Application of a neural fuzzy model combined with simulated annealing algorithm to predict optimal conditions for polyethylene waste non-isothermal pyrolysis

Ruming Pan (1) , João Vitor Ferreira Duque (2) , Márcio Ferreira Martins (2) , Gerald Debenest (1)

1 IMFT - Institut de mécanique des fluides de Toulouse
2 UFES - Universidade Federal do Espirito Santo

Adaptive neural fuzzy model Simulated annealing algorithm A B S T R A C T In the present study, the waste polyethylene (PE) pyrolysis under different non-isothermal conditions was investigated to estimate the optimal conversions and pyrolysis rates. The pyrolysis study was carried out using Thermogravimetry (TG) of the virgin and the waste PE under different heating rates of 5, 10, 15 and 20 C/min. The TG experiments indicated that the virgin and the waste PE pyrolysis processes mainly underwent in the temperature range of 390-510 C. Subsequently, the adaptive neural fuzzy model was adopted to predict the conversions and the pyrolysis rates of the virgin and the waste PE. The optimal operating conditions in different temperature ranges were optimized by the simulated annealing algorithm (SA). Moreover, the R-squared values of the virgin PE conversions (~1) and pyrolysis rates (> 0.999), and the waste PE conversions (~1) and pyrolysis rates (> 0.999) revealed the high accuracy of the adaptive neural fuzzy model predicted results.

Mots clés: Chemical engineering; Chemical reaction engineering; Environmental science; Energy sustainability; Fuel technology; Waste; Waste treatment-; Non-isothermal pyrolysis; Adaptive neural fuzzy model; Simulated annealing algorithm; Waste polyethylene

Domaines: Mécanique des fluides [physics.class-ph]

Ruming Pan, João Vitor Ferreira Duque, Márcio Ferreira Martins, Gerald Debenest. Application of a neural fuzzy model combined with simulated annealing algorithm to predict optimal conditions for polyethylene waste non-isothermal pyrolysis. Heliyon, 2020, 6 (11), pp.e05598. ⟨10.1016/j.heliyon.2020.e05598⟩. ⟨hal-03292080⟩