A prototype system, which detects wrinkling and tearing in sheet metal parts, has been developed. A spatial variation of the blankholding force (BHF) is achieved by hydraulic adjustment of the force in the BHF pins which are subdivided into groups. Neural networks predict the presence of wrinkles. Tears are predicted from the punch force signals using feature extraction techniques. A fuzzy logic controller (FLC) recommends changes to the BHF and lubrication to improve the quality of the next part. The results obtained when drawing the back panel of a washing machine are discussed.