APPLICATION OF ADAPTIVE NEURO FUZZY INFERENCE SYSTEM TO MODELING OXIDATIVE COUPLING OF METHANE REACTION AT ELEVATED PRESSURE
<span style="font-family: Calibri;"><span style="font-size: medium;">The oxidative coupling of methane (OCM) performance over Na-W-Mn/SiO<sub>2</sub></span><span style="font-size: medium;"> at elevated pressures has been simulated by...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Reaserch Institute of Petroleum Industry
2014-10-01
|
| Series: | Journal of Petroleum Science and Technology |
| Subjects: | |
| Online Access: | https://jpst.ripi.ir/article_400_4276561ee5ff6f8a5bae33d9be07eaa3.pdf |
| Summary: | <span style="font-family: Calibri;"><span style="font-size: medium;">The oxidative coupling of methane (OCM) performance over Na-W-Mn/SiO<sub>2</sub></span><span style="font-size: medium;"> at elevated pressures has been simulated by adaptive neuro fuzzy inference system (ANFIS) using reaction data gathered in an isothermal fixed bed microreactor. In the designed neuro fuzzy models, three important parameters such as methane to oxygen ratio, gas hourly space velocity (GHSV), and reaction temperature were considered as inputs and methane conversion and the selectivity of product hydrocarbons (C</span><sub><span style="font-size: medium;">2+</span></sub><span style="font-size: medium;">) were chosen as outputs. Two five-layer neuro fuzzy models based on the partitioning algorithm were designed at each reaction pressure to predict the product hydrocarbons (C</span><sub><span style="font-size: medium;">2+</span></sub><span style="font-size: medium;">) selectivity and methane conversion separately as a linear combination of inputs by the optimal selection of number and type of the membership functions. Moreover, to evaluate the ability and accuracy of the developed neuro fuzzy models in the prediction of OCM reaction performance, the results of ANFIS models were compared with experimental data and artificial neural network outputs. The comparison was carried out by the calculation of some statistical parameters such as correlation coefficient (</span><span style="font-size: medium;"><em>R</em><sup>2</sup></span><span style="font-size: medium;">), mean squared error (MSE), and average relative deviation (ARD). The results show that there are excellent agreement between model predictions and experimental data and the proposed ANFIS model can predict the methane conversion and product hydrocarbons (C</span><sub><span style="font-size: medium;">2+</span></sub><span style="font-size: medium;">) selectivity under different operating conditions by high accuracy.</span></span> |
|---|---|
| ISSN: | 2251-659X 2645-3312 |