Intelligent Processing of E-nose Information for Fish Freshness Assessment

Intelligent odour detection is a challenging research topic and requires an in depth study into types of sensors, advanced signal processing and pattern recognition methods. Electronic nose (e-nose) can be used to identify odour variety, aroma intensity and degree of freshness. The e-nose consists of an array of conductometric chemical sensors which change resistance when exposed to vapours. The applications of commercial e-noses are broad and continuously expanding to various areas such as environmental monitoring, medical instrumentation, and food industries. This paper presents intelligent processing of information collected by a portable 32-sensor e-nose for fish freshness assessment. We used artificial neural networks (ANNs) to design an intelligent system for measuring of the freshness of spoiling fish. It was found that the degree of selectivity and the type of odours that could be detected by the e-nose are largely dependent on the choice and number of sensors in the sensor array. We identified the most sensitive sensors, which provide a good estimation of freshness assessment in a laboratory environment. The information processing method was applied to collected data from four species. The procedure was repeated every day during a period often days after catching the fish. It was found that the proposed intelligent system can successfully identify the number of days after catching the fish with an accuracy of up to of 91%. The results obtained from this research suggest that the proposed assessment method can be used in the design of an intelligent fish freshness assessment system.