Space-filling optimized excitation signals for nonlinear system identification of dynamic processes of a diesel engine

Abstract

This paper examines the space-filling property and expected model quality of two recently developed excitation signals – the Global Optimized Amplitude Time Signal (GOATS) and the incremental Global Optimized Amplitude Time Signal (iGOATS) – on a real-world nonlinear dynamic system (the diesel engine) characterized by a high degree of system complexity. The diesel engine’s system complexity includes infeasible operating regions, strong nonlinearities, and multiple feedback paths. Additionally, a novel variant of the GOATS – the stacked Global Optimized Amplitude Time Signal (sGOATS) – is developed and investigated. For the analysis of the expected model quality and space-filling property of the GOATS, iGOATS, and sGOATS, they are compared to the state-of-the-art excitation signals Amplitude Pseudo Random Binary Signal (APRBS) and Optimized Nonlinear Input Signal (OMNIPUS) to approximate the exhaust pressure of a diesel engine depending on multiple inputs. The results show that the GOATS, iGOATS, and sGOATS outperform the APRBS and OMNIPUS regarding their space-filling property and expected model quality.