Abstract
The activities of lignin-degrading peroxidases are the primary decomposition indexes in compost bioremediation. In this paper, artificial neural networks (ANNs) have been combined with an enzyme sensor for simultaneous determination of lignin peroxidase (LiP) and manganese peroxidase (MnP) activities secreted by
Phanerochaete chrysosporium in composting of municipal solid waste. The LiP and MnP activities were detected through catalytic redox of H
2O
2, hydroquinone and veratryl alcohol as substrates by an amperometric sensor immersed in the culture filtrate solution. Due to the dynamic, nonlinear and uncertain characteristics of the complex composting system, ANNs have been used as a chemometric tool for overlapping signal deconvolution and modelling to quantify the two enzyme activities separately. Feedforward backpropagation network was used for the training process. The effects of the transfer functions, the amount of current values, the number of hidden neurons and the optimization algorithm were investigated. The LiP activities in the filtrate varied from 8.14 to 29.79
U
L
−1, and from 0.36 to 1.37
U
L
−1 for MnP activities. A good prediction capability was obtained, with correlation coefficients of 0.9936 for LiP activity and 0.9976 for MnP activity between the expected and predicted values of the external test samples. The performance of the ANN model was compared with the linear regression model in respect to simulation accuracy, adaptability to uncertainty, etc. All the results show that the combination of amperometric enzyme sensor and artificial neural networks is a rapid, sensitive and robust method in the quantitative study of composting system.