Michael Hafner

Einsatz schneller neuronaler Netze zur modellbasierten Optimierung von Verbrennungsmotoren

Aufgrund der steigenden Anzahl von Stellgrosen und zu optimierenden Ausgangsgrosen an modernen Verbrennungsmotoren werden modellbasierte Verfahren zur Optimierung des Motorverhaltens immer wichtiger. Der erste Teil dieses zweiteiligen Beitrages enthalt eine Einfuhrung in spezielle neuronale Netze,geht auf geeignete Prufstandsmessstrategien ein und stellt eine Modellbildung des Motor- und Abgasverhaltens mit Hilfe schneller neuronaler Netze vor. Mit Hilfe der Modelle kann dann offline eine Applikation der Motorsteuergeratefunktionen durchgefuhrt werden,ohne Prufstandszeit zu beanspruchen. Die Untersuchungen wurden im Rahmen des DFG-Sonderforschungsbereichs „Integrierte mechanisch-elektronische Systeme fur den Maschinenbau“ am Institut fur Automatisierungstechnik der TU Darmstadt durchgefuhrt.

Multiobjective optimization of feedforward control maps in engine management systems towards low consumption and low emissions

This paper describes a new approach towards a model-based optimization of internal combustion engine control maps. The goals of the optimization are - at the same time - minimum fuel consumption, low emissions and a good driveability. First the structure of a torque-oriented engine management system based on control maps is described. Then, an optimization environment is developed, which calculates the basic control maps for the engine settings based on the modelled emission behaviour of the engine. The underlying nonlinear models are realized by fast neural networks. Results in both simulation and measurements prove the quality of the proposed methodology.

Mechatronic Combustion Engines – from Modeling to Optinlal Control

After a short review on mechatronic systems in general, the development of internal combustion engines is discussed. Since the introduction of microprocessor control around 1980 an increasing number of sensors, actuators and digital control functions were introduced, replacing mechanical devices like ignition breaker and injection. In addition several formerly mechanical camponents with fixed functions became active, manipulated components like the electronic throttle, injection, camshaft, valves. Further, integrated sensor systems came into series production, like knock sensors, speed sensors and A-sensors. Thus an integration of mechanical, electromechanical and electronic components and an integration by software-based functions can be observed which is typical for mechatronic systems. The contribution considers first present control structures and their calibration. As the calibration and parameter tuning has become a crucial part in the timely development because of complex interactions and the many degrees of freedom, it is shown how by specially designed experiments for the identification of the engine a considerable improvement can be reached with the aid of local linear neural networks. The engine models are then used for the optimization of static and dynamic engine control, using a multi-objective performance criterion for fuel consumption and emissions. Several results are shown for Diesel engines with variable geometry turbo-chargers and exhaust gas recirculation. Then it is shown how model-based control systems can be implemented and tested with rapid control prototyping systems. A further typical mechatronic design tool is the hardware-in-the-loop simulation of the real-time simulated engine and the real electronic control unit. An outlook discusses further mechatronic developments of combustion engines.

Fast Neural Networks for Diesel Engine Control Design

Abstract Advanced engine control systems require accurate process models. This paper presents neural net models for combustion engines. After briefly introducing a special local linear RBF network (LOLIMOT) two applications are described. Different methods for developing exhaust gas models are compared and a dynamic model for the charging pressure dynamics of a turbocharger is presented. Finally, an exhaust vs. consumption optimization is presented for optimizing the injection angle dependent on given weighting factors for specific emissions, the fuel consumtion and the current driving situation.

Mechatronic design approach for engine management systems

Internal combustion engines with electronic management systems have developed to mechatronic systems. They basically consist of a thermodynamic process, electromechanic, hydraulic or pneumatic actuators, electronic sensors and one ore more digital control units with their specific software. In order to meet rising demands concerning drivability, emissions and consumption under the restriction of shorter development cycles, there is a rising need for modern identification methods (neural networks) and software/hardware tools (Rapid Control Prototyping (RCP), Hardware-in-the-Loop-Simulation) for the design of engine control units. In this contribution, a software/hardware environment for a mechatronic design approach for engine control systems is discussed. A dynamic engine test stand equipped with a RCP system is described which allows fast and comfortable design and testing of new control functions. Enlarged with an on-line indication system, a cylinder pressure based engine management system can be established, where the desired control settings are calculated by an upper level engine optimization. Time-variant optimization strategies for an improved exhaust-, consumption- and drivability performance were developed by means of adequate models of the engine behavior with fast neural networks.

