It is the background to add the sniffing sense for the robot. The methods to buildup the artificial olfactory system(AOS) were discussed, many relative questions about the manufacturing of gas sensors were investigated and the corresponding resolutions were put forward.Firstly the neurobiological model was introduced for the olfactory system of mammals. The newest advances of computer simulation for the AOS were summed up. The several sensors used in AOS now were analyzed. The properties and sensitive mechanism were studied thoroughly. The new type SnO2 based gas sensors were manufactured with nanometer thin sensitive film. Higher sensitivity and selectivity were gotten at lower temperature.The feature extraction of the responding curves is the base of the pattern recognition of odor identification. The quasi-phase-changing kinetics model was proposed originally, by which the full dynamic information was acquired. The other point of AOS is the method of data preprocessing and pattern recognition. The statistics and neural networks method were employed. The fuzzy-C mean clustering and self organization feature mapping neural networks (SOM) were combined to overcome the limitations of SOM method, with which the correct rate of classification improved.As an example of engineering application, the measuring and controlling system based on electronic nose technology for thermal chemical treatment was reported detailedly in chapter 8. The sensors work at the outside of the furnace, the working conditions are improved and the life is lengthened, producing costs are decreased while the control precision is increased. The control process is visual and the productivity is raised.
|