3:30 PM - 5:30 PM
[PE1-009] Design of Identification System Based on Gaussian Mixture Model for Black Plastic Wastes
Keywords:Gaussian Mixture Model (GMM), Preprocessing Technique, Laser Induced Breakdown Spectroscopy (LIBS), Black Plastics Identification System, Black Plastic Waste Recycling
Recycling of plastic wastes offers an efficient solution to reduce plastic pollutions in the environment as well as the consumption of the natural resources. Generally, Near Infrared Ray (NIR) sensor has been used to acquire the characteristic spectrum of plastics. However, when it comes to black plastics, NIR sensor cannot be used as a sensor for acquiring the spectrum of black plastics. NIR sensor cannot obtain the characteristic spectrum of a black material.
In this study, Laser-Induced Breakdown Spectroscopy (LIBS) has emerged as the substitution of NIR sensor to obtain and analyze the characteristic spectrum of black plastics. LIBS is used to obtain and analyze the spectrum of the black plastic wastes. In addition, the acquired spectrum by using LIBS is a very high dimensional data. When dealing with high dimensional data, data preprocessing technique is need to reduce the dimension and to extract the very important features from the original spectrum. Among various preprocessing algorithms, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Locally Linear Embedding (LLE) are applied to effectively extract the appropriate features from the original spectrum.
In order to analyze the features which are extracted by using the preprocessing technique (viz. PCA, LDA, and LLE) and to determine what the resins of black plastics are, a remarkable classification algorithm is also needed. Especially, Gaussian Mixture Model (GMM) is introduced to identify black plastics wastes according to their resins. GMM is one of the statistical methods which are based on the statistical inference. The outstanding identification system for black plastic wastes, which is introduced in this research, can improve the recycling rate of plastic wastes especially black plastic wastes.
In this study, Laser-Induced Breakdown Spectroscopy (LIBS) has emerged as the substitution of NIR sensor to obtain and analyze the characteristic spectrum of black plastics. LIBS is used to obtain and analyze the spectrum of the black plastic wastes. In addition, the acquired spectrum by using LIBS is a very high dimensional data. When dealing with high dimensional data, data preprocessing technique is need to reduce the dimension and to extract the very important features from the original spectrum. Among various preprocessing algorithms, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Locally Linear Embedding (LLE) are applied to effectively extract the appropriate features from the original spectrum.
In order to analyze the features which are extracted by using the preprocessing technique (viz. PCA, LDA, and LLE) and to determine what the resins of black plastics are, a remarkable classification algorithm is also needed. Especially, Gaussian Mixture Model (GMM) is introduced to identify black plastics wastes according to their resins. GMM is one of the statistical methods which are based on the statistical inference. The outstanding identification system for black plastic wastes, which is introduced in this research, can improve the recycling rate of plastic wastes especially black plastic wastes.
講演PDFファイルダウンロードパスワード認証
講演集に収録された講演PDFファイルのダウンロードにはパスワードが必要です。
現在有効なパスワードは、［資源・素材学会会員専用パスワード］です。
※［資源・素材学会会員専用パスワード］は【会員マイページ】にてご確認ください。（毎年１月に変更いたします。）
［資源・素材学会会員専用パスワード］を入力してください