In this paper we present an approach to detecting and locating defects based on an indirect supervision using Principal Component Analysis (PCA) and Average Dynamics of Metrics (ADM). Our contribution lies in the combined use of the fuzzy logic technique for detection and the ACP for the diagnosis method. In a unity of textile spinning, we have defined some indicators of product quality. ADM permits the valuation of the product quality to different levels of the manufacturing procedure. An indication of calculated detection from the last principal components is used to detect defects. Once a defect is detected, its variable tally is located by calculating its contribution.