Efficient Soft Sensor Modelling for Advanced Manufacturing Systems by Applying Hybrid Intelligent Soft Computing Techniques

Ali Al-Jlibawi (Universiti Putra Malaysia Selangor, Malaysia); Ts. Ir. Mohammad Lutfi Othman (Universiti Putra Malaysia & Advanced Lightning, Power and Energy Research (ALPER), Malaysia); Muayed S AL-Huseiny (University of Wasit, Iraq); Ishak Bin Aris (Universiti Putra Malaysia Selangor, Malaysia); Samsul Bahari Mohd Noor (Coauthor & Universiti Putra Malaysia, Malaysia)

Data-driven soft sensor are inferential models that use on-line available sensor to predict quality variables which cannot be automatically measured at all, or can only be measured at high cost, infrequently, or with high cost and delay (laboratory analysis or on-line analyzer). In soft sensors development, the main issues should deal with treatment varying data, selection of input variables, model training, validation, and soft sensor maintenance to adopt the heavy duty of oil refineries in the aim to improve products and increase yield. In this research improvement on virtual sensor on hybrid soft computing methods fuzzy logic system and neural network which employ to construct the modelling and use rough set theory and differential evolution. This study will work on data of refining of crude oil for two different sources and combine database of them to improve the quality of data discover the knowledge inside the data pattern. The contribution of this study will help to break the barriers of privacy between manufacturers and improve the adoptability of soft sensor modelling to the changes of data sources.

Journal: International Journal of Simulation: Systems, Science and Technology IJSSST V19

Published: Jun 30, 2018

DOI: 10.5013/IJSSST.a.19.03.15