Design And Recognition Algorithm Research Of Handheld Raman System Based On ARM

Raman spectrum is a kind of inelastic scattering spectrum.The scattered light carries the "fingerprint" information of the substance and can characterize the molecular structure of the substance.Raman spectroscopy is a non-invasive optical detection technology,which has been widely used in various research fields due to its advantages of fast,nondestructive,non-contact,small sample consumption,simple preparation and so on.In recent years,as many research fields begin to cross,merge and promote each other,a series of new requirements for Raman detection technology are put forward,such as fast,convenient,non-destructive,suitable for field operation,and so on.Therefore,hand-held Raman spectrometer has attracted extensive attention.However,at present,the market of hand-held Raman spectrometers is mainly monopolized by foreign major optical companies,and few domestic manufacturers have launched handheld Raman spectrometers.Therefore,this thesis designed a handheld Raman spectrometer.It has low cost,small volume,easy to operate,easy to carry,intelligent acquisition system and Raman spectral database,which is of great significance to promote the development of handheld Raman system in China.Firstly,this thesis studies the hardware composition principle of the handheld Raman system,and analyzes the working principle of each major module.Finally,according to the design purpose and application scenarios,the suitable laser,Raman probe,micro spectrometer and embedded hardware development platform are selected to complete the hardware design of the handheld Raman system.Then,the software function of the handheld Raman system is developed.Firstly,the embedded Linux system is transplanted as the operating system of the handheld Raman system.Then,based on the design principle of the character device driver,the USB program of the micro spectrometer is designed.Finally,the overall humancomputer interaction framework of the system is analyzed,the main interface is designed by Qt/Embedded development framework,and the overall function of the system is adjusted and tested by low pressure mercury and argon lamps and plastic samples.In addition,this thesis also studied the recognition algorithm of hand-held Raman system with plastic samples as the research object.In this study,we combined Raman spectroscopy with machine learning to identify 11 kinds of plastic samples.Firstly,a785 nm handheld Raman system designed in this this thesis is used to collect the Raman spectral data of plastic samples,and principal component analysis is used to extract the features.The plastic recognition models based on three machine learning algorithms are compared.The results show that the three machine learning algorithms can identify11 kinds of plastics well,with an average accuracy of 96.59%.It is proved that machine learning algorithm combined with Raman spectroscopy has great potential for fast and nondestructive identification of plastics This study not only provides a unique method for the recycling of waste plastics,but also provides a new idea for the establishment of spectral database and the development of identification algorithm of hand-held Raman system.