Empirical models in the investigations of internal combustion engines - limitations and examples of application

International Automotive Conference 2018

Dziedzik, P.^1 ; Danilecki, K.^1

West Pomeranian University of Technology Szczecin, Faculty of Mechanical Engineering and Mechatronics, Al. Piastów 19, Szczecin

70-310, Poland^1

关键词: Approximation function;    Approximation methods;    Comparative tests;    Engine performance;    Measurement uncertainty;    Parameter variability;    Polynomial functions;    Several variables;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/421/4/042016/pdf
DOI  :  10.1088/1757-899X/421/4/042016

来源: IOP

【 摘 要 】

The empirical model of engine is determined by set of functions that constitute dependences approximating the measurement results of engine parameters corresponding to the adopted control value system. The use of appropriate approximation method, enabling the description of parameters using polynomial functions with several variables, ensures a large convergence of the calculated results with the current running engine operation parameters. However, the adequacy of such a model is limited to range of parameter variability and engine regulation taken into account during the identification experiment. With the right range selection of factors variability, which have controlling effect on engine operation, those approximation functions can be used to calculate the average of the parameters in the whole field of engine operation. However, the number of factors affecting the engine's properties is very high. The impact of some may be very significant, others less than a measurement uncertainties or production and regulatory inaccuracies. The paper presents an example of empirical model that was used to optimize the parameters of selection for turbochargers to engine, obtaining a high convergence of calculations with the results of the experiment, but not exceeding 5%. By determining the set of inputs for baseline model, the current knowledge of the phenomena occurring in the cylinder of the piston engine and the results of the preliminary comparative tests were based on. The principle of "redundancy" was followed, striving to adopt as many factors as possible, whose effects on engine performance indicators were considered to be dominant. In order to determine the final form of the approximation function, verification of the correlation between pre-accepted variables was carried out using the sensitivity analysis. The influence of single factors and their interactions on output value in the tested range of factor variability was evaluated using sensitivity indicators and to determine them, a statistical method of regression analysis was used.