Vehicle gear shifting point real-time solving method based on model prediction for electric multi-gear automobile

Abstract

The invention discloses a model prediction-based vehicle gear shifting point real-time solving method for an electric multi-gear automobile, and belongs to the technical field of automobile control. A vehicle economy nonlinear equation is established, a gear shifting strategy of a vehicle is formulated by taking the vehicle speed and an accelerator pedal as input parameters, and vehicle economy optimization of the vehicle is completed by utilizing optimization of a gear shifting point; establishing a cost function of the vehicle economical gear shifting strategy, requiring to find a group of optimal gear sequences in a prediction time domain, and solving the established cost function through an LM algorithm; weight coefficients in the cost function are optimized and adjusted, an optimal combination of the weight coefficients in the cost function is obtained by adopting a genetic algorithm, and an optimal solution of a vehicle real-time gear shifting strategy objective function is obtained, so that the optimal optimization effect of the whole vehicle performance is achieved.

Description

Technical Field

[0001] The invention discloses a real-time solution method for vehicle shift points of an electric multi-gear vehicle based on model prediction, and belongs to the technical field of vehicle control. Background Art

[0002] The real-time solution of the vehicle's shift point has a significant impact on the overall performance of the vehicle. The vehicle's shift strategy plays a vital role in predictive cruise control by optimizing the shift timing, improving power performance, improving driving comfort and enhancing system stability. In addition, the real-time solution of the vehicle's shift point also provides the driver with a more intelligent, efficient and comfortable driving experience. The shift point during vehicle driving is a typical nonlinear solution problem. Choosing a suitable nonlinear solution method is crucial to quickly, accurately and stably solve the vehicle's driving gear.

[0003] At present, the methods for solving nonlinear equations are mainly divided into the following methods: least squares method, gradient descent method, Gauss-Newton method, and Levenberg-Marquardt (LM) method. The least squares method is stable in iteration and usually has a unique solution, but it may cause excessive deviation for a certain abnormal value, and the amount of calculation is large when solving; the gradient descent method converges slowly in high-dimensional space, and may require a large number of iterations to converge, and the gradient descent method can easily fall into the local optimal solution instead of the global optimal solution; the Gauss-Newton method solves the equation by solving the first-order derivative information to approximate the second-order derivative, which reduces a lot of calculation, but it may have too many iterations in the iterative process; the LM method can achieve the effect of combining the advantages of the Gauss-Newton method and the gradient descent method by modifying the parameters in the calculation, and the coefficient value introduced can be adjusted according to the actual situation. Summary of the invention

[0004] In order to achieve real-time solution of vehicle shift points, improve vehicle power performance, improve economy, enhance driving experience, and adapt the vehicle to different road conditions and driving needs. The present invention proposes a real-time solution method for vehicle shift points based on model prediction for electric multi-gear vehicles. The technical solution provided by the present invention is:

[0005] A real-time solution method for vehicle shift points of an electric multi-gear vehicle based on model prediction is characterized by comprising the following steps:

[0006] Establish a nonlinear equation for vehicle economy, formulate the vehicle's shift strategy by taking the vehicle speed and accelerator pedal as input parameters, and optimize the vehicle's economy by optimizing the shift point;

[0007] Establishing the cost function of the vehicle economic shifting strategy requires finding a set of optimal gear sequences in the prediction time domain and solving the established cost function through the LM algorithm;

[0008] The weight coefficients in the cost function are optimized and adjusted, and the optimal combination of the weight coefficients in the cost function is obtained by using a genetic algorithm to obtain the optimal solution of the vehicle's real-time shifting strategy objective function, thereby achieving the best optimization effect on the vehicle's overall performance.

[0009] In some of these examples, the establishment of the nonlinear equation for the economy of the whole vehicle needs to consider the factors that affect the economy of the vehicle, which are mainly related to the characteristics of the drive motor and the efficiency of each component in the transmission system. The motor efficiency is mainly related to the motor speed and motor torque, which are related to the vehicle speed and the accelerator pedal opening. Therefore, the vehicle speed and the pedal opening are used as inputs of the nonlinear equation for the economy of the whole vehicle.

