| Hybrid technology is an effective way to achieve energy saving and emission reduction of vehicles,has been widely used.As the key technology of hybrid electric powertrain,energy management should not only coordinate the characteristics of different power sources to improve the energy-saving effect,but also face different road environment and drivers to improve the adaptability.The existing calibration methods for energy management strategy are mainly based on test adjustment,which have some disadvantages,such as long adjustment period,poor adaptability to environment and drivers,and limited energy-saving effect.In recent years,with the rapid development of digital twinning technology,the problems existing in the traditional calibration methods can be effectively solved by constructing the digital twinning model of powertrain and operating environment and carrying out the calibration work in the virtual environment.However,with the increasing complexity of energy management strategy for hybrid vehicles,many parameters need to be calibrated,which leads to an exponential increase in the complexity of calibration,the determined calibration parameters are difficult to adapt to the variable driving cycles and different driving styles,which is not conducive to deeply tap the potential of energy-saving powertrain.To solve the above-mentioned problems,this paper presents a calibration parameter optimization method considering the driving cycle and driving style,aiming at the energy management strategy of the power-split hybrid electric powertrain,the optimization of multi-dimension calibration parameters and its adaptability to different driving cycles and driving style were studied.The full text of the work can be summarized as follows:(1)A Driver-vehicle-road forward simulation model for driving cycle and driving style is established.First of all,the key components of the powertrain are modeled using theoretical and numerical modeling methods;Secondly,the data of light commercial vehicle driving in a given area are collected,by fusing kernel density estimation,metropolis-hastings sampling method and genetic algorithm,the optimal driving model is established,and the high-precision comprehensive representative and local representative driving cycles are obtained;and then,a longitudinal driver model with driving style is established by using the elastic network regression method and adaptive network fuzzy inference system with real driver test data as input;Finally,the key component model,driving mode and driver model are coupled to the driver-vehicle-road forward simulation model,and the model accuracy is verified by experiments.(2)Aiming at the problems of numerous calibration parameters and unclear optimization directions for the energy management strategy of a real vehicle,an analysis of the factors affecting the fuel consumption of powertrain was carried out considering different driving cycles and driving styles.Based on the energy flow characteristics of a power-split hybrid electric powertrain,a comprehensive fuel consumption model for the entire vehicle was established by integrating mechanisms and numerical modeling methods,The variation rules of powertrain energy consumption under different driving cycles and driving styles are revealed.Through quantitative analysis of the main and interactive effects of various fuel consumption influencing factors,a reference basis is provided for the selection and optimization of calibration parameters for actual vehicle energy management strategies.(3)Aiming at the problem of numerous and interrelated calibration parameters for real vehicle energy management strategies,a Cooperative Coevolution Particle Swarm Optimization(CCPSO)algorithm was designed to achieve in-depth optimization of calibration parameters.Firstly,based on the main effect analysis results of powertrain fuel consumption influencing factors,the actual vehicle energy management strategy is simplified to determine calibration parameters significantly related to comprehensive fuel consumption,and the feasible region of calibration parameters is determined based on the characteristics of powertrain components;Based on this,the calibration parameter optimization problem is decomposed into multiple sub problems according to the analysis results of the interaction effects of the powertrain fuel consumption influencing factors,and a parallel optimization model based on chaotic particle swarm optimization algorithm is established for each sub problem;Then,the dynamic allocation method of computing resources is introduced into the optimization model,enabling subpopulations to automatically adjust the size and increase or decrease particles based on particle fitness and population change trends,achieving real-time optimal allocation of computing resources and collaborative coevolution of various subpopulations.Using the CCPSO algorithm,the optimal calibration parameter combinations corresponding to different driving cycles and driving styles are obtained.At the same time,quantitative results show that the proposed CCPSO algorithm achieves a fuel saving effect of 20.90% under typical driving cycles and aggressive driving styles.(4)An Optimal Calibration Parameter Adaptive(OCPA)energy management strategy for differentiated driving cycles and driving styles is proposed.Firstly,according to the timing characteristics of driving cycles and the slow changing characteristics of driving style,an offline identification model for driving cycles and driving style is established based on relevant data using a bi-directional long-term memory network and a support vector machine,respectively;Then,select an appropriate identification time window and establish an online identification model;Further,a calibration parameter library is established based on the optimal calibration parameter combination,and the identification model is fused to achieve online adaptation of the calibration parameters of the energy management strategy.The effectiveness of the strategy is verified through simulation;Finally,a hardware in the loop test platform based on d SPACE/Simulator is established,fully verifying the real-time control effect of the OCPA strategy.The results show that the proposed OCPA strategy ensures real-time performance while reducing the overall fuel consumption by 4.93%～15.14% in different scenarios compared to the original strategy,achieving better economy. |
PDF Full Text Request