Quantitative study of thermal barrier models for paper-based barrier materials using adaptive neuro-fuzzy inference system

Abstract

Abstract A composite silicone emulsion-biomass polymer paper-based barrier coating material with high barrier performance was prepared by double-layer coating, and the material was tested for oil repellency. The composition-structure-property data set of the paper-based barrier materials was constructed based on the experimental data. An adaptive neuro-fuzzy inference system (ANFIS) was used to construct a prediction model of the coating structure in high-temperature environments to achieve quantitative analysis of the barrier performance in high-temperature environments. The ANFIS prediction model was constructed based on two algorithms, the grid partitioning algorithm and the subtractive clustering algorithm, and the accuracy of the model determined by the two algorithms was compared for training, validation and testing of this experimental data. The results showed that the prediction model of the grid partitioning method had a better fit with the experimental data, with a root mean square error (RMSE) value of 7.00383 and a R-squared (R 2) of 0.9644 between the model prediction data and the actual data.