Research On Estimation And Control For Combustion Timing Of HCCI Engine

The homogeneous charge compression ignition(HCCI) engines have become the focus in internal combustion engine research, due to its advantages of high fuel efficiency and low emission of nitrogen oxide and particulate matter. Nonetheless, performance of HCCI engine extremely depends on the location of combustion timing, and late or early combustion will has a great effect on performance of HCCI engine, which causes that control of the combustion timing becomes the key and difficult point in HCCI engine research. However, it is worth noting that combustion of HCCI engine is different with traditional internal combustion timing, which is determind by cylinder charge characteristics instead of ignition timing or fuel injection timing, including charge temperature, pressre, and composition, etc. Thus, it is necessary to control the charge characteristics in order to control the combustion timing. And one of effective methods to change charge characteristics is intake valve closure timing in variable valve timing strategy, which could control mass of fresh charge and effective compression ratio of cylinder. Although the combustion timing could be controlled by intake valve closure timing, the combustion timing could not be measured directly due to high temperature and pressure in cylinder. Moreover, the performance is optimal when combustion timing is controlled to a pretty narrow range. Hence, it is necessary to design appropriate control algorithms based on combustion timing estimated accurately to control the combustion timing.Therefore, the research has done aimed at estimation and control for combustion timing of HCCI engine based on MATLAB, and the main works in this thesis are as following:(1) The research about estimation model for combustion timingAiming at that combustion timing estimated by nonlinear chemical kinetic model for HCCI engine is hard to meet real-time performace, this thesis establishes a linear model based on parameter identification for state space model. The numerical experiments show that the error between experimental data combustion timing estimated is 0.5421(CAD). Considering that the error is large to some extent, this thesis establishes a correction model by BP neural networks to reduce the error.(2) The research about PID controller based on radial basis function neural networksAiming at problem about control of combustion timing, a PID controller is chosen to control combustion timing. However, considering that traditional PID controller has deficicency of fixed parameters, this thesis choses gradient descent algorithm and radial basis function neural networks to adjust the parameters of PID controller. The numerical results show that, compared with traditional PID controller, PID controller based on RBF neural network can reduce the setting time about 4 cycles; besides, overshoot and tracking error are reduced to a certain extent, which illustrates the control performances of designed controller are better than the traditional PID controller.(3) The research about fuzzy neural network based PID controllerConsidering the deficiencies about relatively big overshoot of positional PID controller and shorter time for adjustment of PID parameters by gradient descent in(2), thus based on(2), an incremental PID controller is chosen to control combustion timing, and fuzzy neural network is chosen to adjust the parameters of PID controller. Eventuallly, experimental results indicate that, the fuzzy neural network based PID controller has better performance with zero-shoot and shorter setting time compared with traditional PID controller and radial basis function neural networks based PID controller.(4) The research about discrete sliding mode controller based on reaching lawAiming at that the empirical controller could not control the combustion timing by the internal dynamics of HCCI engine, which leads to long time for control.Therefore, based on linear model of HCCI engine, a discrete sliding controller is designed to control the combustion timing, where error and change ratio of error are chosen as inputs of controller; and ideal switch dynamic based reaching law is chosen as reaching law for sliding mode. Finally, the experimental results indicate that, the controller could achieve the control of combustion timing; besides, compared with the control methods in(2) and(3), control result of sliding mode controller has no overshoot, and setting time reduced by 5 cycles, which shows that the control performances of disctere slide mode controller are better.