Research On Model-Based Control Of Partially Premixed Combustion Engine

Partially Premixed Combustion(PPC),as an advanced combustion concept,achieves higher thermodynamic efficiency and lower soot and nitrogen oxides(NOx)emissions,and achieves higher controllability than homogeneous charge compression ignition concept.PPC approaches longer ignition delay with the use of low reactivity fuel and high exhaust gas recirculation(EGR)rate,and ignites mixture with high compression ratio.Furthermore,combustion process can be optimized flexibly by manipulating multiple injection strategies.In order to expand the practical use of PPC and meet the demands of stricter emission regulations,control-oriented modeling and advanced control methods are investigated regarding the control problem of real-time per-formance optimization and emission constraints for PPC engine.The experimental system of PPC engine was developed based on Scania D13 heavy duty 6-cylinder diesel engine.Steady state experimental research was conducted to deeply investi-gate the effects of injection parameters on combustion process,engine performance and emis-sions in a PPC engine operated with double injection strategy.Experimental results revealed that the effects of pilot injection timing on ignition delay,combustion duration,pressure rise rate and soot emission presented nonlinear relationships,while pilot injection amount and main injection timing showed linear relationships with each parameter.With the use of sensitivity analysis method,the significance of input parameters on output parameters was quantified,and the method of selecting control path for different optimization targets was proposed.Based on Wiebe function,a novel linearization analysis method was proposed in this study.By analyzing experimental data with the proposed method,the piecewise characteristics of PPC heat release were identified,along with the corresponding characteristic parameters.Conse-quently,the control-oriented combustion model was developed with the inputs of injection strat-egy,and the on-line prediction of cylinder pressure and heat release rate during combustion process was achieved.In the steady experimental validation,the prediction error of cylinder pressure was within 5%.By combining the predictive soot and NOx models,on-line estimations of soot and NOx were achieved,of which the error of cumulative emissions in a transient cycle were both below 10%.The accuracy of the proposed control-oriented model meets the demands of control purpose.Based on the steady state experimental research results and control-oriented combustion model,a model-based PI controller was designed to control PPC engine during transient opera-tions to solve the trade-off problem between soot emission and pressure rise rate.To eliminate the influence of noise disturbance soot measurement delay on the feedback loop,gain schedul-ing strategy and Smith predictor were designed respectively to optimize the PI controller per-formance.Simulative and experimental results showed that the proposed controller tracked well with the desired target of indicated mean effective pressure(IMEP)and combustion phas-ing(CA50)for PPC engine during transient operations,meanwhile soot emission and pressure rise rate compromised the trade-off relationship and were constrained within desired limits by manipulating pilot injection strategy.Furthermore,on the basis of the control problem of soot emission and pressure rise rate,the control of NOx emission was introduced and a multiple-input multiple-output(MIMO)problem was formulated,where outputs are more than inputs,Conse-quently,it was hard to decouple the control system.Regarding this,a Model Predictive Control(MPC)structure was designed to achieve optimal control of IMEP and CA50 with pressure rise rate,soot and NOx emissions implemented as indirect constraints during transient operations.By introducing a softening term,the trade-off relationship between target tracking and state constraint was improved,and the control error of IMEP and CA50 was reduced.Experimental results showed that under the transient operating cycle,the cumulative soot and NOx emissions were reduced by 19.8%and 10.2%.In addition,apart from the approach of PPC optimization through injection strategy,re-search shows that EGR also significantly contributes to thermodynamic efficiency of PPC en-gine.Considering the dynamic of airpath system,it is necessary to carry out the investigation on EGR control in transient operations.In this study,an MPC controller was designed and a novel MPC solving method was proposed based on the error updating algorithm from iterative learning control(ILC).Control error and control input were updated within the receding horizon in the form of filtering,consequently the optimal control sequence was solved.The proposed control algorithm was applied to the engine EGR control,and evaluated in co-simulation envi-ronment of GT-Power and Simulink.Transient simulation results confirmed the controllability of the proposed controller.By investigating the effects of filter parameters on the dynamic per-formance of the control algorithm,it was found that the phase delay of L filter for error updating showed a significant influence on the dynamic response and robustness.