Improvement of radial basis function neural network with accelerated particle swarm optimization for corrosion rate prediction of 3C steel in seawater environment

For accurately prediction of 3C steel corrosion rate in seawater environment, this paper establishes a radial basis function neural network (RBFNN) and improves it with accelerated particle swarm optimization (APSO). Specifically, the centers, spreads and connection weights of each radial basis function (RBF) were automatically tuned by the APSO, and the number of RBFs in the RBFNN was minimized by choosing a special fitness function. The APSO-optimized RBFNN was proved through a case study to have good prediction accuracy and self-learning ability. The research findings provide an accurate, adaptive and easily-to-train prediction model for 3C steel corrosion rate in the seawater environment.