Abstract: Based on control system characteristics for high-dimension, coupled, and redundant data, we propose a new method that combines dynamic principal component analysis with a weighted fuzzy C-mean clustering algorithm. By considering the system's dynamic characteristics, the data dimensions are reduced. We use the principal components as weights and discuss the degree that different characteristics contribute to the system. We use the fuzzy C-means clustering algorithm to obtain the clustering center of normal data and establish the weight difference model using the fault data for detection in the control system. The experimental results show that the accuracy and effectiveness of control system fault detection can be improved by this method.