An Ensemble Classifier Based on Feature Selection Using Ant Colony Optimization

An ensemble classifier based on feature selection is proposed to provide an effective methodology for classifying device states. We used Support Vector Machine (SVM) as base classifiers, and defined the classification performance and similarity measure to construct models using serial and parallel classifiers associated with the feature selection. We put forward Ant Colony Optimization (ACO) algorithms with three models, and verified the performance of the proposed ensemble classifier on vibration signals with five working states from a certain engine's cylinder head.