Process Optimization Studies Based on BP Neural Network in Electroless Plating of Ni-Fe-Co-P on Carbon Fiber
It presents a method to predict and optimize the electroless plating process, and to compare the predictive ability of the network to the experimental results. It combined with the neural network and orthogonal experiment and used a small step searching method to optimize the chemical process of plated Ni - Co-Fe-P on carbon fiber within the scope of the process parameters, got more optimized process recipe: the temperature is 88°C, ratio of the main salt concentration is 0.46, the concentration of sodium citrate is 46 g/L, and the pH value is 9.03, the concentration of sodium hypophosphite is 24 g/L. Through validating with experiments, the error between them is 2.39%, the linear correlation between the method of calculation and experimental program of the target is very good, and the correlation coefficient R =0.99943 which indicated that the training results are reliable and the BP neural network optimizing the process recipe is indeed feasible.