Speed planning method based on space-time hybrid model predictive control

Abstract

The invention discloses a speed planning method based on space-time hybrid model predictive control, and aims to improve the energy-saving performance of a vehicle on roads with different slopes, and the method specifically comprises the following steps: (1) combining a time dimension and a space dimension, building a space-time hybrid model predictive control architecture, replacing a conventional MPC predictive time domain with a space domain distance, and building a space-time hybrid model predictive control architecture; the road gradient information serves as external disturbance quantity input of a control system, optimization solution of an optimal control sequence is carried out, inconvenience caused by the problem that time and space dimensions are not unified is avoided, and meanwhile model simplification is facilitated; (2) performing linearization processing on the established model based on a kinetic energy theorem, combining a speed quadratic term of vehicle kinetic energy with a speed quadratic term of air resistance, and eliminating a corresponding nonlinear part of a state-space equation; according to the method, the relation between the adjacent states of the vehicle is constructed in the spatial domain, and the state-space equation is subjected to linear processing, so that the solving difficulty is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of vehicle energy saving, and specifically relates to a speed planning method based on time-space hybrid model predictive control. Background Art

[0002] As a key technology for unmanned driving, speed planning has a significant impact on the fuel economy of vehicles. At present, commonly used speed planning methods mainly include PID control, LQR control and model predictive control (MPC). Among them, MPC, as a type of controller that solves constrained optimization problems in a rolling manner, has outstanding advantages in solving speed planning problems. However, the traditional MPC research on speed planning is complex when considering the road slope. Considering the combination of time dimension and space dimension, a time-space hybrid model predictive control architecture is established, and the spatial domain distance is used instead of the traditional MPC prediction time domain. The road slope information is used as the external disturbance input of the control system to optimize the optimal control sequence and solve it, avoiding the inconvenience caused by the problem of inconsistent time and space dimensions, and at the same time facilitating the simplification of the model. Based on the kinetic energy theorem, the model is linearized, and the quadratic term of the vehicle kinetic energy is combined with the quadratic term of the air resistance to eliminate the corresponding nonlinear part of the state space equation. The present invention constructs the connection between adjacent states of the vehicle in the spatial domain, and linearizes the state space equation, which ultimately reduces the difficulty of solving.

[0003] Some existing patents, such as the invention patent with patent number CN116409338A, proposes a speed planning algorithm for automatic driving based on slope, and the invention patent with patent number CN115520188A proposes an energy-saving vehicle speed planning method. The former calculates the current slope based on the information perceived during the vehicle's driving process, and the latter obtains the current vehicle information and the slope information and spacing information of multiple sub-sections of the road ahead, and generates the first planned speed of the vehicle in each of the sub-sections. Both do not consider the slope as an external disturbance input into the system, and the model is relatively complex, making it difficult to ensure the real-time solution. Summary of the invention

[0004] The present invention aims to improve the energy-saving performance of vehicles on roads with different slope conditions. A speed planning method based on time-space hybrid model predictive control is proposed. This method combines the time dimension and the space dimension to establish a time-space hybrid model predictive control architecture, using the spatial domain distance instead of the traditional MPC prediction time domain, and the road slope information as the external disturbance input of the control system. Then, based on the kinetic energy theorem, the established model is linearized, the speed quadratic term of the vehicle kinetic energy is combined with the speed quadratic term of the air resistance, and the corresponding nonlinear part of the state space equation is eliminated. The connection between adjacent states of the vehicle is constructed in the spatial domain, and the state space equation is linearized to reduce the difficulty of model solution.

[0005] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions:

[0006] A speed planning method based on spatiotemporal hybrid model predictive control, characterized in that it comprises the following steps:

[0007] S1: Combine the time dimension and space dimension to establish a spatiotemporal hybrid model predictive control architecture:

[0008] At each control sampling moment, the high-precision map and GPS obtain the precise position of the vehicle, thereby obtaining the road slope information of a certain spatial distance ahead. The spatial domain distance is used instead of the traditional MPC prediction time domain. The road slope information is used as the external disturbance input of the control system to optimize and solve the optimal control sequence.

