Internet-based expert system for monitoring and control of chemical processes

An innovative Inverse Response and Fault-diagnosis Expert System with the capability of being performed on the Internet for remote monitoring, advanced control, and productivity improvement of chemical processes has been developed and tested. The Inverse Response Expert System increases the performance of certain control systems significantly by forcing standard controllers to take correct control action when inverse response occurs. The Fault-diagnosis Expert System along with multimedia data management detects and predicts failures in a timely manner and provides an effective troubleshooting procedure for abnormal operating conditions. The real-time Internet-based application, which integrates the Inverse Response and the Fault Diagnosis Expert System, greatly improves productivity and decision-making quality of existing control systems by allowing control engineers to remotely monitor and perform adjustments on the process.; The Expert System encodes the knowledge-base of a chemical process using a set of rules and an inference engine. The inference engine uses forward and backward chaining techniques for deterministic cases such as inverse response, and the Bayesian theory that predicts the probability of the truth of some hypothesis, for heuristic cases such as fault-diagnosis. In the event the inference engine identifies inverse response, a compensation term is dynamically computed based on inverse response models and then added to the standard controller output to cancel its wrong control action. Since the inference engine for fault-diagnosis uses the Bayesians theory for inexact reasoning that is based upon probabilities, it detects and predicts abnormal operating conditions. Then, by using multimedia data management techniques the Expert System provides an effective troubleshooting procedure.; The proposed Internet-based control system has been designed to integrate the expert system through a multi-tier scheme using client/server architecture. Thus, clients using a web browser assisted with a graphical user interface can remotely access, supervise and perform real-time changes and adjustments on the local control system.; Implementing the Expert System on a 75-kLbs/h coal-fired boiler unit where the drum level exhibited inverse response tested the proposed strategy. As a result, the new level controller performance was significantly improved. A second test was performed on a binary distillation column but this time along with the Internet-based application. A variety of abnormal operating conditions were successfully resolved and an improved supervision and control from remote sites has resulted.