The widespread availability of Distributed Control Systems (DCS) not only provides the framework for advanced control applications but also greatly facilitates the continuous monitoring of chemical processes to maintain safe and profitable plant operations. In most facilities, the plant operators are asked to manage the operation in such a way as to ensure optimal production levels, while attending to occasional alarm situations that may result from equipment malfunctions. It is critical to identify such abnormal situations in a timely manner as there may be a potential for a safety hazard that may affect not only the plant and its personnel but also the surrounding communities. Most operators traditionally relied on personal expertise for such a task, and in some cases, the events exceeded the capabilities of any human operator, thus leaving the plant vulnerable to costly shutdowns and, in the worst case scenario, to possibly fatal accidents [216]. Today, human expertise is complemented by computerized support systems that comprise various data analysis and interpretation strategies that can provide guidance to the plant personnel for handling abnormal situations. The key component of such a system is fault detection and diagnosis (FDD) that monitors the occurrence of process failures and identifies their root causes.

This section builds on the techniques described in Sections 6.2 and 6.3 to offer a strategy for process trend analysis (Figure 7.1). The problem can be stated simply as follows: given a set of known models of the process operating conditions, determine the likelihood of a new set of observations.