Abstract:Aiming at the problem of fault detection caused by the dynamic characteristic of batch process data, a double weight multiple neighborhood preserving embedding (DWMNPE) algorithm is proposed. Firstly, by finding time neighbors for each sample point, the time correlations between samples are reflected. By defining angle neighbors, sample points are reconstructed to represent the similarity in the space by finding time neighbors, angle neighbors and distance neighbors for sample points. Three different manifold features can fully extract the essential structure of original data. Then, considering the minimum error and three kinds of neighbor order information, a new objective function is constructed to further prevent the loss of neighbor order information when the reconstructing weights are calculated by using NPE algorithm. The data dynamic is solved, meanwhile, the essential local structure is achieved. Finally, the LOF statistic of the dimensionality reduction data is constructed to monitor the process and eliminate the bad influence of data non-Gaussian for monitoring effect. The results of a numerical example and the penicillin fermentation process simulation demonstrate that DWMNPE algorithm is effective for fault detection in dynamic batch process.