The Effect of Natural Gas Substitution Ratio and Diesel Injection Timing on Accuracy of In-cylinder Pressure Prediction DNN Model from Vibration Signal in a CNG-Diesel Dual-Fuel Engine

Abstract

In-cylinder pressure of a single-cylinder dual-fuel engine using compressed natural gas(CNG) and diesel was predicted from engine block vibration. A supervised learning model was built using a deep neural network (DNN). Vibration and in-cylinder pressure signals measured by the accelerometer and piezoelectric pressure sensor were used as input and output labels. A parametric study was conducted in order to check the effects of the composition of train conditions on the accuracy of the model. CNG substitution ratio and diesel injection timing were selected as main parameters. Train conditions were divided by combustion modes, which are the pilot-dual fuel(pilot-DF) and reactivity-controlled compression ignition(RCCI), in order to study the effects of parameters on the accuracy of the model. The results showed that the prediction accuracy of the model is related to the amount of train data which have similar trends of in-cylinder pressure with the test conditions, regardless of the total amount of train data.