Modeling and Predicting the Important Properties of the PVC/Glass Fiber Composite Laminates in the Production Process by the TLBO-ANFIS Approach

In this paper, by considering the temperature, time, and process pressure, as the most important factors in producing the thermoplastic composites, an experimental design was performed. An adaptive neuro-fuzzy inference system (ANFIS) was utilized to estimate the important characteristics containing flexural strength, porosity volume ratio, fiber volume ratio, and flexural modulus. Then, the parameters of the ANFIS network were optimized by the teaching-learningbased optimization (TLBO) algorithm. For the purpose of modeling material behavior in the process, the experimental results were utilized for the training and validation of the adaptive inference system. The accuracy of the obtained model has been investigated by using different graphs, based on the statistical criteria of the mean absolute error, correlation coefficient, mean square error, and the percentage of mean absolute error. Based on the obtained results, the TLBOANFIS approach has been very effective in estimating the above-mentioned properties in the production process. The network error for estimating flexural strength, porosity volume ratio, fiber volume ratio, and flexural modulus in the teaching section was equal to 0.159%, 0.0003%, 1.074%, and 0.0001%, and the corresponding values were equal to 0.852%, 42.413%, 33.95%, and 4.894% in the testing section.