Optimal Design And Energy Management Of The EVT-based Hybrid Electric Powertrain

Electric variable transmission(EVT)is a novel compound electric machine,which can be used in hybrid electric vehicle(HEV),equivalent to a planetary gear set plus one generator and one electric machine.It can realize power split in series-parallel powertrain and greatly simplify the powertrain configuration.The EVT based hybrid powertrain contains multiple components such as electric machines and battery,whose sizes have influence on the powertrain performance.Most of the design and energy management of the EVT hybrid powertrain focus on the single objective optimization and rule-based strategies,whose sizing and control require further improved.This dissertation investigates on the multi-objective sizing and energy management for the EVT powertrain.The dissertation starts with the optimal design of the EVT powertrain.To quickly obtain the optimal design,a fast design methodology using the drivability Pareto lines is put forward,which minimizing the sizes of EM1 and EM2 while powertrain performances,such as acceleration ability,top speed and fuel economy,are still guaranteed.Meanwhile,after analyzing the powertrain features,control subproblems using different level of dynamics are formulated to search different powertrain performances.To reduce the computational time of optimizing fuel economy and acceleration time,fast DP is formulated by vraiables inverting,which reduces the state/control variables to be gridded.When searching the top speed,static optimization is applied to improve the accuracy of the results.Pareto line of each performance is obtained by nested optimization method,which allows to find out the major influence of the EVT parameters on the fuel economoy,acceleration ability and top speed.Simulation results give out the fuel economies of different driving cycles with different gear combinations as well as the EVT sizes.It is verified th at the proposed fast design method can reduce 45% and 11.8% of the peak power of EM1 and EM2,while still maintain the same fuel economies.Also,different gear combinations have great effect on the the fuel consumptions and EVT power ratings.The optimization of the EVT Plug-in hybrid electric vehicle involves not only the battery sizing but also the energy management strategy.To reduce the computation time,convexification of the EVT powertrain in different operating modes is proposed to enable convex programming to optimize the battery size and the control simultaneously.In hybrid mode,the engine and the EVT are viewed as a compound unit,whose fuel power could be approximated as quadratic function as its net output power.With the compound unit,the optimization in hybrid mode could be solved by a bilevel program,whose the upper level optimizes the power split between the battery and the compound unit while the lower level searches the optimal engine speed.In this way,the optimization in hybrid mode can be convexified and solved by convex optimization.The EVT Plug-in hybrid powertrain is optimized by different charge depleting strategies and different machine driven ways.Compared to EM2 single driving,dual machine driving can improve the all el ectric range in CDCS strategy by 4%,while very little in Blended strategy.Blended strategy can reduce the daily driving cost by 8.3% compared to CDCS.Energy management for the EVT powertrain is a nonlinear mix-integer program.To realize the realtime implementation,predictive control and adaptive control based on equivalent consumption minimization strategy(ECMS)both are proposed.An iterative ECMS-DP program is put forward to fast solve the global optimization of the EVT powertrain,where ECMS and DP optimizes the power split and the engine on/off control respectively,which greatly reduces the computational time to optimize the receding horizon.A linearized adaptive law is also proposed to determine the equivalent factor online,which uses the upper and lower boundaries of the equivalent factors and the state of the charge.With the equivalent factor,the Hamilton values in different modes could directly decide the power split and engine on/off control,which reduces the complexity of parameter tuning in the non-predictive energy management.Both of these two startegies calculate the operating points of the components after obtaining the optimal power split,which further reduces the computation burden.Simulation results show that,under three different driving cycles,ECMS-DP could reduce the computation time by 99% compared to DP.Also,the optimality of predictive energy management could achieve 97% while that of the adaptive energy management is within 92%?96%.Finally,a reduced-scale hardware-in-the-loop simulation platform for the EVT powertrain is built to verify the design parameters as well as the control strategies.In the platform,the permanent magnetic EVT machine is connected to two induction machines.The engine and the road load are emulated by the two induction machines.Using Matlab/Simulink,the control loops of the two emulation systems and the EVT powertrain,together with virtual battery pack are modelled and downloaded to d SPACE.Experimental tests validate the realtimeness of the predictive energy management,the viability of the platform and the previous deisgn and control methodology.