激光质谱中基于数据挖掘的激光输出功率预测技术研究  

Research on Laser Output Power Prediction Technology Based on Data Mining in Laser Mass Spectrometry

作　　者：刘莲花 杨文喜[2] 张晓卫 但勇军 刘彬[2] Liu Lianhua;Yang Wenxi;Zhang Xiaowei;Dan Yongjun;Liu Bin(Key Laboratory of Science and Technology on Particle Transport and Separation,Tianjin 300180,China;Research Institure of Physical and Chemical Engineering of Nuclear Industry,Tianjin 300180,China)

机构地区：[1]粒子输运与富集技术国防重点实验室,天津300180 [2]核工业理化工程研究院,天津300180

出　　处：《计算机测量与控制》2019年第9期137-140,共4页Computer Measurement &Control

摘　　要：由于激光质谱系统逻辑结构复杂多样,激光输出功率是激光质谱系统进行的关键条件之一,提前掌握激光输出功率未来状态的发展趋势可为激光质谱系统运行决策提供重要依据,因此进行激光质谱系统激光输出功率的预测技术研究非常必要;采用M5预测模型、线性回归模型、向量机模型对质谱系统的激光输出功率历史数据进行了建模及预测分析,通过比较几个预测模型的预测误差及平均误差,结果表明M5预测模型的预测结果相对最优;通过对激光输出功率历史数据的分析及预测,确定了激光质谱系统激光输出功率的研究预测模型。Because the logic structure of laser mass spectrometry system is complex and diverse,laser output power is one of the key conditions for laser mass spectrometry system to carry out.To grasp the development trend of laser output power in advance can provide an important basis for the operation decision of laser mass spectromery system.Therefore, it is necessary to study the prediction technology of laser output power.The system adopted the M5 prediction model,the linear regression model and the vector machine model to modele and predicte by history data.By comparing the prediction errors and the average errors of several prediction models, the prediction results of the M5 prediction model are relatively optimal.Through the analysis and prediction of the output power historical data,the research and prediction model of the output power of laser mass spectrometry system is determined.

关 键 词：数值预测 预测模型 预测算法 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]