2024
Tezis
Oil refining is an energy-intensive process that continuously separates petroleum products. Refining companies are interested in products that meet all regulations and the efficiency of their production. It is possible to achieve cost minimization and increase the quality of the output product by using accurate soft sensors that can reliably predict the quality of the output product in real time. In this regard, research on the construction of reliable soft sensors in industrial production is necessary and relevant. To improve the estimation capability, the alternating conditional expectation algorithm, which is based on nonparametric optimal transformations, was used for nonparametric soft sensor design. As a result, the designed nonparametric soft sensor shows a better efficiency in predicting the quality index of the target distillation product with a significantly reduced root mean square error compared to multiple linear regression analysis.