[0010] In some of the examples, the nonlinear equation of vehicle economy is characterized in that the prediction model of the input parameters vehicle speed and accelerator pedal opening is as follows:

[0011] v(t 0 +t k )=Mar(v 0 , t k );t k <t p ;t k , t p ∈N

[0012]

[0013] Among them, Mar(v, t) is the driving state predicted at time t based on the initial speed v using the Markov prediction model. p is the prediction step length; τ(t k ) is t k The accelerator pedal opening at time t, ξ is the accelerator pedal opening change rate, and κ is the decay rate. Assuming that ξ>0, ξ decays linearly to 0 in the prediction time domain, then κ=ξ / 2t p .

[0014] In some of the examples, the nonlinear equation of vehicle economy is as follows:

[0015]

[0016] Where P is the power consumed by the driving motor during the vehicle's driving process, which can be expressed as an equation related to the driving motor torque and the driving motor speed f(T e , n e ), where T eis the relationship between the output torque of the drive motor and the accelerator pedal opening, the speed of the drive motor, and the gear position. η is the functional relationship between the efficiency of the drive motor and the output torque of the drive motor and the speed of the drive motor, which can be obtained by interpolation. ρ(i) represents the equation of the automatic transmission ratio with respect to the gear position of the automatic transmission, and θ is a constant. It can be seen that the energy consumption of an electric vehicle during driving is a nonlinear equation with respect to the accelerator pedal opening τ, the vehicle speed v, and the gear position i of the automatic transmission.

[0017] In some of these examples, the cost function of the vehicle shift real-time optimization control strategy is as follows:

[0018] J=min(F E +θ D F D +θ c F c )

[0019] where θ D ,θ c is the weight coefficient, F E 、F D 、F c They are economic cost, power cost and gear shift cost function respectively. The economic cost function F of the vehicle gear shift strategy E The energy consumed by the vehicle during driving in the predicted time domain is used as a reference value for calculation. Therefore, the calculation formula is as follows:

[0020]

[0021] Dynamic cost function F of vehicle shift strategy D The calculation is based on the product of the difference between the current drive motor output torque and the rated torque and the current speed during the vehicle's driving in the prediction time domain as a reference value. The calculation formula is as follows:

[0022]

[0023] in, Indicates the power potential of the vehicle in the current state, represents the power potential of the vehicle after the control action. The gear shift cost function is formulated to limit the frequent gear shifting of the automatic transmission, so the gear shift cost function is related to the control action The calculation formula is as follows:

[0024]

[0025] The real-time optimization problem of electric vehicle shift strategy can be described as: finding a set of optimal shift control sequences in the prediction time domain The target cost function J is minimized to improve the economy of the vehicle and optimize the performance of the whole vehicle while satisfying the vehicle's dynamics and drivability.

[0026] In some examples, the shift strategy cost function is a nonlinear equation. The shift strategy cost function is solved using the LM method. After the cost function is discretized, formula (1-4) can be rewritten as:

[0027]

[0028] If it is assumed that the minimum cost function of the vehicle's driving condition before N-1 moments has been solved, the cost function at the current moment can be adjusted to:

[0029] argmin{J(N)}=argmin{J(N-1)}+argmin{ΔJ(N)}

[0030] Therefore, we only need to discuss ΔJ(N). According to whether the gear shift action occurs and the difference between upshift and downshift operations, ΔJ(N) can be mainly divided into the following two situations:

[0031]

[0032] Among them, ψ is a constant related to the gear shifting action, so

[0033] ψ=θ C F C (k)

[0034] When there is no gear shifting in the automatic transmission, the cost function of the Nth stage of the optimization algorithm can be derived from the cost function of the N-1th stage. When there is a gear shifting in the automatic transmission, it is necessary to calculate all possible transmission ratios i * (N) to minimize the economic cost function J. The LM algorithm can be used to solve the minimum cost function when there is a gear shift in the transmission in formula (1-10).

