The upper controller then implements steps S2 and S3.

And step S2, judging whether the current time is the feedback gain switching time, if so, entering step S3, otherwise, continuing to use the feedback gain at the previous time to calculate the upper control input.

And S3, constructing an economic optimization problem, solving the optimal feedback gain, calculating the optimal upper control input, taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment, and returning to the step S2.

The upper control input calculated in steps S2 and S3 is sent to the lower controller. And the lower controller calculates expected torque and controls the action of the actuating mechanism based on the vehicle longitudinal inverse kinematics model according to the upper control input.

In one embodiment, step S3 specifically includes:

step S31, modeling the vehicle linear longitudinal kinematics.

The present embodiment is described as equation (1) in consideration of the nonlinear longitudinal kinematics model of the fuel vehicle:

wherein p (t), v (t) and a (t) are the self-parking spaces respectivelyThe device, the speed and the acceleration, T (t) is driving torque and is obtained by a self vehicle information acquisition module; eta is the mechanical efficiency of the transmission system, r_wThe radius of a wheel, m is the mass of the vehicle, f is a rolling resistance coefficient, and tau is a time lag coefficient of a longitudinal system of the vehicle and is set based on a vehicle parameter identification result; g is the acceleration of gravity; theta (t) is a road gradient and is obtained through map information; t is_des(t) is the desired torque, which will be calculated by inverse kinematic equation (4); lumped coefficient of air resistance C_A(t) is calculated by the following formula (2):

wherein, C_DIs the coefficient of air resistance, ρ_aIs the density of air, A_vζ (t) is a resistance correction coefficient calculated by equation (3) for the windward area:

wherein A is_v,lFrontal area of the front vehicle, A_v,rIs the windward area of the rear vehicle,is the distance from the front vehicle,is the distance from the rear vehicle.

To obtain a linear model equivalent to equation (1), a feedback linearization strategy is employed:

where u (t) is the upper control input after linearization, it can be understood as the desired acceleration.

Obtaining a linear longitudinal kinematic model (5) for a parametric economic optimization problem:

wherein, x (t) ═ p (t), v (t), a (t)]^TIs the state vector of the vehicle at the time t.

And step S32, modeling the energy consumption rate.

In this embodiment, a Virginia rational power-based integrated fuel consumption model (VT-CPFM) and a fuel consumption rate E are adopted_C(x (t)) is:

wherein, theta₀、θ₁And theta₂Is a constant coefficient, P_d(t) is the drive system output power, which is described by equation (7):

in the formula, x^p(k | t) is a state where the own vehicle predicts the time t + k at the current time t, v^p(k | t) is the speed of the own vehicle at the predicted time t + k at the current time t, and theta₀、θ₁And theta₂Is a constant coefficient, m is the mass of the vehicle, a^p(k | t) is the acceleration of the own vehicle at the predicted time t + k at the current time t,predicting the lumped air resistance coefficient of the self-vehicle at the t + k moment at the current moment t, wherein f is the rolling resistance coefficient, and g is the gravity accelerationDegree, theta^p(k | t) is the predicted road gradient of the host vehicle at time t + k at the current time t, η_TIs the mechanical system efficiency.

Step S33, the target design is tracked.

The self-vehicle expects to maintain a set distance with the front vehicle and keep the same speed and acceleration, and can be described as an equation (8):

where d (t) is the desired spacing, which is described as equation (9) according to a constant-time-distance strategy:

d(t)＝d₀+hv(t) (9)

wherein d is₀The static distance is set as h is the headway time.

Tracking error vector described by the definition equation (10)

Wherein d (t) ═ d (t),0]^TIs the desired pitch vector.

In step S34, the feedback control input is switched.

Calculating a linear time-varying feedback gain control input u (t) using a function (11):

wherein k (t) ═ k_p(t),k_v(t),k_a(t)]，k_p(t)、k_v(t) and k_a(t) time-varying feedback gains for position, velocity and acceleration at time t, respectively, whose values are solved by the economic optimization problem in step S35, which satisfies the constraintsDetermined empirically.

The linear control input of the embodiment adopts the time-varying feedback gain as an optimization variable, so that the calculation efficiency can be improved.

And step S35, carrying out economic optimization control.

And (4) discretely solving control input according to an economic model predictive control theory. The continuous-time linear longitudinal kinematics model (5) is discretized and described as formula (12):

x(k+1)＝f(x(k),x_l(k),K(k)) (12)

where k is a discrete time instant.

Constructing an economic optimization problem, describing an equation (13), and satisfying a constraint equation (14) -equation (17), wherein the constraint equation (14) requires that the initial value of the prediction state is equal to the real state at the current moment; the constraint (15) is a kinematic constraint; constraints (16) and (17) are feedback gain and control input constraints.

x^p(0|t)＝x(t) (14)

In the formula, N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the predicted feedback gain for position, velocity and acceleration at time t, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t),k_v(t),k_a(t)]，k_p(t)、k_v(t) and k_a(t) are respectively the position, speed and acceleration feedback gains of the vehicle at the moment t,andfeedback gain and control input constraints are respectively, and superscript p represents a predicted value and is used for parameterization optimization.