Mechatronic Design Approach for Engine Management Systems

Abstract Internal combustion (IC) engines with electronic management systems have developed to mechatronic systems. They basically consist of a thermodynamic process, electromechanic, hydraulic or pneumatic actuators, electronic sensors and one ore moredigital control units with their specific software. In order to meet rising demands concerning drivability, emissions and consumption under the restriction of shorter development cycles, there is a rising need for modem identification methods (neural networks) and software/hardware tools (Rapid Control Prototyping, Hardware-in-the Loop-Simulation) for the design of engine control units. In this contribution, asoftware/hardware environment for a mechatronic design approach for engine controlsystems is discussed. A dynamic engine test stand equipped with a Rapid Control Prototyping (RCP) system is described which allows fast and comfortable design and testing of new control functions. Enlarged with an on-line indication system, a cylinder pressure based engine management system can be established, where the desired control settings are calculated by an upper level engine optimization. Time-variant optimization strategies for an improved exhaust-, consumption- and drivability performance were developed by means of adequate models of the engine behavior with fast neural networks

Model Based Control of Diesel Engines with Dynamic Neural Networks

Abstract After a short review on the development towards mechatronic combustion engines in general the model based control of diesel engines is discussed. Then the contribution considers present control structures and their calibration. Because of the complex interactions and the many degrees of freedom, the calibration task itself has become a main bottleneck. It is shown how by specially designed experiments for the identification of the engine a considerable improvement can be reached with the aid of local linear neural networks. For this purpose all the necessary programs are integrated in a Toolbox called OptiMot: Optimization of IC Motors. Several results are shown for a Diesel engine with turbocharger and exhaust gas recirculation.

Experimentelle stationäre und dynamische Motormodelle für DI-Dieselmotoren mit Turbolader und Abgasrückführung

Fur das Abgasverhalten von Verbrennungsmotoren bietet sich eine experimentelle Modellbildung an, da ein theoretischer Ansatz aufgrund der komplexen thermodynamischen und chemischen Prozesse bei heutiger Rechenleistung noch nicht zielfuhrend ist. Das schnelle neuronale Netz LOLIMOT eignet sich hierbei aufgrund einiger charakteristischer Eigenschaften (geringer Entwurfsaufwand, Interpretierbar keit, automatische Anpassung der Modellstruktur, Integrierbar keit von Expertenwissen, dynamische Modellbildung, Extrapolationsverhalten, impliziter Messfehlerausgleich und extrem kurze Trainingszeit) besonders gut.

Multikriterienoptimierung von Verbrauch und Emissionen für die nichtlineare Motorsteuerung von Dieselmotoren

Der Aufwand fur die Kalibrierung moderner ECU-Funktionen steigt seit Jahren bedingt durch die Einfuhrung einer Vielzahl elektronisch gesteuerter/geregelter Technologien im Fahrzeug rapide an. Eine manuelle Applikation der vielen Steuerkennfelder wird daher immer schwieriger. Moderne Verfahren zur Modellierung des Motorverhaltens erlauben hingegen die Implementierung von modellbasierten Ansatzen, die unter Verwendung mathematischer Routinen eine Optimierung der elektronischen Motorsteuerung im Rechner ermoglichen.

A HINGE NEURAL NETWORK APPROACH FOR THE IDENTIFICATION AND OPTIMIZATION OF DIESEL ENGINE EMISSIONS

Abstract Characteristics such as nonlinearity, uncertainty, and time-delays make the identification of dynamic processes challenging. One means of addressing this problem is to develop approaches based on artificial neural networks that are capable of modeling highly nonlinear systems. The aim of this paper is to present the non linear mathematical models of exhaust gas formations achieved with a certain Neural Network (NN) architecture, namely Hinging Hyperplane Trees (HHT). A brief description of HHT is followed by a presentation of identification of nitrogen-oxides and opacity emissions for a Diesel engine. Experimental results that show the effectiveness of the HHT approach are included. The article concludes with the description of an optimization environment for Diesel engine management based upon previously achieved emission models, and which is used in order to obtain an optimal performance regarding fuel consumption, emissions and drivability