A FRACTAL ORIENTED UPGRADING OF RELIABILITY AND SAFETY KNOWLEDGE OF REALISTIC CHEMICAL-ENGINEERING PROBLEMS

Reliability and safety knowledge is very sparse. Its realistic and obj ective application is nearly impossible. Additional knowledge is neede d to increase the reasoning power of reliability and safety knowledge bases. The conventional fractal analysis has been used for the study o f chaos in physical systems. Its possible role in the evaluation of re liability/safety knowledge bases is studied in this paper. The only pr econdition for the application of fractal analysis is an ability to di stinguish between specific and general knowledge items. This enables u s to detect a level of inconsistency between mostly subjective additio nal knowledge items and the existing knowledge bases. The fractal anal ysis can characterise the knowledge base as one integrated complex. Ho wever, knowledge acquisition requires ''local'' evaluations as well. T herefore a discriminative analysis is used. A realistic man-machine di alogue (an evaluation of the mean time between failures of control val ves) supported by the fractal and/or discriminative analysis is presen ted.