[0035] In some examples, the different weight coefficients θ D ,θ C It will not only affect the economy, power and drivability of the shift strategy, but also the performance of the drive motor. Using genetic algorithm, the weight coefficient θ in the objective function is D ,θ C Optimize. Genetic algorithm is an optimization algorithm based on natural selection and genetics principles. By simulating the natural evolution process of organisms in the biological world, genetic genes are crossed, mutated, and selected, and finally the optimal solution in the population is selected.

[0036] The beneficial effects of the present invention are:

[0037] The present invention provides a real-time solution method for vehicle shift points based on model prediction for electric multi-gear vehicles. The present invention constructs a nonlinear equation for solving the vehicle's driving gear position, takes into account the timeliness requirements of the shift strategy, and integrates the solution conditions and advantages and disadvantages of various nonlinear solution methods. Finally, the LM method is used to solve the nonlinear equation for solving the gear position. When formulating the objective function, the genetic algorithm is used to solve the optimal objective function weight coefficient, and finally the online optimization of the gear position during vehicle driving is completed. The real-time solution of the vehicle's shift point can achieve the purpose of optimizing the performance of the entire vehicle.

[0038] The steps of the present invention are: establishing a nonlinear equation for the economy of the whole vehicle, formulating the vehicle's shifting strategy by taking the vehicle speed and the accelerator pedal as input parameters, and optimizing the shift point to optimize the economy of the whole vehicle; establishing a cost function for the vehicle's economy shifting strategy, requiring to find a set of optimal gear sequences in the prediction time domain, and solving the established cost function through the LM algorithm; optimizing and adjusting the weight coefficients in the cost function, using a genetic algorithm to obtain the optimal combination of the weight coefficients in the cost function, and obtaining the optimal solution of the vehicle's real-time shifting strategy objective function, thereby achieving the best optimization effect on the performance of the whole vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The present invention is a flow chart of a method for real-time solution of vehicle shift points based on model prediction for an electric multi-gear vehicle.

[0040] Figure 2 This is the genetic algorithm flow chart. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The present invention is further described in detail below so that those skilled in the art can implement it according to the description.

[0043] This embodiment provides a real-time solution method for vehicle shift points based on model prediction for electric multi-speed vehicles. Figure 1 , including the following steps:

[0044] Step 1: Establish the nonlinear equation of vehicle economy, establish the prediction model of vehicle speed and accelerator pedal opening, and make short-term prediction of the two parameters as the input of the nonlinear equation of vehicle economy. Analyze the energy consumption of multi-gear electric vehicles during driving, and construct the vehicle economy problem into a nonlinear equation about accelerator pedal sticking, vehicle speed and automatic transmission gear.

[0045] Step 2: Establish a cost function for real-time optimization of vehicle shift points. The basic driving function of the vehicle must be guaranteed during the driving process. Therefore, we cannot only focus on the economy of the vehicle and ignore the power of the vehicle. At the same time, in order to limit the frequent shifting caused by the vehicle in pursuit of optimal economy, consider adding relevant parameter information to the objective function to complete the overall optimization of vehicle performance. Therefore, the cost function of the real-time optimization control strategy of vehicle shifting can be set. See formula 1-4

[0046] Step 3: The cost function of the real-time optimization solution of the vehicle shift point is a nonlinear equation, and keyi uses the LM method to solve it. Since the real-time optimization algorithm for shifting involves the vehicle's driving conditions at the previous moment, and whether the transmission performs a shift action has a certain influence on the solution of the cost function. Therefore, when the automatic transmission does not shift, the N-stage cost function of the optimization algorithm can be derived from the cost function of the N-1 stage. When the automatic transmission has a shift action, the LM algorithm needs to be used to calculate all possible transmission ratios i * (N) to minimize the economic cost function J. The transmission ratio solved by the LM algorithm is only the theoretical optimal solution of the vehicle automatic transmission ratio, which can be any value. However, the actual vehicle automatic transmission ratio is discrete. Therefore, an actual transmission ratio closest to the theoretical transmission ratio is selected as the final optimized transmission ratio gear, and then decide whether to shift gears.

