Intelligent monitoring of sheet forming process

In the sheet forming process, tight QC requirements and strict economic objectives make it necessary for factory to quickly identify sheet defects and take corrective actions. A framework of sensor-based intelligent monitoring system is suggested to perform online monitoring and control of cold rolled strip forming. For successful implementation, a backpropagation neural monitor is employed to recognize the defects of the cold rolled strip and generate appropriate control strategy. The monitoring system first deals with a large amount of raw process data detected by stress sensors. From such real-time data, interesting and important features, stress series are extracted and normalized. The stress series are then trained by the neural monitor for identifying the defects, such as left slope, right slope, central buckle, edge wave, quarter-wave and compound wave et al. The output of the neural monitor will activate corresponding feedback control actions such as CVC shift, screws, adjustment of bending pressure and selection of cooling sprays. To improve the performance of the neural networks, optimal learning parameters are employed to train the neural networks. The results of a case study have shown that the output of the defect recognition is well matched to the practical situation, and have given encouragement to further improvements of the intelligent monitoring system