[0009] S2: Linearize the established model based on the kinetic energy theorem;

[0010] The overall distance S in the prediction space domain is discretized into p stages of equal length, and the distance of a single stage is defined as:

[0011]

[0012] Assume that the vehicle's acceleration is constant within any single stage distance:

[0013]

[0014] Where a(s) is the acceleration of the vehicle in a single stage distance, v(s+Δs) is the final velocity of the vehicle in a single stage distance, and v(s) is the initial velocity of the vehicle in a single stage distance;

[0015] Within this spatial domain step, the vehicle driving force-driving resistance balance equation is organized into the following form:

[0016]

[0017] Among them, δ is the conversion coefficient of the vehicle's rotational mass, and the right side of the equation is the work done by the driving force, the work done by the rolling resistance, the work done by the air resistance, and the work done by the slope resistance;

[0018] The quadratic term of the vehicle's kinetic energy is combined with the quadratic term of the air resistance to eliminate the corresponding nonlinear part of the state space equation. The vehicle's kinetic energy is used as the state quantity, the wheel-end driving force is used as the control quantity, and the trigonometric function value corresponding to the slope is used as the disturbance quantity. The energy balance equation of vehicle motion is transformed into a discrete state space equation as follows:

[0019] x(k+1)=Ax(k)+B u u(k)+B d d(k) (4)

[0020] The state variable x(k), control variable u(k), and disturbance variable d(k) are: x(k) = E(k), u(k) = F t (k)

[0021]

[0022] The parameter matrix corresponding to the state space equation of MPC is:

[0023]

[0024] Avoid adding nonlinear factors that increase the computational burden to the objective function, replace the vehicle fuel consumption in the objective function with the work done by the driving force, and optimize the objective function:

[0025]

[0026] Among them, E(k) is the actual kinetic energy of the vehicle, E ref (k) is the kinetic energy corresponding to the reference vehicle speed, β1 and β2 are weight factors;

[0027]

[0028] Among them, X k =[x(k+1|k) T ,x(k+2|k) T ,x(k+3|k) T ,…,x(k+p|k) T ] T ,

[0029] U k =[u(k+1|k) T ,u(k+2|k) T ,u(k+3|k) T ,…,u(k+p|k) T ] T ,

[0030] D k =[d(k+1|k) T ,d(k+2|k) T ,d(k+3|k) T ,…,d(k+p|k) T ] T ;

[0031] Convert the reference speed set by the driver into a fixed reference kinetic energy value to generate a series of state quantity reference values:

[0032] R k =[r(k+1|k) T,r(k+2|k) T ,r(k+3|k) T ,…,r(k+p|k) T ] T (8)

[0033] The optimization objective function composed of system state quantity and control quantity is in the form of:

[0034]

[0035] Among them, Q, R are the matrix forms of weight factors;

[0036] Substituting (7) into (9) we obtain:

[0037]

[0038] Define the intermediate variable G:

[0039] G=Ψx(k)-R k +ΦD k (11)

[0040] Substituting (11) into (10), we obtain:

[0041]

[0042] Ignore the constant term G T QG, the optimization objective function is expressed in standard quadratic programming form:

[0043]

[0044] Constraints:

[0045]

[0046] Among them, v min is the minimum speed required for the road, v min ≤v≤v max , v max For roads requiring maximum speed, SOC min is the battery SOC lower limit, SOC max is the upper limit of battery SOC, n e,min is the lower limit of engine speed, n e,max is the upper limit of engine speed, T e,min is the lower limit of engine torque, T e,max is the upper limit of engine torque, n m,min is the lower limit of motor speed, n m,max is the upper limit of motor speed, T m,min is the lower limit of motor torque, T m,max is the upper limit of the motor torque, Z is an integer set, gmin The minimum gear of AMT, g max is the maximum gear position of AMT, ω min is the lower limit of the power system speed, ω max is the upper limit of the power system speed, u g It is the gear shift command, -1 means the gear is down, 0 means the gear is unchanged, and 1 means the gear is up.

[0047] Compared with the prior art, the advantages of the present invention are:

[0048] 1. The present invention combines the time dimension and the space dimension to establish a time-space hybrid model predictive control architecture, uses the spatial domain distance instead of the traditional MPC prediction time domain, and uses the road slope information as the external disturbance input of the control system to optimize and solve the optimal control sequence, avoiding the inconvenience caused by the inconsistency of the time and space dimensions, and at the same time facilitating the simplification of the model.

[0049] 2. Based on the kinetic energy theorem, the established model is linearized, and the quadratic term of the vehicle's kinetic energy is combined with the quadratic term of the air resistance to eliminate the corresponding nonlinear part of the state space equation. The present invention constructs the connection between adjacent states of the vehicle in the spatial domain and linearizes the state space equation to reduce the difficulty of solving. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0051] Figure 1 It is a flow chart of a speed planning method based on spatiotemporal hybrid model predictive control according to the present invention;

[0052] Figure 2 A schematic diagram of a spatiotemporal hybrid model prediction architecture in a speed planning method based on spatiotemporal hybrid model predictive control according to the present invention; DETAILED DESCRIPTION

[0053] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0054] The present invention will be further described below in conjunction with the accompanying drawings.