[0047] Step 4: Use genetic algorithm to adjust the weight coefficient θ in the objective function D ,θ C In this example, the optimization objective function of the genetic algorithm is set as:

[0048]

[0049] in, is the average torque change of the driving motor in the working cycle, Δshift is the change in the number of gear shifts in the working cycle, and ΔP is the change in the power consumption in the working cycle. k is a dimensionless constant. For the convenience of calculation, the above variables are set as dimensionless constants. In this example, the genetic algorithm related parameters are given as follows:

[0050]

Claims

1. A real-time solution method for vehicle shift points based on model prediction for electric multi-gear vehicles, characterized by: A nonlinear equation for the economy of the whole vehicle is established. The vehicle's shifting strategy is formulated by taking the vehicle speed and the accelerator pedal as input parameters. The shift point optimization is used to optimize the economy of the whole vehicle. An optimal set of gear sequences is found in the prediction time domain. The weight coefficients in the established cost function are optimized and adjusted through the LM algorithm. The genetic algorithm is used to obtain the optimal combination of the weight coefficients in the cost function, and the optimal solution of the objective function of the vehicle's real-time shifting strategy is obtained, thereby achieving the best optimization effect on the performance of the whole vehicle.

2. A real-time solution method for vehicle shift points based on model prediction for electric multi-gear vehicles according to claim 1, characterized in that: The establishment of the nonlinear equation for the economy of the whole vehicle needs to consider the factors that affect the economy of the vehicle, which are mainly related to the characteristics of the drive motor and the efficiency of each component in the transmission system. The motor efficiency is mainly related to the motor speed and motor torque, and is also related to the vehicle speed and the accelerator pedal opening. Therefore, the vehicle speed and the pedal opening are used as inputs of the nonlinear equation for the economy of the whole vehicle.

3. The method for real-time solution of vehicle shift points based on model prediction for electric multi-gear vehicles according to claim 2, characterized in that: The prediction model of the input parameters vehicle speed and accelerator pedal opening is as follows: v(t0+t k )=Mar(v o t k );t k <t p ;t k ,t p ∈N Among them, Mar(v, t) is the driving state predicted at time t based on the initial speed v using the Markov prediction model. p is the prediction step length; τ(t k ) is t k is the accelerator pedal opening at the moment, ξ is the accelerator pedal opening change rate, and κ is the attenuation rate.

4. The method for real-time solution of vehicle shift points based on model prediction for electric multi-gear vehicles according to claim 3, characterized in that: When the accelerator pedal opening change rate ξ>0, ξ decays linearly to 0 in the prediction time domain, then κ=ξ / 2t p The power consumed by the driving motor during the vehicle's driving process can be expressed as an equation related to the driving motor torque and the driving motor speed. Among them, T e is the relationship between the output torque of the drive motor and the accelerator pedal opening, the speed of the drive motor and the gear position. η is the functional relationship between the drive motor efficiency and the output torque of the drive motor and the speed of the drive motor. ρ(i) represents the equation of the automatic transmission gear ratio with respect to the gear position of the automatic transmission. θ is a constant.

5. According to the method for real-time solution of vehicle shift point based on model prediction for electric multi-gear vehicle in claim 2, the cost function of the real-time optimization control strategy of vehicle shift is: J=min(F E +θ D F D +θ c F c ) where θ D ,θ c is the weight coefficient, F E 、F D 、F c They are economic cost, power cost and shift cost functions respectively.

6. A real-time solution method for vehicle shift points based on model prediction for electric multi-gear vehicles according to claim 5, characterized in that: When there is no gear shifting in the automatic transmission, the cost function of the Nth stage of the optimization algorithm can be derived from the cost function of the N-1th stage. When there is a gear shifting in the automatic transmission, all possible transmission ratios i need to be calculated. * (N) to minimize the economic cost function.

7. The method for real-time solution of vehicle shift points based on model prediction for electric multi-gear vehicles according to claim 1, characterized in that: In some examples, a genetic algorithm is used to adjust the weight coefficient θ in the objective function. D ,θ C Optimization is carried out by simulating the natural evolution process of organisms in the biological world, crossing, mutating and selecting genetic genes, and finally selecting the optimal solution in the population.

Images