Direct Injection Diesel Engine Cylinder Pressure Modelling via NARMA Identification Technique

Future engine control systems need suitable and accurate models for combustion. For this purpose, this paper presents a practical application of nonlinear autoregressive moving average polynomial models with exogenous inputs (NARMAX) technique to model pressure dynamics inside the cylinder of a direct injection compression ignition engine. Two models have been investigated taking two different sets of input variables. The first model only includes basic injection settings available from the electronic control unit. The second model uses the instantaneous crankshaft revolution speed as a main model input. Model parameter identification and validation are performed with experimental data obtained from a test engine equipped with a piezoelectric pressure sensor and with data computed from a thermodynamic-based engine cycle simulation code.