Font Size: a A A

Energy Management Of Hybrid Electric Vehicle Based On Driving Cycle Recognition And Prediction

Posted on:2020-10-23Degree:DoctorType:Dissertation
Country:ChinaCandidate:D Y QiuFull Text:PDF
GTID:1362330602466390Subject:Vehicle Engineering
Abstract/Summary:PDF Full Text Request
As the brain of hybrid electric vehicle,energy management strategy has always been the focus of research in various automotive enterprises and scientific research institutes.In the current mainstream energy management strategies,both rule-based and equivalent fuel consumption minimization strategies have some limitations,such as poor adaptability to driving cycles and difficulty in determining the optimal control parameters values.In the application of energy management strategy based on global optimization algorithm,it is necessary to predict all driving cycles in the future,which leads to poor feasibility in actual control.Aiming at the above problems,the hybrid electric vehicle based on P2 configuration is taken as the research object,and improving the fuel economy of the vehicle is taken as the research purpose.The research of energy management strategy with driving cycle recognition and driving cycle prediction were carried out respectively.The research contents and achievements are as follows:1)A driving cycle recognition algorithm based on particle swarm optimized support vector machine is proposed.In view of the limitations in theoretical application of current driving cycle recognition algorithms,such as the network structure settings in neural network theory;the initial value of clustering algorithm has a great influence on clustering results,which leads to the algorithm easily falling into local optimum and the membership function of fuzzy controller is mostly selected according to experience,only repeated debugging can improve the recognition accuracy,etc.In this paper,the support vector machine algorithm in machine learning theory is applied to the establishment of driving cycle recognition model.At the same time,in order to further improve the recognition accuracy,particle swarm optimization algorithm is used to optimize the core parameters of the support vector machine algorithm.The off-line modeling results show that the accuracy of the proposed algorithm based on particle swarm optimized support vector machine is significantly improved compared with the off-line driving cycle recognition model based on neural network and non-optimized support vector machine algorithm.2)For the optimization of equivalent fuel consumption minimization strategy is a discontinuous and non-differentiable bilevel multi-objective optimization problem,a double-loop multi-objective particle swarm optimization algorithm is proposed and applied to the simultaneous optimization of charging and discharging equivalent factors and power split rule between engine and motor in the equivalent fuel consumption minimization strategy.The optimal parameters or rule base of the strategy are formed,which lay the foundation for the strategy with driving cycle recognition.The optimization results show that,compared with the rule-based energy management strategy optimized by genetic algorithm,the equivalent fuel consumption minimization strategy optimized by inner and outer nested double-loop algorithm improves the vehicle fuel economy and charge sustaining performance with four typical driving cycles.3)Taking a random driving cycle as an example,the influence of recognition cycle and update cycle on online recognition accuracy is analyzed.According to the cross validation method,the optimal recognition period and update period are determined,and the rule-based energy management strategy based on driving cycle recognition and the equivalent fuel consumption minimization strategy are verified.The results show that compared with the energy management strategy without driving cycle recognition,the rule-based energy management strategy based on driving cycle recognition can improve fuel economy by 8.475%,while the equivalent fuel consumption minimization strategy based on driving cycle recognition can improve fuel economy by 9.845%.The SOC changing curve is relative smooth after using two strategies based on driving cycle recognition technology.The number of battery charging and discharging is significantly reduced,and the system efficiency and battery life are effectively improved.4)An energy management strategy based on stochastic model predictive control is proposed.For the application of dynamic programming algorithm in solving hybrid power energy management strategy,it is necessary to predict all driving cycle information in future,a new energy management strategy based on stochastic model predictive control is proposed by combining driving cycle prediction technology with dynamic programming algorithm.Markov chain is used as the prediction module in model predictive control,dynamic programming algorithm is used as the solution algorithm in predictive time domain,and rolling solution is carried out in predictive time domain.The simulation results of random driving cycle show that the energy management strategy based on multi-step Markov model is the best for improving vehicle fuel economy.Compared with the rule-based energy management strategy based on driving cycle recognition and ECMS based on driving cycle recognition,the fuel economy of the vehicle is improved by 19.85%and 9.96%respectively.
Keywords/Search Tags:energy management strategy, driving cycle recognition, support vector machine, dynamic programming, model predictive control, Markov chain
PDF Full Text Request
Related items