Prediction of a diesel engine characteristics by using different modelling techniques

Abstract

In this study, the characteristics of a four-stroke internal combustion diesel engine have been investigated by means of artificial neural networks (ANNs) and adaptive neuro-fuzzy inference system (ANFIS) modelling techniques, using injection pressure, engine speed and torque. Injection pressure of diesel engine has been designed with a pressure of 150 bars for the turbo charger and pre-combustion chamber. The experiments have been implemented for four different pressure values, namely 100, 150, 200 and 250 bars with throttle positions of 50, 75 and 95%. Brake means effective pressure (BMEP), fuel flow (FF), specific fuel consumption (SFC) were obtained from experimental results for four different injection pressure. The proposed ANNs and ANFIS models are composed of the results of implemented measurements. ANNs model of the diesel engine has two subsystem. The first subsystem has two outputs (BMEP, FF) and the second subsystem has single output as specific fuel consumption (SFC). In first subsystem ANNs model, both mean effective pressure and fuel flow parameters are computed concurrently. ANFIS model of system has three inputs and outputs as injection pressure, engine speed, torque, BMEP, FF and SFC, respectively. The performance of ANNs and ANFIS models are compared with each other in same figures for same experimental data. The results of modeling techniques of a four-stroke internal combustion diesel engine are observed to be very close with the experimental results.