Building economic optimization problemThe cost function of (18) includes the unit displacement oil consumption l (x) of each sampling time^p(k | t)), wherein E_C(x^p(k | t)) is the corresponding x^pThe fuel consumption rate of (k | t), which is described as formula (6); p_d(x^p(k | t)) is the corresponding x^pThe drive system output power of (k | t), described as equation (7);

solving result K^*(t) calculating an actual control input u from the function (11) as an actual control parameter at time t^*(t) and sends it to the lower controller. The lower controller calculates a lower control input, i.e., a desired torque, based on (4)And controls the action of the executing mechanism.

Table 1 below shows the selected fuel vehicle model and the parameter values of the ACC controller.

First, k (t) k is empirically predetermined_p(t),k_a(t),k_v(t)]Value range ofThen according toSolving the inequality group (21) to obtain a group P_σ(t) and calculating epsilon from equation (20) and finally determining the residence time pi from equation (19). The parameters of this example are selected as shown in table 1. In consideration of the conservatism of the tracking error convergence analysis process, in practical application, after the residence time is determined according to the formula (19), the residence time can be adjusted to balance the conservatism and the control performance.

TABLE 1

In the above embodiment, the interval pi of the feedback gain switching timing of step S2 is set to equation (19):

wherein α and β are 0<β<A constant of α, ε is represented by formula (20), wherein P_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

wherein, lambda represents a matrix characteristic value, A is a system matrix, B is an input matrix, H is a headway matrix, tau is a vehicle longitudinal system time lag coefficient, and H is a headway:

as shown in fig. 2, the residence time designed by the method can ensure that the polytype Lyapunov function is monotonically decreased when the feedback gain K (t) is kept unchanged, and the jump quantity has an upper bound epsilon when K (t) is switched; and tracking errorsThe norm decreases exponentially at the upper bound, and the tracking error is taken if and only if the time t approaches infinityConverge to zero. The closed-loop error system meeting the residence time condition converges to the zero point gradually, and has higher tracking precision.

The embodiment directly takes the energy consumption as a cost function, and the optimal economy can be achieved within a given feedback gain range.

In another embodiment, for the hub-driven electric vehicle application, the difference of the feedback-optimization-based vehicle energy-saving prediction adaptive cruise control method of the present invention is mainly embodied in step S3, which specifically includes:

step S31, modeling the vehicle linear longitudinal kinematics.

In the embodiment, the nonlinear longitudinal kinematic model of the hub-driven electric vehicle is considered as formula (22):

wherein, p (t) and v (t) are the position and the speed of the vehicle respectively; f_t(t) Total longitudinal Driving force, to be solved by feedback linearization relationship (23), F_f(t) and F_r(t) front and rear wheel drive forces, respectively; f_g(t)、F_w(t) and F_c(t) longitudinal component of gravity, air resistance and rolling resistance, respectively; m, f, g, θ (t) and C_AThe meaning of (t)As in the embodiment in which the nonlinear longitudinal kinematics model of the fuel vehicle is considered.

To obtain a linear model equivalent to equation (22), a feedback linearization strategy is employed:

F_t(t)＝mu(t)+F_g(t)+F_w(t)+F_c(t) (23)

where u (t) is the upper control input after linearization, it can be understood as acceleration.

Obtaining a linear longitudinal kinematics model (24) for a parametric economic optimization problem:

wherein, x (t) ═ p (t), v (t)]^TIs a self-vehicle state vector, A is a system matrix, B is an input matrix:

and step S32, modeling the energy consumption rate.

Calculating formula (26) taking into account wheel dynamics (25) and tire longitudinal force:

wherein the content of the first and second substances,or r, is a front or rear wheel index; j. the design is a square_wAs the moment of inertia of the wheel, is,is the longitudinal stiffness of the tire, r_wIs the radius of the wheel based onSetting a vehicle parameter identification result;is the tire slip ratio;is the wheel speed (rpm),f to be supplied from the upper layer model predictive controller in step S35 for the wheel driving torque (Nm)_t(t) and v (t) calculation.

x^p(k | t) is the state of the vehicle at the current time t, the predicted time t + k, and the energy consumption index of the vehicle at the current time t, the predicted time t + k is taken as the total powerIs described as equation (27), Δ T is the sampling interval;

wherein the content of the first and second substances,andthe driving efficiency of the hub motor is predicted when the vehicle predicts t + k at the current moment t,predicting the brake regeneration efficiency of the hub motor at the t + k moment at the current moment t for the self-vehicle according to a motor efficiency map;andare all fromThe vehicle predicts the output power of the motor at the moment t + k at the current moment t,predicting the motor input power of the self-vehicle at the t + k moment at the current moment t, wherein the motor input power does not act simultaneously, and the value isWherein the content of the first and second substances,andthe wheel driving torque and the rotating speed at the moment t + k are respectively predicted for the self vehicle at the current moment t.

