Anselm Schwarte

Model-based fault detection of a diesel engine with turbo charger - a case study

Abstract Modern Diesel engines with direct fuel injection and turbo charging have shown a significant progress in fuel consumption, emissions and driveability. Together with exhaust gas recirculation and variable geometry turbochargers they became complicated and complex processes. Therefore, fault detection and diagnosis is not easily done and need to be improved. This contribution shows a systematic development of fault detection and diagnosis methods for two system components of Diesel engines, the intake system and the injection system together with the combustion process. By applying semiphysical dynamic process models, identification with special neural networks, signal models and parity equations residuals are generated. Detectable deflections of these residuals lead to symptoms which are the basis for the detection of several faults. Experiments with a 2.01 Diesel engine on a dynamic test bench as well as in the vehicle have demonstrated the detection and diagnosis of several implemented faults in real time with reasonable calculation effort.

Model-based fault detection and diagnosis for Diesel engines

Due to the increasing complexity of diesel engines with more and more electrical and electronic components and sophisticated control strategies, automatic fault detection and diagnosis is becoming increasingly important. Within the scope of the FVV project “Model-Based Fault Detection and Diagnosis for Diesel Engines” at the Institute of Automatic Control, Darmstadt University of Technology, new model-based methods for monitoring the intake system, the injection, the combustion and the exhaust system have been developed.

NEURAL NETWORK APPLICATIONS FOR MODEL BASED FAULT DETECTION WITH PARITY EQUATIONS

Abstract The rising complexity of modern automotive engines with an increasing number of actuators and sensors to minimise emissions and fuel consumption and to maximise engine driveability require a detailed supervision for fault detection and on-board diagnosis. The European Community Directive 98/69/EC requires on-board diagnosis for spark ignition engines and will require it for diesel engines as of January 2003, mainly to prevent excessive emissions. Beside this regulation it is also in the interest of the automobile manufactures to establish capable diagnosis systems for maintenance, repair and the benefit of their customers. This paper will describe applications of neural networks for modelling complex fluid- and thermodynamics with unknown physical model structure. Reference models, which describe the fault free process, are set up and identified with the special neural network LOLIMOT (Local-Linear-Model-Tree). Fault detection algorithms, which employ the method of parity equations, were successfully implemented and tested in real time with a 2 litre diesel engine and a Rapid Control Prototyping System. Measurements of online fault detection are shown for several built-in faults in the intake system of this diesel engine.

Modellbasierte Fehlererkennung und -diagnose für Dieselmotoren

Aufgrund der wachsenden Komplexitat des Dieselmotors durch Zunahme von elektrischen/elektronischen Komponenten sowie aufwandiger Steuerung und Regelung erhalt die automatische Fehlererkennung und -diagnose des Dieselmotors eine grosere Bedeutung. Im Rahmen des FVV-Projekts „Modellgestutzte praventive Diagnosemethoden (Fehlerfruherkennung) fur Dieselmotoren“ am Institut fur Automatisierungstechnik der TU Darmstadt wurden daher neue modellbasierte Methoden zur Uberwachung von Ansaugsystem, Einspritzung und Verbrennung sowie Abgassystem entwickelt.

Modellgestützte Fehlerdiagnose des Ansaugsystems eines Dieselmotors

Aufgrund der Sensibilitat der Abgasemissionen auf Fehler im Ansaugsystem eines Verbrennungsmotors steht die Diagnose des Ansaugsystems eine Kernaufgabe der Motoruberwachung dar. Es wird gezeigt wie durch den Einsatz modellgestutzter Fehlerdiagnosemethoden fur das Ansaugsystem kleine Fehler schnell und fruh erkannt werden sowie eine weitere Fehlereingrenzung zur Fehlerisolation moglich ist. Hierzu werden geeignete, vereinfachte Modelle hergeleitet, am Motorprufstand identifiziert und zusammen mit Fehlererkennungsalgorithmen auf einem Rapid Prototyping System implementiert. Letztlich werden die Ergebnisse der online Fehlererkennung ausgewertet.