Review and big data perspectives on robust data mining approaches for industrial process modeling with outliers and missing data

Industrial process data are usually mixed with missing data and outliers which can greatly affect the statistical explanation abilities for traditional data-driven modeling methods. In this sense, more attention should be paid on robust data mining methods so as to investigate those stable and reliable modeling prototypes for decision-making. This paper gives a systematic review of various state-of-the-art data preprocessing tricks as well as robust principal component analysis methods for process understanding and monitoring applications. Afterwards, comprehensive robust techniques have been discussed for various circumstances with diverse process characteristics. Finally, big data perspectives on potential challenges and opportunities have been highlighted for future explorations in the community.