Optimisation of fermentation process using data mining techniques for small-medium industry

Fermentation monitor is widely used by industries to control the process effectively. Various online method are used such as flow injection method (FIR), liquid chromatography (HPLC), infrared spectroscopy (IR), gas chromatography (GC), mass spectrometry (MS) and others developed techniques. Those instruments are quite expensive to be used by small and cottage industries e.g. ldquotapairdquo, ldquotemperdquo, ldquopekasamrdquo and yoghurt industries. This paper will propose to use simple sensors/probes to gather information using certain variables such as pH value and temperature. From the collected information a pattern or a classification for fermentation bioprocess will be extracted using data mining technologies. The knowledge will be used to monitor fermentation using cheap sensor/probe and also can be used in intelligence system to predict the fermentation process for example fermentation duration and optimal fermentation condition. Data mining is a combination of statistical analysis, machine learning, and database management to extract information from large database systems. Different techniques used in data mining will produce different outcomes based on user specification. The outcomes are classification, association, clustering, prediction, estimation and deviation analysis. With this advantages, suitable data mining techniques will be used in this paper to analyse and classify the fermentation database obtained through the experiments. Certain pattern can be isolated and categorised. A prototype intelligent knowledge based system will then be developed to optimise the fermentation process. The system would be able to facilitate certain components of the fermentation for example estimation of complete fermentation cycle, fermentation status, temperature regulation, etc.