Multimodal Data Based Regression to Monitor Air Pollutant Emission in Factories

Air pollution originating from anthropogenic emission, which is an important factor for environmental policy to regulate the sustainable development of enterprises and environment. However, missing or mislabeled discharge data make it impossible apply this strategy in practice. In order solve challenge, we firstly discover that energy consumption a factory air pollutants are linearly related. Given observation, propose support vector regression based Single-location recovery model recover pollutant emission by using factory. To further improve precision estimation, proposed Gaussian process multiple-location estimate surrounding available quality readings, collected government’s monitoring station. Moreover, optimally combine two approaches achieve accurate estimation. our best knowledge, first paper taking both factory’s readings into account. The research article uses actual data(10,406,880 entries including weather, PM 2.5, date, etc.) parts Shandong Province, China. dataset contains 33 factories (5 types) use co-located station as ground truth. results show that, single-location recovery, multi-location combined method could acquire mean absolute error 8.45, 9.69, 7.25, respectively. has consistent prediction behavior among 5 different types, shows promising potential be applied broader locations application areas, outperforms existing spatial interpolation methods 43.8%.