Prediction of polishing roughness of aluminum alloy hard anodized film based on BP and GA-BP neural network

Abstract

In order to obtain better-polished surface quality of hard anodic oxide film, two deep learning models of the BP neural network and GA-BP neural network were used to establish a roughness prediction model for aluminum alloy hard anodic oxide film polishing. The experimental data was divided into two groups, one group of data was used for model training, and the other group of data was used for model testing. The results showed that the mean square error between the polishing roughness predicted by the BP neural network model and the experimental results was 1.33E-2, the maximum relative error was 18.84 %, the minimum relative error was 0.77 %, and the average relative error was 10.46 %. The error is relatively large, and the degree of variation of the error is relatively large; the mean square error of the polishing roughness predicted by the GA-BP neural network model and the test results is 0.58E-2, the maximum relative error is 14.28 %, the minimum relative error is 0.51%, the average relative error is 6.61 %, the error is smaller, and the degree of error change is smaller; the prediction accuracy of the GA-BP model is the highest, and the generalization ability strongest.