| The short driving range and high battery cost are important factors restricting the popularization and development of pure electric vehicles at this stage.The pure electric vehicles with dual-energy source can further extend the driving range and improve the dynamic performance by adding additional energy storage devices.This paper proposes an energy management strategy based on deep fuzzy control network for a dual-energy source pure electric vehicle equipped with lithium iron phosphate battery pack and super capacitor as a composite energy storage system to optimize the power distribution between the lithium iron phosphate battery pack and super capacitor,thus improving the working efficiency and safety of the composite energy storage system.The main contents of this paper are as follows:Firstly,according to the types and test characteristics of common power batteries in electric vehicles,it is determined that the composite energy storage system is composed of lithium iron phosphate batteries and super capacitors.The series,parallel and hybrid connection forms of the composite energy storage system and their respective advantages and disadvantages are introduced in detail.According to the working principle and design requirements of DC/DC converter,it is proposed to control the output proportion of super capacitor by adding the arrangement of DC/DC converter in super capacitor branch.The change of power demand for pure electric vehicles during driving and the four corresponding working modes of the composite energy storage system are analyzed in detail,thus determining the super capacitor as the main control body to realize power distribution in the energy management strategy.The energy management problem of composite energy storage system is put forward,the mathematical model of power distribution is established,and the optimization objectives,control variables and constraints of the model are further determined.Secondly,the vehicle model and control strategy model are established in AVL Cruise and MATLAB Simulink environment,and the key components such as lithium iron phosphate battery pack,super capacitor,DC/DC converter and motor are introduced and analyzed in detail.The environment variables and simulation software of the computer operating system are preset,and the operation steps of realizing AVL Cruise and MATLAB joint simulation are introduced in detail.Based on NEDC cycle condition debugging,the correctness of the vehicle model is verified,and the rationality of the working mode of the composite energy storage system is verified.The distribution of vehicle demand power is divided into 7 fuzzy intervals,the control strategy under the 7 intervals is formulated,and the two control objectives to be realized preferentially are put forward.The fuzzy controller with Mamdani structure is adopted to determine the input and output variables,universe of discourse and fuzzy subsets,and 175 fuzzy control rules are formulated according to the coupling relation of fuzzy subsets.Then,in order to reduce the influence of expert experience in the process of making fuzzy control strategies,this paper proposes an optimization method based on deep neural network to train fuzzy rules.175 fuzzy control rules are used as samples for model training,verification and prediction through interval normalization,and the accuracy of model training is further improved by adding hidden layers.The optimal training model is determined by comparing the relevant parameters of nine neural network models under the three different training strategies of LaiWenberg-marquardt method,quantitative conjugate gradient method and Bayesian regularization method,and the joint simulation of deep fuzzy control network(DFCN)control strategy is realized through the steps of model construction,online compilation of DLL files,etc.Finally,based on NEDC,FTP75,HWFET and WLTC four cycle test conditions,the control effects of DFCN and charge-depleting(CD)control strategies in three aspects of composite energy storage system changes,lithium iron phosphate battery output current and motor operating point efficiency distribution are compared.The simulation results show that DFCN control strategy can slow down the downward trend of the curve and increase the proportion coefficient of super capacitor.The peak current output and inrush of the lithium iron phosphate battery pack can be effectively limited.The efficiency of the motor working point can be improved as a whole,and the working efficiency of the braking energy recovery working point at medium and high rotating speeds can be improved. |
