Signal Processing Parameters for Estimation of the Diesel Engine Combustion Signature

Document Type

Article

Publication Date

5-17-2011

Department

Department of Mechanical Engineering-Engineering Mechanics

Abstract

Research into the estimation of diesel engine combustion metrics via non-intrusive means, typically referred to as "remote combustion sensing" has become an increasingly active area of combustion research. Success in accurately estimating combustion metrics with low-cost non-intrusive transducers has been proven and documented by multiple sources on small scale diesel engines (2-4 cylinders, maximum outputs of 67 Kw, 210 N-m). This paper investigates the application of remote combustion sensing technology to a larger displacement inline 6-cylinder diesel with substantially higher power output (280 kW, 1645 N-m) than previously explored. An in-depth frequency analysis has been performed with the goal of optimizing the estimated combustion signature which has been computed based upon the direct relationship between the combustion event measured via a pressure transducer, and block vibration measured via accelerometers. The goal of this paper is to explore the signal processing parameters involved in this process and to introduce the idea of the use of the singular value decomposition (SVD) to indentify frequency content required to accurately and robustly estimate the combustion signature which may be otherwise unidentifiable in standard frequency domain measurements such as coherence and frequency response. In addition to the singular value decomposition other signal processing techniques are employed with the goal of optimizing the estimated combustion signature, including optimization of the time-domain windowing process and the application of various filtering regimes.

Publisher's Statement

© 2011 SAE International. Publisher’s version of record: https://doi.org/10.4271/2011-01-1649

Publication Title

SAE International Journal of Passenger Cars - Mechanical Systems

Link to Full Text

Share

COinS