Iranian Institute of Research and Development in Chemical Industries (IRDCI)-ACECR Nashrieh Shimi va Mohandesi Shimi Iran 1022-7768 38 3 2019 10 23 Evaluation of Palm Biodiesel Fraud Detection System by Dielectric Spectroscopy and Machine Vision Techniques Evaluation of Palm Biodiesel Fraud Detection System by Dielectric Spectroscopy and Machine Vision Techniques 281 291 31864 FA Ali Reza Zahedi Faculty of New Technologies, Iran University of Science and Technology, Tehran, I.R. IRAN Mahdi Rashvand Colleges of Engineering, University of Tehran, Tehran, I.R. IRAN Ardeshir ShayanNejad School of New Technologies, Iran University of Science and Technology, Tehran, I.R. IRAN Journal Article 2018 02 06 <em>Biodiesel is a clean and alternative fuel for diesel. It produces from renewable sources such as vegetable and animal oils. However, due to the high price of this clean fuel, it is mixed with other inexpensive fuels, such as diesel, and a fraud fuel was sold as a pure one. The purpose of this study is to evaluate an intelligent system for detecting the level of fraud in biodiesel fuels by integrating dielectric spectroscopy and image processing techniques. In order to classify fake biodiesel specimens using frequency and color properties, the principal component analysis, linear discriminant analysis, decision tree, Support vector machine, and Artificial neural network were used. The best network with structure 1-6-36 to predict blended biodiesel palm and diesel samples provided correlation coefficient and mean square error values of 0.944 and 0.006, respectively. In the final step, the combination of dielectric and color properties was used to model the problem. Correlation coefficient and mean squared error values for biodiesel and diesel mixed samples were 0.962 and 0.008 with a structure of 1-2-38. The results of the evaluations show that the designed device has the ability to detect biodiesel fuel fraud with high precision.</em> <em>Biodiesel is a clean and alternative fuel for diesel. It produces from renewable sources such as vegetable and animal oils. However, due to the high price of this clean fuel, it is mixed with other inexpensive fuels, such as diesel, and a fraud fuel was sold as a pure one. The purpose of this study is to evaluate an intelligent system for detecting the level of fraud in biodiesel fuels by integrating dielectric spectroscopy and image processing techniques. In order to classify fake biodiesel specimens using frequency and color properties, the principal component analysis, linear discriminant analysis, decision tree, Support vector machine, and Artificial neural network were used. The best network with structure 1-6-36 to predict blended biodiesel palm and diesel samples provided correlation coefficient and mean square error values of 0.944 and 0.006, respectively. In the final step, the combination of dielectric and color properties was used to model the problem. Correlation coefficient and mean squared error values for biodiesel and diesel mixed samples were 0.962 and 0.008 with a structure of 1-2-38. The results of the evaluations show that the designed device has the ability to detect biodiesel fuel fraud with high precision.</em> https://www.nsmsi.ir/article_31864_aa6d4debefaf1d03a69265997d8a8a6d.pdf