Analysis Of Complex Gas Mixtures By Pattern Recognition With Polymer Based Quartz Microbalance Sensor Arrays

Polymer coated quartz microbalance gas sensor arrays were used to analyse quaternary mixtures of volatile or game compounds. The performance of the sensor arrays in combination with qualitative data evaluation utilising self-organising feature maps is described. The concept of context preservation artificial neural networks is compared to standard topology preservation with respect to its classification properties for model partitioning. In the studied case, context preserving maps yield more compact clusters with smaller map dimensions.