Software sensors for biomass concentration in a SSC process using Artificial Neural Networks and Support Vector Machine

In this work NARX-ANN, NARMAX-ANN and NARX-SVM models are compared when acting as software sensors of a relevant state variable for a Solid-substrate cultivation (SSC) process. Results show that NARX-SVM outperforms the other models with an Index of Agreement close to 1.0 even under very noisy conditions thus confirming the claimed superiority of SVM over other black-box techniques for approximating non-linear functions. NARMAX-ANN outperforms NARX-ANN because of its better predictive capabilities.