Statistical Methods and Mathematical Modeling for On-line Monitoring of Critical Control Points

Methods and software have been developed for statistical monitoring of CCPs and fault diagnosis by using multivariate statistical techniques. These methods indicate accurately and quickly the existence, magnitude and occurrence times of critical changes in process operation and probable causes for abnormal operation. The developed software integrates and automates monitoring of CCPs and determines the probable source causes for abnormal behavior in real-time, in a production environment. These methods will facilitate the determination of target values based on historical process data, monitoring of critical limits, and enable the diagnosis of most likely causes of abnormal process operation. <P> The methodology and software developed were tested first by using simulation studies based on dynamic process models, then by experimental studies in pilot scale experimental systems and industrial processes. Data were collected from the high temperature, short time pasteurization pilot plant located at the NCFST, and the Armour Swift-Eckrich Brown-N-Serve plant.Methodology for complementing the intermittent CCP monitoring of internal sausage temperature was developed by monitoring of external sausage temperature using an infrared thermometer. This system measures continuously the external sausage temperature and is able to detect changes that may indicate an out-of-control process and a violation of the CCP. Using the developed programs, the operating conditions of the process can be continuously monitored and excess waste of product can be reduced. The process monitoring tools are being integrated to a real-time knowledge-based system that provides a supervisory level for coordinating various activities and additional tools for fault diagnosis.