The control of modern engines is becoming increasingly sophisticated and model-based control strategies are now commonly used to provide greater precision of control. Using a model to predict the behaviour of part of an engine system allows the control software to make decisions based on the output of the model and avoids the need for additional sensors. During the design of such models a balance must be found between the model accuracy and the processing resource that is required to run the model. Once calibrated a good model should also maintain an accurate output over the operating range of the engine. This paper describes the optimisation of the design and calibration of a model that predicts the temperature of an actuator within a diesel fuel injector. The accuracy of the fuel quantity injected can be improved once the actuator temperature is known. The optimisation of the model is achieved using the Taguchi 'Robust Engineering' approach, which assesses the model over a range of operating conditions that affect the actuator temperature. The Taguchi approach normally requires multiple design options to be built and tested, which can be time consuming and expensive. The advantage of using the Taguchi method for model optimisation is that engine measurements need only be taken once, and then any number of design options can be evaluated against the same test data. In this way the complexity of the model can be adjusted to achieve the desired level of accuracy. This approach also provides rigorous justification and documentation for the selection of the optimised model.
Optimising predictive models used within engine control software
IAV Symposium Steuerungssysteme für den Antriebsstrang von Kraftfahrzeugen, 6 ; 284-Folienbeitrag
2007
19 Seiten, 9 Bilder, 2 Quellen
Folienbeitrag
Conference paper
English
Optimising engine performance towards emission control
British Library Conference Proceedings | 1995
|Multilevel Predictive Models of IC Engine for Model Predictive Control Implementation
SAE Technical Papers | 2008
|Multilevel Predictive Models of IC Engine for Model Predictive Control Implementation
British Library Conference Proceedings | 2008
|Automotive engineering | 2008
|Optimising engine performance and emissions using Bayesian techniques
Automotive engineering | 1997
|