Research On Manufacturing Monitoring System Of Additional Materials Based On LIBS

Additive manufacturing(AM)technology,characterized by precise reduction prototyping and rapid prototyping etc.,which can be used to manufacture various complex devices,has good application prospects in aviation,machinery and military fields.However,in the process of the additive manufacturing,there exists the problem of material damage caused by equipment precision control,which leads to defects of formed parts.Since the chromium element can enhance the briquetting ratio of the additive manufacturing materials,the forming quality of the parts can be predicted and judged by real-time monitoring of the chromium element.Therefore,it is important to control the forming quality which can realize real-time monitoring of the additive manufacturing process and accurate detection of the constituent elements.Laser induced breakdown spectroscopy(LIBS)has extensive practical prospect in the field of metal composition monitoring due to its characteristics of no pretreatment,rapid multi-element analysis and long-distance real-time online detection,etc.In this paper,LIBS technology is used to monitor the chromium(Cr)elements of additive manufacturing materials,and solves some key factors in the process of monitoring,such as improving control precision of 3D mobile platform,and considering the influence of laser energy,pulse interval,spectrometer acquisition delay,laser focus distance and laser wavelength on monitoring.At the same time,we aim at improving the stability of monitoring Cr element and the precision of identifying the defects in additive manufacturing parts.The specific research contents of this paper are as follows:1.Aiming at the processing process of metal laser additive manufacturing,laser induced plasma spectroscopy technology was introduced to monitor and the real-time spectral acquisition system was built.At the same time,the defect identification method is proposed to improve the monitoring accuracy of metal additive manufacturing and the identification rate of defect recognition.2.In view of the poor control precision of the 3D mobile platform,the hardware circuit and software interface design of the 3D mobile platform is re-designed,which not only effectively solved the problem of low control precision of the 3D mobile platform,but also improved the convenience of software operation;3.To counter the problem of measurement result affected by monitoring parameters,the influences of parameters of spectrometer acquisition delay,pulse interval time,laser focus position,laser energy and laser wavelength are compared and analyzed in terms of spectral intensity,signal-to-back ratio,relative standard deviation and plasma morphology.And selected parameters suitable for monitoring additive manufacturing to avoid data errors caused by monitoring parameters,and to improving monitoring accuracy;4.Aiming at the problem of poor accuracy and stability of the monitoring component data analysis method,the results of component analysis and prediction are improved by optimization algorithm.The absolute strength method,partial least squares(PLS)and the least squares support vector machine(LSSVM)are used to quantitatively predict and analyze the nonlinear relationship between the line intensity ratio of the monitoring data and the element concentration.The results show that the accuracy and stability of component monitoring using the least squares support vector machine is better than the univariate quantitative analysis and partial least squares quantitative analysis,and the correlation coefficient is increased to 0.9993.The predicted root mean square error is 0.0221%,the test root mean square error is reduced to 0.0327%,and the average relative error is also decreased to 1.3112%.Therefore,it is more reasonable to apply the least squares support vector machine to the analysis of additive manufacturing monitoring data;5.For the problem of defect identification of metal additive manufacturing samples,the defects are compared and identified by using PLS and LSSVM.The experimental results show that the identification rate and correct rate of defect recognition by LSSVM are better than by PLS.The recognition probability of identifying an overall defect using LSSVM increased from 91.6% to 98.3%.Based on two kinds of defect recognition models,the results were verified.The results show that the correct rate of spheroidization defect is improved from 73.3% to 100% with LSSVM.The recognition accuracy of defects increased from 76.7% to 90.0%;the recognition accuracy of crack defects also increased from 80.0% to 93.3%.This verifies the effectiveness of the LSSVM classification model for defect classification.