Corrosion resistance characteristics and predictive modeling of electrochemically polished medical NiTi shape memory alloy milled surfaces

The NiTi shape memory alloy is widely employed in medical applications owing to their exceptional corrosion resistance and biocompatibility, with the milling-electropolishing process being a standard surface treatment approach. This study employed a three-factor, three-level Box-Behnken experimental design based on response surface methodology (RSM) to systematically investigate how electrochemical polishing parameters (current density, electrode gap, and polishing time) influence corrosion characteristics including corrosion potential, current density, and polarization resistance in simulated body fluid (SBF). The BP neural network model, developed using full factorial experimental data, provides quantitative predictions of corrosion current density from process parameters. Results demonstrate that current density predominately governs the corrosion characteristics, while the electrode gap-polishing time interaction exhibits secondary influence. The optimized BP neural network (single hidden layer with four neurons, tansig-purelin transfer functions, trainlm training algorithm) achieves exceptional prediction accuracy ( R 2 = 0.9969) for corrosion current density, offering both theoretical guidance and technical support for optimizing surface treatment processes and enhancing the service reliability of NiTi alloy.