随机森林算法在电喷印液滴直径预测中的应用

Application of Random Forest Algorithm in the Prediction ofElectrohydrodynamic Inkjet Printing Droplet Diameter

电流体喷射打印技术是一种新型微纳尺度增材制造工艺,由于电喷印成本低、工艺简单、效率高、精度高,在制备微流控芯片、印刷柔性电子器件和各种光学器件等领域具有广阔的应用前景。精确预测并控制电喷印液滴直径是保证加工质量的关键。因此,以电喷印液滴直径预测为研究对象,运用随机森林回归算法进行建模,来预测在几种关键参数(基电压、高电压、脉冲频率、喷头与基板的距离、油墨电导率及粘度)的条件下打印出的液滴直径。预测结果显示,该模型的预测精度较传统的理论建模及回归分析的结果有明显提高。在几种常见的算法预测模型中,随机森林回归模型在训练样本受限的情况下,预测精度最高。基于随机森林回归算法的液滴直径预测模型为电喷印液滴直径预测及控制提供了一种经济有效的方法。

Electrohydrodynamic inkjet printing technology is a new type of micro-nano-scale additive manufacturing method.Due to the low cost,simple process,high efficiency and high precision of Electrohydrodynamic inkjet printing,it has great poten-tial application in fabrication of microfluidic chips,printed-flexible electronic devices,and various optical devices,etc.Therefore,taking the droplet diameter prediction of electrohydrodynamic inkjet printing as the research object,a model based on the random forest regression algorithm is developed,in order to predict the printed droplet diameter under the condition of several key parame-ters(base voltage,high voltage,pulse frequency,distance between the print head and the substrate,material conductivity and vis-cosity).The prediction result shows that the prediction accuracy of the model is significantly higher than that of the theoretical modeling and regression analysis.Among several common algorithm prediction models,random forest regression model has the highest prediction accuracy when the training samples are limited.The droplet diameter prediction model based on random forest regression algorithm provides an economic and effective method for droplet diameter prediction and control of electrohydrodynam-ic inkjet printing.