Elsevier

Fusion Engineering and Design

Volume 87, Issue 12, December 2012, Pages 2002-2006
Fusion Engineering and Design

Design of a real-time fault diagnosis expert system for the EAST cryoplant

https://doi.org/10.1016/j.fusengdes.2012.04.016Get rights and content

Abstract

The EAST cryoplant consists of a 2 kW/4 K helium refrigerator and a helium distribution system. It is a complex process system which involves many process variables and cryogenic equipments. Each potential fault or abnormal event may influence stability and safety of the cryogenic system, thereby disturbing the fusion experiment. The cryogenic control system can monitor the process data and detect process alarms, but it is difficult to effectively diagnose the fault causes and provide operation suggestions to operators when anomalies occur. Therefore, a real-time fault diagnosis expert system is essential for a safe and steady operation of EAST cryogenic system. After a brief description of the EAST cryoplant and its control system, the structure design of the cryogenic fault diagnosis expert system is proposed. Based on the empirical knowledge, the fault diagnosis model is built adopting fault tree analysis method which considers the uncertainty. The knowledge base and the inference machine are presented in detail. A cross-platform integrated development environment Qt Creator and MySQL database have been used to develop the system. The proposed expert system has a fine graphic user interface for monitoring and operation. Preliminary test was conducted and the results found to be satisfactory.

Highlights

► An expert system of real-time fault diagnosis for EAST cryoplant is designed. ► Knowledge base is built via fault tree analysis based on our fault experience. ► It can make up the deficiency of safety monitoring in cryogenic DCS. ► It can help operators to find the fault causes and give operation suggestion. ► It plays a role of operators training in certain degree.

Introduction

EAST (Experimental Advanced Superconducting Tokamak) has been put into operation for seven experimental campaigns. The cryogenic system is one of the subsystems of EAST, which provides enough cold power for cooling of the superconducting magnets and other components. It involves so many process variables and cryogenic equipments that each potential fault or abnormal event may influence the fusion physics experiment or cause economic losses. The cryogenic control system can supervise the process and send out alarm signals when abnormal situations occur. However, it is difficult for operators to immediately analyze the fault causes and present a rational operation suggestion. Therefore, real-time fault diagnosis is critically important for a safe and steady operation of the cryoplant. Using the knowledge-based expert system is a feasible method to realize a time-efficient fault analysis in industrial process like EAST cryoplant which is difficult to exactly model.

The expert systems used for fault detection and diagnosis have been built in many chemical and biochemical process applications [1]. At other fusion facilities of similar complexity, such as Tore Supra, the expert system has been used for task scheduling as well as for overall monitoring and error management in data acquisition system for control and long pulse operation. It eases the fault prevention and their corrections [2], [3]. However, the one applied in helium cryogenic system has been scarcely reported. In this paper, an expert system for real-time fault diagnosis of EAST cryoplant based on the distributed control system (DCS) is developed to facilitate dealing with abnormal situations. Combining knowledge-based fault diagnosis method with real-time process variables monitoring will improve the efficiency and reliability of detecting fault behavior and overall effectiveness of the system [4].

Section snippets

Descriptions of EAST cryoplant and its control system

The EAST cryoplant consists of a 2 kW/4 K helium refrigerator and a helium distribution subsystem. The refrigerator can be separated into a compressor station and a coldbox which contains four turbines, adsorbers, and all of the heat exchangers in it. The distribution subsystem has four circulation pumps for supplying sufficient forced flow of supercritical helium to cool down the coils [5]. The cryogenic process flowsheet and its control network can be referenced in literature [5]. The two major

Structure design of fault diagnosis expert system

The cryogenic fault diagnosis expert system is developed to identify the fault causes and provide operation suggestions when anomalies occur. It uses the expertise of operators in the cryogenic engineering domain, which is accumulated through experience over years, and solves problems imitating the reasoning process of human experts. The proposed structure is presented in Fig. 1. The expert system includes seven main modules: monitoring module, integrated database, knowledge base (KB),

Fault diagnosis modeling

The main knowledge source of the expert system is experience of domain experts and operators. In this work, interviews and analysis of process flow are the most efficient ways to acquire knowledge. The empirical knowledge table, which consists of a symptom, possible causes and suggested actions, is designed to collect knowledge. In total, 176 empirical knowledge tables are collected based on the existing alarm signals in EAST cryoplant. Fault-tree analysis (FTA) has been widely used for

Conclusions

An expert system of real-time fault diagnosis for the EAST cryoplant is designed in this work. C++ symbol processing, object-oriented and relational database technique as well as a novel IDE Qt Creator have been used in developing the proposed system. The technique of knowledge acquisition based on the empirical knowledge table and the convenient knowledge management user interface help us to enlarge the knowledge base and maintain the whole system. The current diagnosis rules and fault trees

Acknowledgments

Financial support from the Knowledge Innovation Program of the Chinese Academy of Sciences (Grant No. Y05FCQ1124) is gratefully acknowledged.

References (6)

There are more references available in the full text version of this article.

Cited by (0)

View full text