Application of Electronics Nose to Monitor and Extract the Predictive Modeling for Lab-Scale Biogas Digester Conditions

Process control and monitoring in operation of biogas production require various parameters to be measured. Therefore, the methods that express the effect of combination of parameters via robust and inexpensive equipment for fast and accurate measurement have high value.  In this research, the relationship between the indicator of daily production of biogas and the signals of the gas sensor array in an electronic nose system designated for biogas has been investigated to analyze and predict the condition of biogas digester in laboratory scale. To produce biogas, 1 liter batch digesters were fed by different combinations of two substrates at mesophilic condition. The Sensor array signals of each digester were divided into two groups of 1 and 2 by the clustering analysis (CA) method, which correspond to the balanced and imbalanced groups, respectively.  By analyzing score plot and also correlation loading in the PCA, MQ-4 and MQ-136 sensors were determined as main indicators for detecting the two groups and rest of sensors also contribute to interpret conditions of digesters. At final, to predict the conditions, a pattern recognition was defined by sensor array signals of all digesters in linear discriminant analysis (LDA) that the classification accuracy and resubstitution error were 100%   and 0.0476 respectively, also the accuracy of the prediction pattern was 81.25%.