The Electronic Nose System In The Electrical Fire Detection Applications

This dissertation presents an electric fire detecting model and a new intelligent arithmetic that gives a fire prediction probability on the base of neural network and fuzzy logic inference and introduces a new method to detect the electric fire , it analyses the characteristics of the gas sensor. This new combined arithmetic gains the merits of both neural network and fuzzy logic inference, and it improves the performance of the fire detecting system by reducing the miss-warning and failing reporting rate. This dissertation reviews the researches of pioneers, gives the basic features of electric fire and shows a new intelligent electric fire detecting system.The metal oxide gas sensor and voltage following circuit are adopted in this electronic nose system. The single chip controller completes the signal transform from analog signal to digital signal and dispatch it to the home system.Following arithmetic and technologies are applied on the new design. 1. Digital Signal Analysis: FFT and Digital Filters are used to pre-treat the signal detected by gas sensors. 2. Neural network: it is used to identify the patterns of short circuit and overheated of several kinds of wires;and the research illustrates the advantages and disadvantages of it. 3. Fuzzy Inference System: it is used to imitate the instinct of human being for detecting the electric fire probability prediction. 4. Adaptive Neural Fuzzy Inference System (ANFIS): it combines the strongpoints of both NN and FIS, and it greatly improves the identification capability of the system.This dissertation forecasts the further research directions. Super-Sonics, Image Processing and other methods are also being applied to this field.