engleski

Soft Sensors for Estimation and Control of Refinery Plant Emission

One of common problems in industrial facilities is inability of real-time and continuous measurement of key process variables. As an alternative, the use of soft sensors as process analyzers and laboratory testing is suggested. Soft sensor is defined as an analytical or empirical model, which is used for control and estimation of process variables and properties (e.g. product quality, emission) that are difficult to measure, based on available measurements of input and output variables such as temperature, flow and pressure. This paper is about developing soft sensor models for process control and estimation of H2S and SO2 emissions based on experimental data gathered from sulphur recovery unit. The aim is to remove dangerous components from sour gas streams before they are released into the atmosphere and fulfill strict regulations regarding environmental protection. The soft sensor models were developed by application of neural networks. The best results were achieved with MLPs and neural-fuzzy soft sensor models. Soft sensors are to warn about problems in operation and malfunctions of analyser, to replace the analyser during repair time and servicing, and, at the same time, allow for continously monitoring of sulphur compound emissions.

soft sensor; process identification; neural network; sulphur recovery unit

nije evidentirano