Search

The internal combustion engine faces increasing societal and governmental pressure to improve both efficiency and engine out emissions. Currently, research has moved from traditional combustion methods to new highly efficient combustion strategies such as Homogeneous Charge Compression Ignition (HCCI). However, predicting the exact value of engine out emissions using conventional physics-based or data-driven models is still a challenge for engine researchers due to the complexity the of combustion and emission formation. Research has focused on using Artificial Neural Networks (ANN) for this problem but ANN's require large training datasets for acceptable accuracy. This work addresses this problem by presenting the development of a simple model for predicting the steady-state emissions of a single cylinder HCCI engine which is created using an metaheuristic optimization based Support Vector Machine (SVM). The selection of input variables to the SVM model is explored using five different feature sets, considering up to seven engine inputs. The best results are achieved with a model combining linear and squared inputs as well as cross correlations and their squares totaling 26 features. In this case the model fit represented by R 2 values were between 0.72 and 0.95. The best model fits were achieved for CO and CO2, while HC and NOx models have reduced model performance. Linear and non-linear SVM models were then compared to an ANN model. This comparison showed that SVM based models were more robust to changes in feature selection and better able to avoid local minimums compared to the ANN models leading to a more consistent model prediction when limited training data is available. The proposed machine learning based HCCI emission models and the feature selection approach provide insight into optimizing the model accuracy while minimizing the computational costs. [ABSTRACT FROM AUTHOR]

A correlation-based model order reduction algorithm is developed using support vector machine to model NOx emission and break mean effective pressure of a medium-duty diesel engine. The support vector machine-based model order reduction algorithm is used to reduce the number of features of a 34-feature full-order model by evaluating the regression performance of the support vector machine-based model. Then, the support vector machine-based model order reduction algorithm is used to reduce the number of features of the full-order model. Two models for NOx emission and break mean effective pressure are developed via model order reduction, one complex model with high accuracy, called high-order model, and the other with an acceptable accuracy and a simple structure, called low-order model. The high-order model has 29 features for NOx and 20 features for break mean effective pressure, while the low-order model has nine features for NOx and six features for break mean effective pressure. Then, the steady-state low-order model and high-order model are implemented in a nonlinear control-oriented model. To verify the accuracy of nonlinear control-oriented model, a fast response electrochemical NOx sensor is used to experimentally study the engine transient NOx emissions. The high-order model and low-order model support vector machine models of NOx and break mean effective pressure are compared to a conventional artificial neural network with one hidden layer. The results illustrate that the developed support vector machine model has shorter training times (5-14 times faster) and higher accuracy especially for test data compared to the artificial neural network model. A control-oriented model is then developed to predict the dynamic behavior of the system. Finally, the performance of the low-order model and high-order model is evaluated for different rising and falling input transients at four different engine speeds. The transient test results validate the high accuracy of the high-order model and the acceptable accuracy of low-order model for both NOx and break mean effective pressure. The high-order model is proposed as an accurate virtual plant while the low-order model is suitable for model-based controller design. [ABSTRACT FROM AUTHOR]