过程运行数据的稳态检测方法综述

Overview on the steady-state detection methods of process operating data

稳态检测对设备性能评价、运行优化、系统建模和故障检测均具有重要意义,但由于检测方法各具特性,使得各种方法的应用场合不尽相同,目前仍然缺少对实际应用中方法选择的理论指导。因此,根据现有稳态检测方法所采用基础理论的不同对其进行综述,并对每种方法的优缺点进行了分析。从多变量检测能力、趋势定量表征、检测区间、在线检测能力、抗噪能力及可靠性6个方面对已有检测方法进行了分析、对比和总结。对检测方法在实际工业过程中的应用起到了一定的借鉴作用,并为现有方法的改进及新方法的提出指明了方向。

Steady-state detection has great significance to equipment performance evaluation, operation optimization, system modeling and fault detection. However, various detection methods have different application situations because of the differences of the detection methods, and there is a lack of theoretical guidance for selecting proper detection method in practical application. In this paper, existing steady-state detection methods are reviewed according to the differences of their adopted basic theories, and the advantages and disadvantages of various detection methods are an- alyzed. The existing detection methods are compared and summarized in six aspects, which include multivariate detec- tion ability, trend quantitative characterization, testing interval, online detection ability, noise immunity ability and re- liability. The overview provides certain reference for the practical applications of the detection methods in practical in- dustrial processes, and points out the directions for improving existing methods and proposing new methods.