A Machine Learning Based Approach to Analyze the Relationship Between Process Variables and Process Alarms

With the advancement of technology, industrial plants are adopting advanced process control systems and sensors to optimize operations. Alarm systems are utilized to manage the process and ensure safe and reliable operations. Although alarm systems can sometimes produce overhead to the operators if not managed properly, alarm systems can contribute to economic, environmental, and safety added value in industrial sectors. The relationship between alarm systems and process sensors may not be clearly known in complex plants. In this paper we propose an approach to iteratively improve the plant process management by quantifying the influence of process variables on the triggered alarms using machine learning and AI tools. This approach thus contributes to root cause analysis, continuous improvements and effective process controls. The proposed approach groups related alarms by analyzing times series alarm data and process data. In addition, the approach quantifies the influence between the process variables and related alarms triggered.