6837092894

Diagnostyka’ 4(48)/2008

KEKEZ, AMBROZIK, RADZISZEWSKI, Modeling of Cylinder Pres surę in Compression Ignilion Engine ...

MODELING OF CYLINDER PRESSURE IN COMPRESSION IGNITION ENGINE WITH USE OF GENETIC-FUZZY SYSTEM PART 1: ENGINE FUELED BY DIESEL OIL

Michał KEKEZ, Andrzej AMBROZIK. Leszek RADZISZEWSKI

Kielce University of Technology, Faculty of Mechatronics and Machinę Building 25-314 Kielce, Al. Tysiąclecia Państwa Polskiego 7, building B, fax: [phone], e-mail:, [e-mail], [e-mail] [e-mail]

Summary

This paper concems measurement and modeling cylinder pressure in diesel engines. The aim of this paper is to build the empirical-analytical model of engine work. The experiments on engine test bench were conducted. The new genetic-fuzzy system GFSm was proposed. By means of GFSm, the engine model was built. This model allows simulation of cylinder pressure, for each allowable crankshaft speed. The model can be used to evaluate tlie ąuality of working cycles of piston engine with accuracy reąuired in practical technical applications.

Keywords: diesel engines, modeling, fuzzy Systems, genetic algoritluns.

MODELOWANIE PRZEBIEGU CIŚNIENIA W CYLINDRZE SILNIKA O ZAPŁONIE SAMOCZYNNYM PRZY POMOCY ALGORYTMU GENETYCZNO-ROZMYTEGO CZĘŚĆ 1: SILNIK ZASILANY ON

Streszczenie

Praca dotyczy pomiaru i modelowania przebiegów ciśnień w cylindrze silnika o zapłonie samoczynnym. Celem pracy jest budowa analityczno-empirycznego modelu pracy silnika. Przeprowadzono badania eksperymentalne na hamowni oraz opracowano nowy system genetyczno-rozmyty GFSm. Przy jego użyciu zbudowano model pracy silnika, który pozwala przeprowadzać symulację przebiegów ciśnień w cylindrze silnika, dla wszystkich dopuszczalnych prędkości obrotowych walu korbowego. Może także służyć do oceny jakości cykli pracy tłokowych silników spalinowych z dokładnością wymaganą w praktycznych zastosowaniach technicznych.

Słowa kluczowe: silniki o zapłonie samoczynnym, modelowanie, systemy rozmyte, algorytmy genetyczne.

1. INTRODUCTION

Intemal combustion engine is a complex mechatronic device that responds to indications frotn many transducers. On the basis of them, the control values are generated in order to obtain reąuired pararneters of engine work: e.g. amount of fuel per cycle, injection angle or ignition angle. The control system should allow checking and manipulating the fuel supply process and combustion of the fuel during eveiy working cycle of the engine. In order to achieve this, the measurement of cylinder pressure rnust be performed. Having known the clianges of diis physical ąuantity, we can calculate a series of important pararneters such as: indicated power, mean indicated pressure, relative air/fuel ratio.

Time-consuming and expensive field as well as bench tests can be partially replaced by appropriate models that describe engine work. Empirical models (based on the first law of thennodynamics [9]) as well as CFD (Computational Fluid Dynamics) models [1] are well known. Expert Systems, artificial neural networks, genetic algoritluns, fuzzy logie and neuro-fuzzy networks [4] are used relatively rare. Neural networks are most freąuently used. for several reasons. They can easily represent nonlinear Systems and they have ability to self-train as well. Moreover, they can be applied in various controllers.

Most publications about intemal combustion engine modeling with use of artificial intelligence methods concem the control of engine work [6] and exhaust gas ernission [2, 3], Spray penetration in diesel engine, depending on fuel pressure and density, was also modeled, by means of neuro-fuzzy system [8], The modeling of cylinder pressure curves in diesel engines, by means of genetic-fuzzy Systems, was not presented yet. The review of literaturę shows that existing engine work models, botli numerical and analytical, have some limitations which make it difficult to use in practice. As a result, the aim of the paper was fonnulated following: to develop a new genetic-fuzzy system for engine work modeling.

In order to achieve this, the necessary measurements of cylinder pressure curves were tnade on the test bench. The engine operated in