Committee machine for LPG calorific power classification

Abstract

This work shows the results of the development of a robust system as an alternative to recognize the quality of an alcohol vapor fuel sample and liquid petrol gas (LPG) calorific power in an electric nose. Two experimental methodologies were implemented to extract the features of alcohol vapor fuel and LPG gas patterns. The first approach used the multi-layer perceptron (MLP) topology of artificial neural networks (ANN) to recognize alcohol vapor fuel patterns. The second approach processed data to develop an LPG calorific power recognizing system that is robust to the loss of a random sensor. Three systems were used. The first implemented an MLP to recognize all data that simulated the failure of a random sensor. This system had 97% of right responses. The second implemented seven MLPs trained with input data subsets, so that six MLPs were trained with a different failure sensor, and the seventh MLP was trained with data considering all sensors without failure. This system had 99% of right responses. The third implemented an ensemble static learning machine containing 10 parallel MLPs. This system had 97% of right responses.