Relational model of compositions of enamel coatings and sintering temperature based on back propagation neural network algorithm

Abstract

Declining the sintering temperature of enamel coatings is traditionally by reducing the content of network-forming agents (such as SiO2) and increasing the content of flux (for example, alkali metal oxides), which will lower the density of the [-Si-O-] network structure, and adversely affect the corrosion resistance of enamel coatings. To ensure a low sintering temperature of enamel coatings as much as possible under a high SiO2 content, it is essential to develop novel enamel coating formulas. However, abundant compositions of enamel coatings cause enormous verification workloads. To address this issue, this paper establishes a relational model of the compositions of enamel coatings and the sintering temperature using the back propagation (BP) neural network algorithm and proposes a formula that meets the requirement of lower the sintering temperature under a high SiO2 content. When the SiO2 content is 55 %, the sintering temperature can be diminished to around 400 °C. Furthermore, the influences of common enamel components on the sintering temperature are identified in this paper using the sensitivity algorithm. The obtained general law of formulas is that the total weight value of the formula at low sintering temperatures should be around −0.3. This paper can provide valuable theoretical guidance for subsequent research and expand the application range of enamel coatings.