Prediction of shield machine tool failure and control strategy design based on ant colony optimization and neural network

It is important to cut down the erection time and the operation guidance by studying the shield machine tool failure. In this paper, an ACO-BP algorithm based tool failure prediction model is established by utilizing the nonlinear mapping characteristics of neural network and mining data characteristics from the subway. According to the practical problems, the dependent variables and the independent variable are systematically selected and compared. The wear and life are chosen as the independent variables in the light of the effect of the cutting tool failure. The factors under different targets combined with the actual situation are selected as the independent variables by applying the principle component analysis and gray correlation. After that, the ant colony algorithm is used to reduce the problem of low efficiency because of ant colony large-scale iterations. The final weights, threshold, and the number of hidden layer nodes by properly adjusting the feedback link. Finally, the analysis results show that the ACO-BP prediction model is effective and accurate. In addition, the control strategy for the abnormal damage is given.