Perturbation/correlation based optimization and its application to internal combustion engines

This paper addresses the application of the perturbation/correlation method to optimizing the torque output of internal combustion engines. This application was inspired by observations of the limitations in current techniques of the automotive performance tuning industry. The engine simulated is the Co-operative Fuel Research Engine, which is an industry standard testing engine. Issues such as selection of parameter perturbation frequency and amplitude are addressed as well their effect on input parameter stability. Based on these parameters, a basis for correlation multiplication terms will be chosen. Performance issues to take into consideration are errors from true optimum spark timing and fuel injector pulse width values as well as convergence. The ability of the system to deal with engine cycle to cycle variations and its effect on input parameters is also analyzed.