Nowadays, industries worldwide are looking forward to enlarge their profits and become more competitive. A good management is a key factor to accomplish the company’s target, however all management decisions are supported by tools that provide good and relevant
information for the process, which usually influences decision making strategically. Soft Sensors have been applied in industries which are aiming that target and its use has been growing lately. A soft sensor can be adapted to any application regarding variable
measurement, therefore reducing operational costs without compromising the current information quality, and in some cases, better results can be obtained. Since they are software based, they are not subjected to physical damages as real sensors are, so they can be adapted virtually to hostile environments. The key of this kind of sensor success is the use of computational intelligence techniques, which has been heavily used in nonlinear and highly complex process modeling. Currently, many industries already use them successfully, and this work exploits its use with Neural Networks in a chemical process in an important Brazilian Aluminum Smelter whose control is very hard to maintain once it is not easy to retrieve information from the plant due to its corrosive nature and whose measurements require some operational resources. The usage of soft sensors within it may reduce costs and delays of measures drastically. A case of use of the soft sensor for temperature measure is presented on this work, since its design through implementation at production, according
to a researched methodology.
|
