Aiming at the multimode process data which follow complex distribution in industrial process monitoring

this paper proposes a fault detection method based on local entropy component analysis (LECA) algorithm. In order to deal with the multimode characteristic of operating data

k nearest neighbor Parzen window (KNN-Parzen) method is used to estimate the local probability density of each sample. Then

a local relative density estimate function is constructed to decrease the sensitivity to window width parameter. To effectively extract the feature information hidden in the non-Gaussian data

the local entropies of process data are calculated by using information entropy theory. Independent component analysis (ICA) is applied to establish the local entropy component statistic model for fault detection. The simulation results of numerical example and continuous stirred tank reactor (CSTR) system indicate that LECA can show superior performance in process monitoring.