Design And Implementation Of Real-time Data Analysis Platform For Powder Diffractometer

Powder diffractometer is a kind of scientific instrument which is widely used in the field of material science.The analysis of powder diffractometer data is generally based on the analysis system and software carried by the manufacturer.Its analysis equipment and analysis software are sold with powder diffractometer equipment and have one to one characteristics with powder diffractometer.After the diffraction data is generated,it directly enters the analysis equipment corresponding to the powder diffractometer,the analysis equipment can analyze the data according to the needs of the experimenters.After a long period of development,the analysis software has the characteristics of high speed and high accuracy,and the function is complete.All kinds of data preprocessing function and the qualitative and quantitative analysis function of the sample are mature.But at present,the data analysis mode is based on the single computer system,and there are some disadvantages that can not be eliminated.With the progress of bigdata technology,the data analysis platform of distributed architecture has been used in business,industry and other fields,and tends to be mature.However,it has not been applied to the analysis of the diffractometer for the time being.The research on this application field will be beneficial to the supplement of the existing analysis system as well as the exploration of the improvement direction of the existing system.Based on Hadoop,a real-time data analysis platform for powder diffractometer is designed and implemented.It changes the one to one service mode of the existing powder diffractometer and the computer,and forms a new working mode of a new multi to one(multi powder diffractometer corresponding to an analysis cluster).The platform includes the functions of pretreatment,storage and analysis of powder diffractometer data.In the data analysis module,several important aspects,such as single matter distinction,mixture composition and proportion analysis,equipment error diagnosis and equipment state prediction,are emphatically designed and implemented.With the use of mature Hadoop platform and the application of mature machine learning algorithms in the fields of classification and time series prediction,it makes the system maintenance simple and greatly reduces the use cost.The application of mature open source technology also allows many required functions to be added to the platform through similar methods and processes.