Nonlinear Identification Modeling for PCCI Engine Emissions Prediction using Unsupervised Learning and Neural Networks 2020-01-0558
Premixed charged compression ignition (PCCI) is an advanced combustion strategy which has the potential to achieve ultra-low nitrogen oxides and soot emissions while maintaining diesel level efficiencies. PCCI combustion is characterized by complex nonlinear chemical-physical process which indicates a physical description involves significant development times and also high computation cost. This paper presents a method to use cylinder pressure data and engine operations parameters for prediction of PCCI engine emissions by nonlinear identification and unsupervised learning techniques. The proposed method uses principal component analysis to reduce the dimension of the cylinder-pressure data. A multi-input multi-out model was developed for nitrogen oxides and soot emissions prediction by multi-layer perceptron (MLP) neural network. Before the training process, another principal component analysis was done to reduce input dimension with hyper-parameters there by reducing memory requirements of the models. The algorithm was applied to an experimental data set from a single-cylinder light-duty engine with piezo injection system. By comparing the model predictions with experimental results, it was shown that the neural network coupling with unsupervised learning method could successfully captured the nonlinear relationship between the state parameters and the emissions of PCCI combustion system.
Wang Pan, Metin Korkmaz, Joachim Beeckmann, Heinz Pitsch