Energy management control method of hybrid electric vehicle based on DMPC

Abstract

The present invention provides a hybrid vehicle energy management control method based on DMPC, comprising the following steps: S1. collecting driving information of the hybrid vehicle, wherein the driving information includes the position, speed and torque information of the current vehicle in the driving queue; S2. constructing a nonlinear fleet control model and predicting the predicted required torque of the hybrid vehicle at the next moment based on the driving information; S3. constructing a required torque optimization model and determining the optimal required torque based on the predicted required torque; S4. constructing an EF-DQL energy distribution model, inputting the optimal required torque into the EF-DQL energy distribution model to determine the optimal torque of the motor and the optimal torque of the engine of the hybrid vehicle under the optimal required torque, and controlling the operation of the power system of the hybrid vehicle based on the optimal torque of the engine and the optimal torque of the motor. Through the above method, the energy of the vehicle can be analyzed and managed according to the queue status of the vehicle, and the reasonable torque distribution of the hybrid vehicle can be effectively improved, thereby effectively improving the fuel economy of the hybrid vehicle.

Description

Technical Field

[0001] The present invention relates to a hybrid electric vehicle control method, and in particular to a hybrid electric vehicle energy management control method based on DMPC. Background Art

[0002] The emergence of intelligent connected vehicle technology has provided a new solution for real-time energy management and optimization of hybrid vehicle platoons. If cars on the road can communicate with surrounding vehicles and adjust their driving status in real time based on information from neighboring vehicles, forming an automated platoon of multiple vehicles traveling in the same lane and direction, it will be safer and more economical than a single vehicle.

[0003] However, in the existing technology, the control strategies for the fuel economy of hybrid vehicles are all based on the state of the hybrid vehicle itself, including battery status, motor efficiency and driver's operating status as control factors to control the hybrid vehicle. However, cars on the road are not independent individuals. The driving of the vehicle will be affected by information such as the driving speed, acceleration and position of the vehicles in front and behind. Therefore, the existing methods cannot accurately analyze and control the fuel economy of hybrid vehicles.

[0004] Therefore, in order to solve the above technical problems, a new technical solution is proposed to solve them. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a hybrid vehicle energy management control method based on DMPC, which can analyze and manage the energy of the vehicle according to the vehicle's queue status, and can effectively improve the reasonable torque distribution of the hybrid vehicle, thereby effectively improving the fuel economy of the hybrid vehicle.

[0006] The present invention provides a hybrid electric vehicle energy management control method based on DMPC, comprising the following steps:

[0007] S1. Collecting driving information of a hybrid vehicle, the driving information including the current position, speed and torque information of the vehicle in the driving queue;

[0008] S2. Build a nonlinear fleet control model and predict the next required torque of the hybrid electric vehicle based on driving information.

[0009] S3. Construct a demand torque optimization model and determine the optimal demand torque based on the predicted demand torque;

[0010] S4. Construct an EF-DQL energy distribution model, input the optimal required torque into the EF-DQL energy distribution model to determine the optimal torque of the motor and the optimal torque of the engine of the hybrid vehicle under the optimal required torque, and control the operation of the power system of the hybrid vehicle based on the optimal torque of the engine and the optimal torque of the motor.

[0011] Furthermore, step S2 specifically includes:

[0012] S21. Constructing a nonlinear fleet control model:

[0013]

[0014] Where: i represents the i-th hybrid vehicle in the driving queue, s i (t) is the displacement of the i-th vehicle, which is used to represent the position of the i-th vehicle in the driving queue, v i (t) represents the speed of the i-th vehicle, T i (t) represents the actual torque of the i-th vehicle, u i (t) is the expected torque of the i-th vehicle, m i is the mass of vehicle i, g is the acceleration due to gravity, f i is the rolling damping coefficient, τ i is the time lag coefficient, A is the frontal area of ​​vehicle i, r i is the wheel radius of vehicle i, η i is the mechanical efficiency of the transmission system of vehicle i, c i-car is the air damping coefficient, · represents the derivative;

