Information Technology Infrastructure for Augmented Problem Solving in Process Systems

点击翻译按钮获取中文摘要

Intelligent decision support systems (DSSs) and applications of artificial intelligence (AI) play a key role in improving productivity of Canadian industries. The Information Technology Association of Canada (ITAC) advocated that Canada expands her development and adoption of information and communications technology (ICT) so as to drive stronger productivity across all sectors of the economy. It is estimated that the Canadian market for the "internet of things" solutions would grow from $5 billion in 2013 to $21.1 billion in 2018 [ITAC report, Feb 16, 2015]. Speaking to CNN (F. Zakaria) at the World Economic Forum at Davos Switzerland in 2017, Ginni Rometty, current IBM CEO, described AI's role in the partnership between humans and machines as “augmented cognition”. In other words, AI not only supports but augments human cognition so that humans can be more efficient and “do a better job” (http://machinelearning.ai/artificial-intelligence/davos-2017-ibm-ceo-ginni-rometty-on-artificial-intelligence/, retrieved on July 23, 2017). In building intelligent DSSs that support human problem solving, a fundamental and most difficult step is knowledge acquisition, mainly due to the implicit nature of knowledge [Breuker, 2013]. The objective of the proposed research is to build an information technology infrastructure for knowledge acquisition (KA), which will support building intelligent DSSs that augment the problem solving abilities of human operators when they encounter operational problems in engineering process systems. The rule extraction approach will be adopted because it can generate interpretable rules from a well-trained artificial neural network (ANN) model. The explicit knowledge will be integrated with human expertise for building intelligent DSSs. This approach can effectively make use of the vast amount of data generated by the Canadian process industries while also overcoming the KA bottleneck. The KA infrastructure will be applied to the carbon dioxide capture process system and the cyclic solvent injection process for heavy oil recovery. The former is a key technology for cutting harmful carbon dioxide emissions from coal-fired power generation, and the latter system is a recovery method of heavy oil, a key resource important for sustaining the Canadian economy for many decades to come. The research results of the KA method, algorithms, and DSSs will benefit industrial engineers and operators. The insights and lessons learned in the research project would enable them to better understand the complex engineering process systems, and the DSSs would support them in solving operational problems in a timely manner and avoid downtime. The research results and HQP trained will contribute to information technology advancement of industrial operations related to energy production and greenhouse gas reduction in the Canadian economy.