Optimization of environmental and medium parameters is an important step for bioprocess engineering. In the present study, the efficacies of Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) were compared for their utility in estimating biosurfactant production, surface tension reduction, and emulsification under different environmental and medium parameters. Biosurfactant was collected from the bacterial isolate FKOD36. In these models, temperature, pH, incubation period, carbon, nitrogen, and hydrocarbon sources were used as input variables, whereas surface tension reduction, emulsification index, and biosurfactant production were the dependent output variables. Models were trained for six inputs and three ANFIS sub-networks were developed for each output. Each of three ANFIS models was then used to predict one of the three outputs. The performance indices of both ANN and ANFIS illustrate that proposed ANFIS network produced better results with coefficient of determination (R2) values ranging from 0.96 to 0.99 for the training dataset and 0.90–0.99 for the validation dataset as compared to ANN which had R2 values of 0.95–0.99 for the training set and 0.89–0.98 testing set. Based on the results, the multilayer ANFIS model with its fuzzy application rules proved to give better prediction results than the ANN model.
|Journal||Stochastic Environmental Research and Risk Assessment|
|Number of pages||11|
|Publication status||Published - 1. Jan 2016|
Bibliographical noteFunding Information:
The authors gratefully acknowledge support from the Higher Education Commission of Pakistan for providing an IRSIP Fellowship for the lead author and financial support for the project. The authors also thank Tehran University for its support of this project.
© 2015, Springer-Verlag Berlin Heidelberg.
- Adaptive neuro-fuzzy inference system
- Artificial neural network
- Biosurfactant production