[0015] S22. Discretize the nonlinear fleet control model: Let the state variable x i (t)=[s i (t),v i (t),T i (t)] T , the control quantity is u i (t), the output is y i (t)=[s i (t),v i (t)] T , the discrete step length is Δt, and formula (1) is discretized into:

[0016]

[0017] Further, step S3 specifically includes:

[0018] Set the prediction step size to N p , then in each prediction time domain [t,t+N p ] Construct the prediction optimization objective function:

[0019]

[0020] The constraints are:

[0021]

[0022]

[0023]

[0024] Among them: i_pre (0|t),…,u i_pre (N p -1|t)] is the control variable to be optimized, s i_des (N p |t) and v i_des (N p |t) are the expected position and expected speed of vehicle i, Δs min and Δs max are the lower and upper limits of the displacement error, Δv min , and Δv max are the lower and upper limits of vehicle speed error, s i_des (t) and v i_des (t) are the expected position and speed respectively; a i (t) is acceleration; j i (t) is the rate of change of acceleration;

[0025] When the target optimization function reaches its minimum u i The first term of (t) is the optimal required torque And convert the optimal required torque into the optimal required power of the hybrid vehicle.

[0026] Furthermore, the cost function for vehicle i optimization is

[0027] g i (y i_pre (k|t),u i_pre (k|t),y i_ass (k|t),y j_ass (k|t)) is determined by the following formula:

[0028]

[0029] Where: Q i () is the expected state weight matrix, F i () is the assumed state weight matrix, R i () is the acceleration and deceleration penalty weight matrix of vehicle i, G i () is the control output weight of the neighboring vehicles of vehicle I, yi_ass (k|t) is the assumed control output of vehicle i at time t+k, y i_des (k|t) is the expected output of the prediction terminal.

[0030] Furthermore, step S4 specifically includes: constructing an EF-DQL energy allocation model, which has a reward function L with the goal of minimizing total fuel consumption and maintaining battery SOC balance:

[0031]

[0032] Where: r(s p ,a p )=-1×[β1(f e +f m )+β2(abs(SOC-SOC ref )];f e is the fuel consumption of the engine under the current state and action; f m is the fuel energy equivalent to electric energy; β1, β2 are weight coefficients; SOC ref is the reference value of the SOC of the power battery of vehicle i; wherein the state variables include the motor demand power P req and the remaining power SOC of the power battery. The action variables include the motor torque T m and battery energy and fuel consumption conversion coefficient S _eq ;

[0033] The optimal motor power requirement of the hybrid vehicle is determined based on the EF-DQL algorithm. The optimal engine power requirement is obtained by subtracting the optimal motor power requirement from the optimal power requirement. The optimal motor power requirement and the optimal engine power requirement are then converted into corresponding optimal torque requirements. The hybrid vehicle is controlled according to the optimal torque requirements.

[0034] In DQL, a deep neural network (DNN) is used to estimate the optimal Q value, where:

[0035] Q * (s t ,a t )≈Q(s t ,a t ;w t ),

[0036] Loss function L(w t ) is expressed as:

[0037]

[0038] Furthermore, the battery energy and fuel consumption conversion coefficient S is determined by the following formula _eq

[0039] in: is the average efficiency of the engine, is the average efficiency of the motor, and Indicates the average discharge and charge efficiency of the battery.

[0040] Furthermore, the electric energy equivalent fuel energy f is calculated according to the following formula m :

[0041] f m =S_eq·P m / (Hη m );

[0042] Where: H is the lower calorific value of fuel; η m is the motor efficiency, P m is the motor power of vehicle i.