Step S33, the target design is tracked.

The self vehicle expects to maintain a set distance from the front vehicle and keep the same speed, and can be described as formula (28):

where d (t) is the desired pitch, the description is the same as in the embodiment in which the non-linear longitudinal kinematics model of the fuel vehicle is considered.

Tracking error vector described by the definition (29)

Wherein d (t) ═ d (t),0]^TIs the desired pitch vector.

In step S34, the feedback control input is switched.

Calculating a linear time-varying feedback gain control input u (t) using a function (30):

wherein k (t) ═ k_p(t),k_v(t)]，k_p(t) and k_v(t) time-varying feedback gains for position and velocity, respectively, whose values are solved by the economic optimization problem in step S35, which satisfies the constraintsDetermined empirically.

The linear control input of the embodiment adopts the time-varying feedback gain as an optimization variable, so that the calculation efficiency can be improved.

And step S35, carrying out economic optimization control.

The economic optimization problem construction idea is the same as that in the embodiment considering the fuel vehicle nonlinear longitudinal kinematics model.

Problem of economic optimizationDescribed as equation (31), and satisfying constraints (32) to (36), the constraints (32) to (35) have the same meaning as in the embodiment considering the nonlinear longitudinal kinematic model of the fuel vehicle, and the constraint (36) is the in-wheel motor output/input power constraint:

x^p(0|t)＝x(t) (32)

in the formula, N_pTo predict the time domain, x^p(k +1| t) is a state where the own vehicle predicts the time t + k +1 at the current time t,predicting the assumed state of the preceding vehicle at the current time t at the time t + K, K^p(t) is the predicted feedback gain for position, velocity and acceleration at the current time t, u^p(k | t) is a higher control input of the vehicle at the current time t, which predicts the time t + k, and k (t) [ k ]_p(t),k_v(t)]，k_p(t)、k_v(t) is the position and speed feedback gain of the vehicle at the current moment t respectively,andrespectively the output and input constraints of the in-wheel motor.

Problem of economic optimizationThe cost function (37) of (a) describes the power consumed at each sampling instant, where x^p(k | T) is the state of the vehicle at the current time T and the predicted time T + k, and delta T is a sampling interval;

solving result K^*(t) calculating an actual control input u from the function (30) as an actual control parameter at time t^*(t) and sends it to the lower controller. The lower controller calculates a lower control input, i.e., torque, based on (23), (25), and (26)MomentAnd controls the action of the executing motor.

The following table 2 shows the parameters of the selected in-wheel motor vehicle model and the ACC controller.

TABLE 2

In the above embodiment, the interval pi of the feedback gain switching timing of step S2 is set to equation (38):

wherein α and β are 0<β<A constant of α, ε is represented by formula (39), wherein P_σ(t) is a symmetric positive solution of the linear matrix inequality set (40);

wherein λ represents a matrix eigenvalue, H is a headway matrix, and H ═ H,0]^TAnd h is the headway.

The closed-loop error system meeting the residence time condition converges to the zero point gradually, and has higher tracking precision.

The embodiment of the invention also provides a vehicle energy-saving prediction adaptive cruise control device based on feedback optimization, which comprises a state acquisition unit and a superior controller, wherein the state acquisition unit is used for acquiring the predicted state of the self vehicle, the predicted state of the front vehicle and the predicted state of the rear vehicle.

The upper controller has a switching time judgment unit, a first upper control input calculation unit, and a second upper control input calculation unit, wherein: the switching time judging unit is used for judging whether the current time is the feedback gain switching time. And the first upper control input calculation unit is used for constructing an economic optimization problem, solving an optimal feedback gain, calculating optimal upper control input and taking the self-vehicle state corresponding to the optimal control input sequence as the self-prediction state at the next moment under the condition that the switching moment judgment unit judges that the switching moment judgment unit is positive. And a second upper control input calculation unit configured to calculate an upper control input by using the feedback gain at the previous time when the switching time determination unit determines that the switching time is negative.

In one embodiment, the building economic optimization problem of the first upper control input computing unit includes:

linear time-varying feedback gain control input u (t) described with function (11):

wherein K (t) is a time-varying feedback gain;is the tracking error vector of the self vehicle.

In one embodiment, the interval pi of the feedback gain switching time of the switching time judgment unit is set to equation (19):

wherein α and β are 0<β<A constant of α, ε is represented by formula (20), wherein P_σ(t) is a symmetric positive definite solution of the linear matrix inequality set (21);

wherein, lambda represents a matrix characteristic value, A is a system matrix, B is an input matrix, H is a vehicle head time distance matrix, and tau is a vehicle longitudinal system time lag coefficient:

finally, it should be pointed out that: the above examples are only for illustrating the technical solutions of the present invention, and are not limited thereto. Those of ordinary skill in the art will understand that: modifications can be made to the technical solutions described in the foregoing embodiments, or some technical features may be equivalently replaced; such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.