System identification of the smoke opacity vs. fuel metering in a diesel engine

The use of diesel engine has been increasing in agricultural and industrial applications. The unfavorable diesel smoke, however, discourage the wide-spread use due to health hazards. The smoke also limits the maximum power output. Reduction of diesel smoke by adaptive control was the ultimate goal of this research. The objective of this study was to find an empirical model of diesel engine transfer function for control purpose. With an assumption of linear diesel operation in a limited region, the linear theory was adopted. Since a smoke problem generally appears during unsteady operation, fuel metering rate was selected as an input variable. Binary input signals (pseudo-random binary sequence and multi-frequency binary sequence) were selected as input signals. Step input signal was also used for transient study of smoke production. The system identification method using binary signals was tested by computer simulation before applying it to the engine experiment.; An engine control and monitoring package was developed using an 8088 based personal computer. This package included a host computer, a multipurpose interface board for signal processing, a fuel metering valve perturbing mechanism with a modified fuel metering pump, an in-line smoke opacity meter, and an operating software written in C and Assembly languages. Temperatures at various engine locations and air/fuel flow were also measured. The spectral analysis was implemented to find the frequency response of system for the pseudo-random binary signal. The multi-frequency binary input and the output signals were Fourier transformed. The resulting frequency domain signals were plotted on Bode plot and postulated transfer functions were fitted using the Box nonlinear optimization procedure.; The higher order of transfer function between smoke opacity and fuel metering rate was obtained for 9 operating points. These models can be utilized in development of an electronic injection pump for the reducing smoke level and increasing fuel economy. With sufficiently fast computation, on-line identification might be possible and adaptive control of fuel metering could be implemented. Adaptive control with on-line identification and appropriate sensors leads to an integrated engine management system.