@article { author = {Mohammadi Doust, Akbar and Rahimi, Masoud and Feyzi, Mostafa}, title = {Prediction and Optimization of the Effects of Combining Ultrasonic Waves and Solvent on the Viscosity of Residue Fuel Oil by ANN and ANFIS}, journal = {Physical Chemistry Research}, volume = {4}, number = {3}, pages = {333-353}, year = {2016}, publisher = {Iranian Chemical Society}, issn = {2322-5521}, eissn = {2345-2625}, doi = {10.22036/pcr.2016.14578}, abstract = {In the present work, the influences of temperature, solvent concentration and ultrasonic irradiation time were numerically analyzed on viscosity reduction of residue fuel oil (RFO). Ultrasonic irradiation was applied at power of 280 W and low frequency of 24 kHz. The main feature of this research is prediction and optimization of the kinematic viscosity data. The measured results of eighty-four samples, including 336 data points, were developed by artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS). The ANN predictions were also compared with the ANFIS approach by means of various descriptive statistical indicators, including absolute average deviation (AAD), average relative deviation (ARD) and coefficient of correlation (R2). The AAD and R2 of the developed ANN model for kinematic viscosity prediction of overall set were 0.0107 and 0.99384, respectively. On the other hand, for ANFIS approach, the AAD of 0.02112 and R2 of 0.99279 were attained. Although accuracy and precision of the ANN model were more than the ANFIS approach, it has been illustrated that the proposed ANN and ANFIS models have a superior performance with acceptable errors on the RFO kinematic viscosity estimation. Findings of this research clearly revealed that the neural network and neuro-fuzzy approaches could be successfully employed for prediction and optimization of kinematic viscosity of RFO and high viscosity materials in oil processes.}, keywords = {residue fuel oil,Kinematic viscosity,Ultrasonic irradiation,Artificial Neural Network,adaptive neuro-fuzzy inference system}, url = {https://www.physchemres.org/article_14578.html}, eprint = {https://www.physchemres.org/article_14578_f09955b146252e4ef20285867fde8370.pdf} }