[0055] See also Figure 1 The present invention provides a speed planning method based on spatiotemporal hybrid model predictive control, which specifically includes the following steps:

[0056] S1: See Figure 2 , combining the time dimension and the space dimension, a spatiotemporal hybrid model predictive control architecture is established:

[0057] At each control sampling moment, the high-precision map and GPS obtain the precise position of the vehicle, thereby obtaining the road slope information of a certain spatial distance ahead. The spatial domain distance is used instead of the traditional MPC prediction time domain. The road slope information is used as the external disturbance input of the control system to optimize and solve the optimal control sequence.

[0058] S2: Linearize the established model based on the kinetic energy theorem;

[0059] The overall distance S in the prediction space domain is discretized into p equal-length stages, and the distance of a single stage is defined as:

[0060]

[0061] Assume that the vehicle's acceleration is constant within any single stage distance:

[0062]

[0063] Where a(s) is the acceleration of the vehicle in a single stage distance, v(s+Δs) is the final velocity of the vehicle in a single stage distance, and v(s) is the initial velocity of the vehicle in a single stage distance;

[0064] Within this spatial domain step, the vehicle driving force-driving resistance balance equation is organized into the following form:

[0065]

[0066] Among them, δ is the conversion coefficient of the vehicle's rotational mass, and the right side of the equation is the work done by the driving force, the work done by the rolling resistance, the work done by the air resistance, and the work done by the slope resistance;

[0067] The quadratic term of the vehicle's kinetic energy is combined with the quadratic term of the air resistance to eliminate the corresponding nonlinear part of the state space equation. The vehicle's kinetic energy is used as the state quantity, the wheel-end driving force is used as the control quantity, and the trigonometric function value corresponding to the slope is used as the disturbance quantity. The energy balance equation of vehicle motion is transformed into a discrete state space equation as follows:

[0068] x(k+1)=Ax(k)+B u u(k)+B d d(k) (4)

[0069] The state variable x(k), control variable u(k), and disturbance variable d(k) are: x(k) = E(k), u(k) = F t (k)

[0070]

[0071] The parameter matrix corresponding to the state space equation of MPC is:

[0072]

[0073] Avoid adding nonlinear factors that increase the computational burden to the objective function, replace the vehicle fuel consumption in the objective function with the work done by the driving force, and optimize the objective function:

[0074]

[0075] Among them, E(k) is the actual kinetic energy of the vehicle, E ref (k) is the kinetic energy corresponding to the reference vehicle speed, β1 and β2 are weight factors;

[0076]

[0077] Among them, X k =[x(k+1|k) T ,x(k+2|k) T ,x(k+3|k) T ,…,x(k+p|k) T ] T ,

[0078] U k =[u(k+1|k) T ,u(k+2|k) T ,u(k+3|k) T ,…,u(k+p|k) T ] T ,

[0079] D k =[d(k+1|k) T ,d(k+2|k) T ,d(k+3|k) T ,…,d(k+p|k) T ] T ;

[0080] Convert the reference speed set by the driver into a fixed reference kinetic energy value to generate a series of state quantity reference values:

[0081] R k =[r(k+1|k) T ,r(k+2|k) T ,r(k+3|k) T ,…,r(k+p|k) T ] T (8)

[0082] The optimization objective function composed of system state quantity and control quantity is in the form of:

[0083]

[0084] Among them, Q, R are the matrix forms of weight factors;

[0085] Substituting (7) into (9) we obtain:

[0086] J(U k )=(Ψx(k)+ΘU k +ΦD k -R k ) T Q(Ψx(k)+ΘU k +ΦD k -R k )+U k T RU k (10)

[0087] Define the intermediate variable G:

[0088] G=Ψx(k)-R k +ΦD k (11)

[0089] Substituting (11) into (10), we obtain:

[0090]

[0091] Ignore the constant term G T QG, the optimization objective function is expressed in standard quadratic programming form:

[0092]

[0093] Constraints:

[0094]

[0095] Among them, v min is the minimum speed required for the road, v min ≤v≤v max , v max For roads requiring maximum speed, SOC min is the battery SOC lower limit, SOC max is the upper limit of battery SOC, n e,min is the lower limit of engine speed, n e,max is the upper limit of engine speed, T e,min is the lower limit of engine torque, T e,max is the upper limit of engine torque, n m,min is the lower limit of motor speed, n m,maxis the upper limit of motor speed, T m,min is the lower limit of motor torque, T m,max is the upper limit of the motor torque, Z is an integer set, g min The minimum gear of AMT, g max is the maximum gear position of AMT, ω min is the lower limit of the power system speed, ω max is the upper limit of the power system speed, u g It is the gear shift command, -1 means the gear is down, 0 means the gear is unchanged, and 1 means the gear is up.

