Neural network-based source tracking of chemical leaks with obstacles

作     者：Qiaoyi Xu Wenli Du Jinjin Xu Jikai Dong Qiaoyi Xu;Wenli Du;Jinjin Xu;Jikai Dong

作者机构：Key Laboratory of Advanced Control and Optimization for Chemical ProcessesMinistry of EducationEast China University of Science and TechnologyShanghai 200237China Shanghai Institute of Intelligent Science and TechnologyTongji UniversityShanghai 200092China 

出 版 物：《Chinese Journal of Chemical Engineering》 (中国化学工程学报（英文版）)

年 卷 期：2021年第34卷第5期

页      面：211-220页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 0817[工学-化学工程与技术] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：The work was supported by the National Natural Science Foundation of China(Basic Science Center Program:61988101 21706069),Natural Science Foundation of Shanghai(17ZR1406800) National Science Fund for Distinguished Young Scholars(61725301) 

主　　题：Obstacle Optimization Neural networks Feature extraction Source term estimation Computational fluid dynamics (CFD) 

摘      要：The leakage of hazardous gases poses a significant threat to public security and causes environmental *** effective and accurate source term estimation(STE)is necessary when a leakage accident ***,most research generally assumes that no obstacles exist near the leak source,which is inappropriate in practical *** solve this problem,we propose two different frameworks to emphasize STE with obstacles based on artificial neural network(ANN)and convolutional neural network(CNN).Firstly,we build a CFD model to simulate the gas diffusion in obstacle scenarios and construct a benchmark ***,we define the structure of ANN by searching,then predict the concentration distribution of gas using the searched model,and optimize source term parameters by particle swarm optimization(PSO)with well-performed cost ***,we propose a one-step STE method based on CNN,which establishes a link between the concentration distribution and the location of ***,we propose a novel data processing method to process sensor data,which maps the concentration information into feature *** comprehensive experiments illustrate the performance and efficiency of the proposed methods.