Mixed Gases Recognition Based on Feedforward Neural Network

The three gas sensors array was developed in this paper, and it was encapsulated through the Micro-Electro-Mechanical Systems (MEMS) technique. The gas sensors applied the heating unit to improve the sensitivity. The gas sensor which was sensitive to the special gas could be selected in the different application fields. The sampling experiments showed that the gas sensors have the higher sensitivity and better repeatability and cross sensitivity. Moreover, the pattern recognition algorithms based on a feedforward neural network were studied in the paper. They have the higher pattern recognition capacity, the convergence rate and simpler training method. The intelligent recognition system which adopted the gas sensor array and feedforward neural network was design. The recognition experiments showed the system has better identification effect and higher accuracy.