Research on a prediction model for gas insulation performance based on Pareto optimisation

Abstract

Abstract Predicting the insulation performance of SF6 substitute gases through gas molecular structures has been a popular topic worldwide. The difficulty is that the relationships between the molecular structure and the gas insulation strength, global warming potential and boiling temperature are not clear, and general linear methods cannot be used to effectively extract the key factors. Based on published molecular structure parameters, the grey correlation method is used to extract the factors that affect the gas dielectric strength, global warming potential and boiling temperature in a dynamic (non‐linear) approach. Further, to predict the dielectric strength, global warming potential and boiling temperature of gases, a linear regression method and the factors with high correlations are used as independent variables. Through the Pareto optimal solution, the dielectric strength is set as the target, the global warming potential and boiling temperature are set as the constraints, and the ranges of the molecular structure parameters of the SF6 substitute gas are obtained. This research study provides an important reference regarding the SF6 substitute gas analysis and provides a research foundation for the design and synthesis of new environmentally friendly gases used in power equipment.