Font Size: a A A

Research On Online Detection Of Atmospheric Smoke In Industrial Scenario

Posted on:2024-02-08Degree:MasterType:Thesis
Country:ChinaCandidate:E L WanFull Text:PDF
GTID:2531307106478684Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
Atmospheric pollution is one of the most important environmental issues in the world,and it is also one of the biggest environmental issues facing China.The harm of air pollution is multifaceted,not only damaging human health,but also affecting the environment and climate,directly or indirectly causing a series of ecological and environmental problems.An important cause of air pollution is industry.The properties of pollutants discharged into the atmosphere by industrial production are extremely complex and there are various types of these pollutants.At present,conventional methods for detecting atmospheric smoke in industrial environments,such as inductively coupled plasma mass spectrometry,differential optical absorption spectroscopy,ion chromatography and other technologies,often require complex sample pretreatment,and cannot achieve on-line in situ detection.In this paper,based on LIBS and SPAMS technology,combined with PCA and BP-ANN,on-line detection of atmospheric smoke in industrial environment is carried out.The research contents are as follows:(1)Online laser detection of smoke from burning coal.Under the optimized experimental conditions,online detection with LIBS technology was conducted for the combustion smoke of lignite,anthracite and bituminous coal respectively.According to the spectrum obtained,the main components in smoke were Al,Fe,Ca,Mg,Mn,Si,Sr and Na.In addition,the CN molecular band signal can also be observed from the spectrogram.By comparing the spectra of different samples,PCA was used to reduce the dimension of the raw data.Then the data was distributed in the three-dimensional space,and different types of soot were successfully classified.Combined with BP-ANN,different kinds of soot were accurately identified,and results are very good,with recognition accuracy of more than 92%.(2)Online laser detection of smoke from electric welding operation.Online detection of smoke from leaded tin wire,lead-free tin wire and rosin was carried out by LIBS technology.According to the spectrum obtained,Pb,Sn,Cu,Ca,K and other metal elements were detected in the leaded tin wire.No Pb characteristic peak was observed in the lead-free tin wire spectrum,and strong C signal and CN molecular band signal were observed in the rosin spectrum.With the help of SPAMS technology,the isotope and abundance information of Pb and Sn in smoke were obtained.Then,two scenarios of welding aluminum products and welding circuit boards were simulated.In addition,the C concentration in the continuous operation process was monitored online,and quantitative analysis of Pb in smoke was performed.The Pb concentration can be calculated by using the internal standard method according to the spectral line strength in the spectrum.Finally,combined with PCA and optimized BP-ANN model,different types of smoke and dust were simply classified.(3)Online laser detection of plastic and its combustion smoke.According to the LIBS spectrum obtained,the main components of plastics were Ca,Mg,Na,K,C,H and O.The characteristic peaks of C,H and O were observed in the spectrum of smoke from burning plastic.Then,the spectral differences between different samples were compared.In addition,the heavy metal pollution in the plastic treatment process was detected online.Finally,PCA and further improved BP-ANN were applied to the accurate identification of different plastics and their combustion smoke.The above work is important application in the photoelectric detection field,which not only provides a solid theoretical support and experimental basis for the real time monitoring of atmospheric smoke pollution,but also gives a new idea for the treatment of current atmospheric environmental pollution issues.
Keywords/Search Tags:Laser-induced breakdown spectroscopy, atmospheric pollution, industrial smoke, error back propagation artificial neural network, principal component analysis
PDF Full Text Request
Related items