Model-based adaptive observers for intake leakage detection in diesel engines

This paper studies the problem of diesel engine diagnosis by means of model-based adaptive observers. The problem is motivated by the needs of guarantee high-performance engine behavior and in particular to respect the environmentally-based legislative regulations. The complexity of the intake systems of this type of engine makes this task particularly arduous and requires to constantly monitor and diagnose the engine operation. The development and application of two different nonlinear adaptive observers for intake leakage estimation is the goal of this work. The proposed model-based adaptive observers approach allows estimating a variable that is directly related to the presence of leakage, e.g., hole radius. Monitoring and diagnostic tasks, with this kind of approach, are straightforward. Two different approaches, whose main difference is on observer adaptation law structure are studied. One approach is based on fixed gains while the other method has variable gain. The paper also includes experimental results of the two studied methods on a four cylinder diesel engine testbed.