Failure Detection and Recovery in the Batch Process Automation Domain Using Automation Agents

Abstract

The process industry needs to cope with permanently changing environments where sudden component failures can significantly influence the system performance if not treated in an appropriate amount of time. Agent-based manufacturing control systems are recognized as a promising technology that can handle such anomalies. However, current agent approaches are lacking real-time capabilities and solutions that support distribution and agility. We address this issue by introducing an automation agent architecture that consists of the low level control layer, which is based on IEC 61499 and responsible for directly controlling the physical component, as well as of the high level control layer which includes a world model repository and is responsible for organizational and supervisory functions. The world model serves as an accurate internal representation of the agent’s external environment and is used to detect anomalies in the environment. In this paper we present the monitoring and diagnostics abilities of the introduced automation agent architecture within the batch process domain.