Pressure Sensor Data-Driven Optimization of Combustion Phase in a Diesel Engine

Abstract

In this article, a pressure sensor data-driven optimization by using extremum seeking (ES) technique is applied to optimize the combustion phase in a diesel engine. In particular, a cost function is constructed for the ES optimization with the in-cylinder maximum temperature $\hat{T}_{\text{max}_n}$ , which is an indication of NOx emission performance, and the thermal efficiency $\eta _n$ as tradeoff considerations. The data-driven nature lies in the fact that $\hat{T}_{\text{max}_n}$ is estimated from the in-cylinder pressure measurement within each engine cycle by assuming the in-cylinder process to be a quasi-steady state process, which is, then, compensated with the model-free ES optimization. It is noted that the approach developed in this article does not require the use of real NOx sensor data, which could otherwise jeopardize the ES optimization practically due to the unignorable and hard-to-predict gas transportation delay. Therefore, it is argued that this article actually delivers an alternative soft sensor solution to real NOx sensors for engine optimization purpose. The effectiveness of the proposed method is demonstrated on a single-cylinder diesel engine test bench with the help of steady-state engine tests. The method can be extended to data-driven optimization applications that also utilizes physical sensors with measurement delay.