Methods of computational intelligence to give qualitative and quantitative statements of gas concentrations at a high temperature sensor

This paper describes property evaluation and clustering of sensor data that depends on different gas types and their concentrations. The application subject is to find gas specific information by a multidimensional feature set to give qualitative and quantitative statements. This is investigated by different clustering methods, the fuzzy-c-means algorithm and its derivatives and neural networks. The resulting patterns will be deployed as recognition field for gas detection, that takes recalibration mechanisms into consideration.