Prediction of Manufacturing System Using Improved Infomax Method Based on Poor Information

Xintao Xia, Wenhuan Zhu, Bin Liu

The improved infomax method consists of the bootstrap methodology, the grey system theory, and the information entropy theory. The bootstrap methodology is adopted to imitate the generated information vector of large size by bootstrap resampling from the current information vector of small size, the grey system theory is used to introduce the generated information vector into the grey prediction model for forecasting the future information vector of large size, and the information entropy theory is employed to predict a probability distribution of the future information via the maximum entropy criterion. Information prediction of a manufacturing system is put into practice under the condition of poor information. Case studies of the diameter and roundness of a tapered rolling bearing raceway present that the method is able to make reliably information prediction of a manufacturing system only with the current information of small size and without any prior information of probability distributions.