Using a combination of RBFN, MLP and kNN classifiers for engine misfire detection

In this paper, we apply radial basis function networks (RBFN), multilayer perceptron (MLP) and a conventional statistical classifier, k-nearest neighbour (kNN), to the detection of misfires in a petrol engine. Used alone, each classifier is shown to provide a similar level of performance. We then demonstrate that by combining these techniques using a simple `majority voting' algorithm, the overall performance of the system is improved by approximately 10%