[0043] The beneficial effects of the present invention are as follows: the energy of vehicles can be analyzed and managed according to the vehicle queue status, and the reasonable distribution of torque of hybrid vehicles can be effectively improved, thereby effectively improving the fuel economy of hybrid vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0045] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0046] The present invention is further described in detail below:

[0047] The present invention provides a hybrid electric vehicle energy management control method based on DMPC, comprising the following steps:

[0048] S1. Collecting driving information of a hybrid vehicle, the driving information including the current position, speed and torque information of the vehicle in the driving queue;

[0049] S2. Build a nonlinear fleet control model and predict the next required torque of the hybrid electric vehicle based on driving information.

[0050] S3. Construct a demand torque optimization model and determine the optimal demand torque based on the predicted demand torque;

[0051] S4. Construct an EF-DQL energy distribution model, input the optimal demand torque into the EF-DQL energy distribution model to determine the optimal torque of the motor and the optimal torque of the engine of the hybrid vehicle under the optimal demand torque, and control the operation of the power system of the hybrid vehicle based on the optimal torque of the engine and the optimal torque of the motor. Through the above method, the energy of the vehicle can be analyzed and managed according to the queue status of the vehicle, which can effectively improve the reasonable torque distribution of the hybrid vehicle, thereby effectively improving the fuel economy of the hybrid vehicle.

[0052] In this embodiment, step S2 specifically includes:

[0053] S21. Constructing a nonlinear fleet control model:

[0054]

[0055] Where: i represents the i-th hybrid vehicle in the driving queue, s i (t) is the displacement of the i-th vehicle, which is used to represent the position of the i-th vehicle in the driving queue, v i (t) represents the speed of the i-th vehicle, T i (t) represents the actual torque of the i-th vehicle, u i (t) is the expected torque of the i-th vehicle, m i is the mass of vehicle i, g is the acceleration due to gravity, f i is the rolling damping coefficient, τ i is the time lag coefficient, A is the frontal area of ​​vehicle i, r i is the wheel radius of vehicle i, η i is the mechanical efficiency of the transmission system of vehicle i, c i-car is the air damping coefficient, · represents the derivative;

[0056] S22. Discretize the nonlinear fleet control model: Let the state variable x i (t)=[s i (t),v i (t),T i (t)] T , the control quantity is u i (t), the output is y i (t)=[s i (t),v i (t)] T , the discrete step length is Δt, and formula (1) is discretized into:

[0057]

[0058] In this embodiment, step S3 specifically includes:

[0059] Set the prediction step size to N p , then in each prediction time domain [t,t+N p ] Construct the prediction optimization objective function:

[0060]

[0061] The constraints are:

[0062]

[0063]

[0064]

[0065] Among them: i_pre (0|t),…,u i_pre (N p -1|t)] is the control variable to be optimized, s i_des (N p |t) and v i_des (N p |t) are the expected position and expected speed of vehicle i, Δs min and Δs max are the lower and upper limits of the displacement error, Δv min , and Δv max are the lower and upper limits of vehicle speed error, s i_des (t) and v i_des (t) are the expected position and speed respectively; a i (t) is acceleration; j i (t) is the rate of change of acceleration;

[0066] in, In the above:

[0067] s i_ass (k+1|t): the assumed position of vehicle i at time t+k+1;

[0068] v i_ass (k+1|t): the assumed speed of vehicle i at time t+k+1;

[0069] T i_ass (k+1|t): the assumed torque of vehicle i at time t+k+1;

[0070] u i_ass (k|t): assumed control input for vehicle i at time t+k;

[0071] y i_ass (k|t): Assumed control output of vehicle i at time t+k

[0072] v i_pre (k+1|t): predicted speed of vehicle i at time t+k+1;

[0073] s i_pre (k+1|t): predicted position of vehicle i at time t+k+1;

[0074] v i_pre (k+1|t): predicted speed of vehicle i at time t+k+1;

[0075] T i_pre (k+1|t): predicted torque of vehicle i at time t+k+1;

[0076] u i_pre (k|t): predicted control input of vehicle i at time t+k;

[0077] y i_pre (k|t): The predicted control output of vehicle i at time t+k.

