Risk analysis of hitech membrane companies based on predictions of future prices

conference paper

English

A classical quantitative (analytical and/or statistical) analysis is not appropriate if ill structured, vague and complex cost engineering problems are studied. A fuzzy model can utilise semi-qualitative and semi subjective data sets which are partially inconsistent and subjective. To minimise the information loss of information primary knowledge items (e g. company cost files) are used directly by elimination of the conventional pre processing (e.g. the best-fit approximation) of primary data. There are two different types of cost information items available, namely absolute and relative. An absolute information item is characterised by a vector of parameters (e.g. a stainless steel high-pressure heat exchanger, heat transfer area XY costs 10 000 USD (1980)). A relative knowledge item represents a difference. It can be characterised by two or more absolute items. A relative item is usually represented by a general and inaccurate heuristics (e g. a labour cost in Germany is considerably higher then a Russian labour cost). An absolute item can be characterised by a multidimensional fuzzy set or by a fuzzy conditional statement. A flexible integration of specific conditional statements and general meta heuristics is proposed to increase the reasoning and discriminative power of cost fuzzy knowledge bases. Each relative item is used to generate a set of absolute items. The final result is a set of absolute items represented by a set of fuzzy conditional statements. The set of statements can be easily treat by any fuzzy reasoning algorithm. A case study (complexity factor evaluation, 37 fuzzy statements) is given in full detail.

risk, hitech company, fuzzy

KUČEROVÁ, V., DOHNAL, M.

1. 1. 2002

Česká společnost chemického inženýrství

Praha

80-86059-33-2

Chisa 2002