Title: Soft measurement of dioxin emission concentration based on deep forest regression algorithm

Authors: Tang Jian; Xia Heng; Qiao Junfei; Guo Zihao

Addresses: Faculty of Information Technology, Beijing University of Technology, Beijing, 100024, China; Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing, 100124, China; Beijing Laboratory of Smart Environmental Protection, Beijing, 100124, China ' Faculty of Information Technology, Beijing University of Technology, Beijing, 100024, China; Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing, 100124, China; Beijing Laboratory of Smart Environmental Protection, Beijing, 100124, China ' Faculty of Information Technology, Beijing University of Technology, Beijing, 100024, China; Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing, 100124, China; Beijing Laboratory of Smart Environmental Protection, Beijing, 100124, China ' Faculty of Information Technology, Beijing University of Technology, Beijing, 100024, China; Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing, 100124, China; Beijing Laboratory of Smart Environmental Protection, Beijing, 100124, China

Abstract: Dioxin (DXN) is an organic pollutant emitted by the municipal solid waste incineration (MSWI) process. In industrial process, the DXN emission concentration is detected by using offline laboratory analysis method with a monthly/seasonal or un-determined period. In this paper, a soft measurement method of DXN based on deep forest regression (DFR) algorithm is proposed. First, the input layer forest model consists of multiple sub-forest models is trained and the layer regression vector is obtained. Then, the augmented layer regression vector that serial combine the layer regression vector with the raw features is used to train the middle layer forest model. Finally, the augmented layer regression vector of the middle layer forest is fed into the output layer forest model to produce the final DXN prediction. The effectiveness of the proposed method was verified by the benchmark data and the DXN emission concentration data of the actual MSWI process.

Keywords: dioxin; DFR; deep forest regression; layer regression vector; augmented layer regression vector; MSWI; municipal solid waste incineration.

DOI: 10.1504/IJSCIP.2021.117695

International Journal of System Control and Information Processing, 2021 Vol.3 No.3, pp.208 - 228

Received: 23 Sep 2020
Accepted: 18 Feb 2021
Published online: 21 Sep 2021 *