[0078] When the target optimization function reaches its minimum u i The first term of (t) is the optimal required torque And convert the optimal required torque into the optimal required power of the hybrid vehicle.

[0079] Where: Cost function for vehicle i optimization

[0080] g i (y i_pre (k|t),u i_pre (k|t),y i_ass (k|t),y j_ass (k|t))g i (y i_pre (k|t) is determined by the following formula:

[0081]

[0082] Where: Q i () is the expected state weight matrix, F i () is the assumed state weight matrix, R i () is the acceleration and deceleration penalty weight matrix of vehicle i, G i () is the control output weight of the neighboring vehicles of vehicle I, y i_ass (k|t) is the assumed control output of vehicle i at time t+k, y i_des (k|t) is the expected output of the prediction terminal.

[0083] In this embodiment, step S4 specifically includes: constructing an EF-DQL energy allocation model, which has a reward function L with the goal of minimizing total fuel consumption and maintaining battery SOC balance:

[0084]

[0085] Where: r(s p ,a p )=-1×[β1(f e +f m )+β2(abs(SOC-SOC ref )];f e is the fuel consumption of the engine under the current state and action; f m is the fuel energy equivalent to electric energy; β1, β2 are weight coefficients; SOC ref is the reference value of the SOC of the power battery of vehicle i; wherein the state variables include the motor demand power P req and the remaining power SOC of the power battery. The action variables include the motor torque T m and battery energy and fuel consumption conversion coefficient S _eq ;

[0086] The optimal motor power requirement of the hybrid vehicle is determined based on the EF-DQL algorithm. The optimal engine power requirement is obtained by subtracting the optimal motor power requirement from the optimal power requirement. The optimal motor power requirement and the optimal engine power requirement are then converted into corresponding optimal torque requirements. The hybrid vehicle is controlled according to the optimal torque requirements.

[0087] In DQL, a deep neural network (DNN) is used to estimate the optimal Q value, where:

[0088] Q * (s t ,a t )≈Q(s t ,a t ;w t ),

[0089] Loss function L(w t ) is expressed as:

[0090] The process of estimating the optimal Q value is a prior art and will not be described in detail here.

[0091] Specifically: The battery energy and fuel consumption conversion coefficient S is determined by the following formula _eq

[0092] in: is the average efficiency of the engine, is the average efficiency of the motor, and Indicates the average discharge and charge efficiency of the battery.

[0093] The equivalent fuel energy of electric energy f is calculated according to the following formula m :

[0094] f m =S_eq·P m / (Hη m );

[0095] Where: H is the lower calorific value of fuel; η m is the motor efficiency, P m is the motor power of vehicle i.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A hybrid electric vehicle energy management control method based on DMPC, characterized by: The following steps are involved: S1. Collecting driving information of a hybrid vehicle, the driving information including the current position, speed and torque information of the vehicle in the driving queue; S2. Build a nonlinear fleet control model and predict the next required torque of the hybrid electric vehicle based on driving information. S3. Construct a demand torque optimization model and determine the optimal demand torque based on the predicted demand torque; S4. Construct an EF-DQL energy allocation model, input the optimal demand torque into the EF-DQL energy allocation model to determine the optimal motor torque and the optimal engine torque of the hybrid vehicle under the optimal demand torque, and control the power system operation of the hybrid vehicle based on the optimal engine torque and the optimal motor torque; Wherein: Step S3 specifically includes: Set the prediction step size to N p , then in each prediction time domain [t,t+N p ] Construct the prediction optimization objective function: The constraints are: Among them: i_pre (0|t),…,u i_pre (N p -1|t)] is the control variable to be optimized, s i_des (N p |t) and v i_des (N p |t) are the expected position and expected speed of vehicle i, Δs min and Δs max are the lower and upper limits of the displacement error, Δv min , and Δv max are the lower and upper limits of vehicle speed error, s i_des (t) and v i_des (t) are the expected position and speed respectively; a i (t) is acceleration; j i (t) is the rate of change of acceleration; When the target optimization function reaches its minimum u i The first term of (t) is the optimal required torque and converting the optimal required torque into the optimal required power of the hybrid vehicle; Cost function for vehicle i optimization g i (y i_pre (k|t),u i_pre (k|t),y i_ass (k|t),y j_ass (k|t)) is determined by the following formula: Where: Q i () is the expected state weight matrix, F i () is the assumed state weight matrix, R i () is the acceleration and deceleration penalty weight matrix of vehicle i, G i () is the control output weight of the neighboring vehicles of vehicle I, y i_ass (k|t) is the assumed control output of vehicle i at time t+k, y i_des (k|t) is the expected output of the prediction terminal.

