Data Preprocessing and Modeling of Electronically- Controlled Automotive Engine Power Performance using Kernel Principal Components Analysis and Least-Squares Support Vector Machines

Modern automotive engines are controlled by the electronic control unit (ECU). The electronically-controlled automotive engine power & torque are significantly affected with effective tune-up of ECU control parameters. Current practice of ECU tune-up relies on the experience of the automotive engineer. The engine tine-up is usually done by trial-and-error method, and then the vehicle engine is run on the dynamometer to test the actual engine output power and torque. Obviously the current practice costs a large amount of time and money, and may even fail to tune up the engine optimally because a formal power and torque model of the engine has not been determined yet. With an emerging technique, Least Squares Support Vector Machines (LS-SVM), the approximated power and torque model of a vehicle engine can be determined by training the sample data acquired from the dynamometer. The number of dynamometer tests for an engine tune-up can therefore be reduced because the estimated engine power and torque functions can replace the dynamometer tests to a certain extent. Moreover, Bayesian inference is also applied to automatically infer the hyperparameters used in LS-SVM so as to eliminate the work of cross-validation, and this leads to a significant reduction in training time. Besides, the actual number of adjustable parameters in the ECU for an automotive engine is very huge. Kernel Principal Components Analysis (KPCA) is therefore proposed to reduce the number of unimportant adjustable parameters. KPCA is a well known data pre-processing method under support vector machines (SVM) formulation. KPCA can transform automotive engine adjustable parameters into a more compact subset, while retaining the information content as much as possible. With fewer input parameters, the training time for model construction can be shortened and also the prediction accuracy of the model can be improved. Experimental results show that the integration of KPCA and LS-SVM can really improve the training time and accuracy of an automotive engine model as compared with the LS-SVM method without data preprocessing.