Browsing by Author "Levanto, Santeri"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Data driven modelling for viscose quality characterisation: a machine learning approach
    (2019-03-12) Levanto, Santeri
     Kemian tekniikan korkeakoulu | Master's thesis
    Demand for textile fibers is increasing, and cellulosic man-made fibers can be utilized as an alternative substance for oil-based end products in textile industry. To compete with oil-based products, a more accessible quality characterization could be helpful. The aim of this study is to examine the possibilities of a machine learning method called Random Forest in the viscose fiber production and to find out, if the machine learning method Random Forest is applicable for the viscose quality modelling. This is due to traditional regression methods such linear regression not having been successfully applied for the quality characterisation. The study consists of literature review and an applied part. The literature review considers dissolving pulp and viscose production as well as machine learning and more precisely an algorithm called Random Forest. The applied part consists of data analysis, data handling and other methods required in order to achieve the most accurate Random Forest model. The study shows, that the Random Forest algorithm has a potential to model the quality behaviour, especially in comparison to traditional linear regression. The Random Forest model can predict with 95% confidence if the viscose quality classifies as good or bad, but the numerical prediction for the quality parameter has a large error margin for the 95% confidence. It is suggested, that the error margin could be lower, if the utilized data was whole and the number of data points was larger.
  • Kuitutuotteiden hydrofobiliimaus: menetelmät ja käyttösovellukset
    (2016-05-08) Levanto, Santeri
     Kemiantekniikan korkeakoulu | Bachelor's thesis