Development of a Multi-block Modified Independent Component Analysis based Process Monitoring Strategy

The article highlights the Development of a Process Monitoring Strategy for a multi-stage manufacturing facility laden with non-normal data. A multi-block variant of the modified independent component analysis (MICA)-based technique has been proposed for building of the said monitoring strategy. For validation of the monitoring strategy thus developed, a case study pertaining to a Copper Cathode Manufacturing Unit (CCMU) has been taking into account. The multi-block MICA (MBMICA)-based monitoring strategy provided a means for hierarchical monitoring and effectively negotiating with non-normal data for the chosen multi-stage CCMU. The detection of faults were achieved by employment of MBMICA score-based Hotelling T2 control chart whose control limit was estimated via application of bootstrap procedure. The fault detection was succeeded by fault diagnosis which was performed via application of appropriate fault diagnosis statistic. The said process monitoring strategy thus developed was able to detect and diagnose the detected fault(s) with appreciable accuracy for a multi-stage CCMU laden with non-normal data.