Data-driven monitoring of multimode continuous processes: A review

The Internet of Things benefits connectivity and functionality in industrial environments, while Cloud Computing boosts computational capability. Hence, historical data from processes allows developing automatic strategies for improving efficiency and security. Automatic process monitoring tasks are important due to the complexity of real processes where multiple operating conditions must be considered for achieving a satisfactory performance. In the last 20 years, over 100 publications on data-driven monitoring for multimode continuous processes have been released. Therefore, this work provides a review on this topic. The data-driven modeling problem for monitoring multimode continuous processes is introduced. The analysis of the clustering methods and monitoring approaches is the main contribution of the review. This study includes advantages and drawbacks of every analyzed strategy. Finally, promising research directions towards the Industry 4.0 and the Big Data era are discussed.