Trend recognition of process data of a refinery using wavelets

摘要

The goal of the project "Forecasting of state parameters in refineries", which is supported by the Federal Ministry of Education and Research of Germany, is to reduce the emission of a refinery by use of intelligent methods and logic instruments for analysis of the refinery-wide gas-network using the real-time expert system G2. Excxeptional production events and changes in continual use of the systems often lead to an increased emission of fuel gas. A premature trend recognition with event localisation and representations to exploit alternatives of intervention is especially necessary if the attention of the production staff is turned towards malfunction in primary production processes. A trend recognition of process data using wavelets is described in this paper.