Fault Detection Method Based on the Monitoring State Synchronization for Industrial Process System

Abstract

Currently, with the requirements of quality, safety and service in many modern industrial process systems, fault detection and diagnosis become a significant issue to ensure the high control performances. Under these circumstances, this paper presents a fault detection method based on the monitoring state synchronization to perceive and evaluate the system operation situation. This model firstly considers the system states measured by some sensors as the nodes and defines the distributed monitoring state network with changing dynamically over time, all of which have been given their definition, calculation method and actual physical meaning, and finally its synchronization state of distributed network can be employed to achieve fault detection. Furthermore, this method can be used to provide a novel and feasible research method for the global assessment of the operation states and for monitoring the local operation fault of modern industrial process systems. The application of this method on a simple multivariate process system example shown that it can not only track the operation state of the whole system well to detect the fault, and not only monitor the situation of each distributed network node in real time to achieve fault diagnosis, but also make use of the correlation between the network node states to effectively locate the system operation fault.