Data Driven Application on Waste Stream Data: Industrial Symbiosis Knowledge System and Precautionary Model for Industrial Discharges and Emissions

The applications of A.I. has been booming in many disciplines because the increasing computing power and the enormous growth in data empower the advancement of A.I. applications. Also, the use of the internet of things and ICT technologies accelerates the accumulation of data. Environmental management relies on sufficient information to make sound and strategic decisions. The EPA of Taiwan has made many government-own datasets open to the public and brings more opportunities to discover new insight for environmental management through data mining on the released datasets and other related industrial data. The Waste Disposal Act has been forcing the majority of enterprises in Taiwan to report the material flows of waste streams for 18 years and keep a large amount of waste management data. This project proposes the development of two data-driven applications based on industrial waste data and various related data. The first application is a knowledge system with analytic tools to explore the opportunities of industrial symbiosis, by which firms may turn their wastes into byproducts and offer the byproducts to other firms of potential to consume them as raw materials. We will develop an optimization algorithm to examine the grouping of various industries that might generate most by-product synergies in an industrial area or region. The second application is to develop two smart predictors for abnormal wastewater discharging and flue gas emission. Based on a set of machine learning algorithms, this project will identify several factors that may induce the behaviors of illegal polluting, through the data mining on generation records of wastewater treatment sludge and ash from air pollution control devices in combination with other environmental, geological and industrial statistics. With the information of our predictors, the effort to inspect polluting behaviors can concentrate on the industries, the areas, and the periods of the high probability that occur.