Modeling of dielectric behavior of polymers nanocomposites using adaptive neuro-fuzzy inference system (ANFIS)

Abstract The research presents a theoretical study on modeling of dielectric properties polymer nanocomposites based adaptive neuro-fuzzy inference system (ANFIS). contributes better utilizing ANFIS model in the prediction behavior polymers nanocomposites. In this respect, three different samples are trained (Mg 1− x Cu O/PMMA, PPy-DBSA-Y 2 O 3 and PVC/PEMA with [Zn(CF SO ) ]). Inputs obtained from earlier experimental studies. Takagi–Sugeno type is trained. applied weighted average as defuzzification method. optimal network structures, which produce most acceptable results, implemented MATLAB. Six networks to simulate predict permittivity loss terms nanocomposite weight % (0–0.2%, 0–8% 10–30% for each sample, respectively) frequency (10 −2 –10 kHz). simulation results very close their targets. Predictions at weights that measured experimentally testing step predictions other values not implicated data extent achieved. Also, individual points processed using rule viewer. It found provide excellent results. Three-dimensional illustrations represent mapping weight% MATLAB surface Histogram error plot indicate degree noisy. Mean error, mean squared root-mean-squared standard division calculated. Their improve efficiency process. A key goal paper develop mechanism accordance technique achieves purpose. can also form great link between practical domains.