Developing a state space model for a turbocharged diesel engine using the subspace identification method

Abstract

In this article, a state space representation of a turbocharged diesel engine is provided on the basis of the subspace identification method. Identification is the methods of deriving mathematical expressions for dynamic systems based on input–output data. Providing such data usually leads to costly time-consuming test procedures. Any effort to decrease the number of tests is favourable for engine designers since it reduces both the engine controller development time and the cost. In this article a validated mean-value model of the engine is employed to generate demanded input–output data for identification purposes. Using this method, it is necessary to perform a smaller number of tests. The subspace method is employed to identify a linear time-invariant discrete state space model. State space identification methods such as subspace methods lead to non-physical states which are not suitable for control and fault detection purposes. An appropriate matrix transformation is used to convert the non-physical states to the desired physical states. Comparison between the non-linear model results and the state space data show good correlation.