[0096] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A speed planning method based on spatiotemporal hybrid model predictive control, characterized in that: The following steps are involved: S1: Combine the time dimension and space dimension to establish a spatiotemporal hybrid model predictive control architecture: At each control sampling moment, the high-precision map and GPS obtain the precise position of the vehicle, thereby obtaining the road slope information of a certain spatial distance ahead. The spatial domain distance is used instead of the traditional MPC prediction time domain. The road slope information is used as the external disturbance input of the control system to optimize and solve the optimal control sequence. S2: Linearize the established model based on the kinetic energy theorem.

2. The speed planning method based on spatiotemporal hybrid model predictive control according to claim 1 is characterized in that: The specific process of linearizing the model based on the kinetic energy theorem in step S2 is as follows: The overall distance S in the prediction space domain is discretized into p equal-length stages, and the distance of a single stage is defined as: Assume that the vehicle's acceleration is constant within any single stage distance: Where a(s) is the acceleration of the vehicle in a single stage distance, v(s+Δs) is the final velocity of the vehicle in a single stage distance, and v(s) is the initial velocity of the vehicle in a single stage distance; Within this spatial domain step, the vehicle driving force-driving resistance balance equation is organized into the following form: Among them, δ is the conversion coefficient of the vehicle's rotational mass, and the right side of the equation is the work done by the driving force, the work done by the rolling resistance, and the air Work done by resistance and work done by slope resistance; The quadratic term of the vehicle's kinetic energy is combined with the quadratic term of the air resistance to eliminate the corresponding nonlinear part of the state space equation. The vehicle's kinetic energy is used as the state quantity, the wheel-end driving force is used as the control quantity, and the trigonometric function value corresponding to the slope is used as the disturbance quantity. The energy balance equation of vehicle motion is transformed into a discrete state space equation as follows: x(k+1)=Ax(k)+B u u(k)+B d d(k) (4) The state variable x(k), control variable u(k), and disturbance variable d(k) are: x(k) = E(k), u(k) = F t (k) The parameter matrix corresponding to the state space equation of MPC is: Avoid adding nonlinear factors that increase the computational burden to the objective function, replace the vehicle fuel consumption in the objective function with the work done by the driving force, and optimize the objective function: Among them, E(k) is the actual kinetic energy of the vehicle, E ref (k) is the kinetic energy corresponding to the reference vehicle speed, β1 and β2 are weight factors; X(k)=ψx(k)+ΘU k +ΦD k Among them, X k =[x(k+1|k) T ,x(k+2|k) T ,x(k+3|k) T ,…,x(k+p|k) T ] T , YOU k =[u(k+1|k) T ,u(k+2|k) T ,u(k+3|k) T ,…,u(k+p|k) T ] T , D k =[d(k+1|k) T ,d(k+2|k) T ,d(k+3|k) T ,…,d(k+p|k) T ] T ; Convert the reference speed set by the driver into a fixed reference kinetic energy value to generate a series of state quantity reference values: R k =[r(k+1|k) T ,r(k+2|k) T ,r(k+3|k) T ,…,r(k+p|k) T ] T (8) The optimization objective function composed of system state quantity and control quantity is in the form of: Among them, Q, R are the matrix forms of weight factors; Substituting (7) into (9) we obtain: Define the intermediate variable G: G=Ψx(k)-R k +ΦD k (11) Substituting (11) into (10), we obtain: Ignore the constant term G T QG, the optimization objective function is expressed in standard quadratic programming form: Constraints: Among them, v min is the minimum speed required for the road, v min ≤v≤v max , v max For roads requiring maximum speed, SOC min is the battery SOC lower limit, SOC max is the upper limit of battery SOC, n e,min is the lower limit of engine speed, n e,max is the upper limit of engine speed, T e,min is the lower limit of engine torque, T e,max is the upper limit of engine torque, n m,min is the lower limit of motor speed, n m,max is the upper limit of motor speed, T m,min is the lower limit of motor torque, T m,max is the upper limit of the motor torque, Z is an integer set, g min The minimum gear of AMT, g max is the maximum gear position of AMT, ω min is the lower limit of the power system speed, ω max is the upper limit of the power system speed, u g It is the gear shift command, -1 means the gear is down, 0 means the gear is unchanged, and 1 means the gear is up.

Images