According to different manufacturing strategies, modern chemical processes always have multiple modes. At the same time, variables within the same mode often follow a mixture of Gaussian and non-Gaussian distributions. In this study, an algorithm using multisubspace principal component analysis (MSPCA) with the local outlier factor (LOF) technique is proposed for process monitoring. Unlike conventional clustering methods, which require iterative processes, a new clustering strategy based on serial correlation and the LOF method is developed. To decrease the complexity of process analysis and simultaneously preserve information, a two-step principal-component selection scheme called full variable expression (FVE) is proposed in this article. Moreover, for the mixed distribution of a single mode, a monitoring statistic is established using LOF in the feature subspace. Then, the results in all feature subspaces are integrated through the Bayesian fusion strategy. Finally, to verify its superiority, the MSPCALOF scheme is applied to the Tennessee Eastman (TE) benchmark problem and a continuous stirred-tank reactor (CSTR) process.