A Salt Spray Corrosion Prediction Model for Aviation Aluminum Alloy Based on MVO-GRNN Algorithm

Abstract

It is very important to focus on the corrosion failure of aluminum alloy materials for airframe because of the increasing of aircraft service time. However, due to the long corrosion test cycle, the number of samples for processing alloy corrosion related data with high cost is very small. In this paper, a corrosion rate prediction model for less sample data sets is proposed: a General Regression Neural Network which using Multi-Verse Optimizer to optimize prarmeters in order to improve accuracy. In this paper, according to the public data of China Corrosion and Protection Network, the content of Mg, Cu, Fe and other components in different types of aluminum alloys, the corrosion environment (Cl− concentration, pH value) and outdoor exposure time are taken as the input and the mass loss per unit area of aluminum alloys is taken as the output of the model. Moreover, using radial basis function for comparsion. The experimental results show that the model has a good fitting effect for predicting the salt spray corrosion of aviation aluminum alloy, and its performance is better than the traditional RBFnetwork.