2. The hybrid electric vehicle energy management control method based on DMPC according to claim 1 is characterized in that: Step S2 specifically includes: S21. Constructing a nonlinear fleet control model: Where: i represents the i-th hybrid vehicle in the driving queue, s i (t) is the displacement of the i-th vehicle, which is used to represent the position of the i-th vehicle in the driving queue, v i (t) represents the speed of the i-th vehicle, T i (t) represents the actual torque of the i-th vehicle, u i (t) is the expected torque of the i-th vehicle, m i is the mass of vehicle i, g is the acceleration due to gravity, f i is the rolling damping coefficient, τ i is the time lag coefficient, A is the frontal area of ​​vehicle i, r i is the wheel radius of vehicle i, η i is the mechanical efficiency of the transmission system of vehicle i, c i-car is the air damping coefficient, represents the derivation; S22. Discretize the nonlinear fleet control model: Let the state variable x i (t)=[s i (t),v i (t),T i (t)] T , the control quantity is u i (t), the output is y i (t)=[s i (t),v i (t)] T , the discrete step length is Δt, and formula (1) is discretized into:

3. The hybrid electric vehicle energy management control method based on DMPC according to claim 1 is characterized in that: Step S4 specifically includes: constructing an EF-DQL energy allocation model, which has a reward function L with the goal of minimizing total fuel consumption and maintaining battery SOC balance: Where: r(s p ,a p )=-1×[β1(f e +f m )+β2(abs(SOC-SOC ref )];f e is the fuel consumption of the engine under the current state and action; f m is the fuel energy equivalent to electric energy; β1, β2 are weight coefficients; SOC ref is the reference value of the SOC of the power battery of vehicle i; wherein the state variables include the motor demand power P req and the remaining power SOC of the power battery. The action variables include the motor torque T m and battery energy and fuel consumption conversion coefficient S _eq ; The optimal motor power requirement of the hybrid vehicle is determined based on the EF-DQL algorithm. The optimal engine power requirement is obtained by subtracting the optimal motor power requirement from the optimal power requirement. The optimal motor power requirement and the optimal engine power requirement are then converted into corresponding optimal torque requirements. The hybrid vehicle is controlled according to the optimal torque requirements. In EF-DQL, a deep neural network (DNN) is used to estimate the optimal Q value, where: Q * (s t ,a t )≈Q(s t ,a t ;w t ), Loss function L(w t ) is expressed as:

4. The hybrid electric vehicle energy management control method based on DMPC according to claim 3 is characterized by: The battery energy and fuel consumption conversion coefficient S is determined by the following formula _eq in: is the average efficiency of the engine, is the average efficiency of the motor, and Indicates the average discharge and charge efficiency of the battery.

5. The hybrid electric vehicle energy management control method based on DMPC according to claim 4 is characterized in that: The equivalent fuel energy of electric energy f is calculated according to the following formula m : f m =S_eq·P m / (Hη m ); Where: H is the lower calorific value of fuel; η m is the motor efficiency, P m is the motor power of vehicle i.

Images