Unified Study on a Computer-Aided Maintenance System

对该算法进行了改进，利用化工过程中状态变量之间的延迟信息，获得了令人满意的诊断结果。开发了基于高阶统计量的工艺数据分析方法。(7)提出了一种确定计算机辅助维修系统部件作业合理分配的方法

(1) Two types of CUSUM method called Crosier Method and Healy Method have been successfully applied to process monitoring problems with a large number of noisy variables.(2) A combinatorial algorithm has been developed to select rationally periods and kinds of works for preventive maintenance of the equipment whose breakdown might give rise to a fatal accident.(3) Several novel methods to detect the propagation of fouling or corrosion have been developed by using clustering and ICA techniques. Synthesis of operating procedures to avoid faulty equipment is also considered.(4) A risk assessment system has been developed for supporting safety design of batch plants, which gives us qualitative risks based on the degree of severity and the frequency of events, as well as makes clear all accident scenarios that are likely to occur in batch processes.(5) An algorithm for searching multiple origins of failures by use of the signed digraph has been proposed. The algorithm has been improved so as to obtain diagnostic results of satisfactory accuracy using information about delays among state variables in a chemical process. The usefulness of this algorithm is demonstrated by its application to a Fluidized Catalytic Cracking Unit dynamic simulator.(6) Countermeasure planning algorithms has been developed to recover the abnormal situations in which fault diagnosis will be finished when a common countermeasure for each candidate can be found. Process data analysis based on high order statistics is also developed.(7) A method has been developed to determine rational allotment of component activities in a computer-aided maintenance system