摘要
引入偏最小二乘回归(PLSR)原理和方法应用于微波效应实验数据的预测,得到的预测精度与自适应神经模糊推理网络(ANFIS)结果基本一致,平均相对误差小于3%。实例分析了PLSR方法与ANFIS方法对建模数据样本量的需求,在建模样本数较少条件下,PLSR所建模型的预测精度均高于ANFIS模型。因此PLSR方法更适用于微波效应小样本数据的预测,更有利于实际应用。
The partial least-square regression（PLSR） method was introduced and implemented in the prediction of microwave effects.The results show that the PLSR method has an accuracy almost consistent with the adaptive neuro-fuzzy inference system（ANFIS） model＇s,and their average relative error is less than 3%.The requirement for sample size of these two methods was analyzed.On the condition of small sample size,the PLSR model is more precise than the ANFIS model.Thus the PLSR method is more effective for data processing and prediction with small sample size of microwave effects.
出处
《强激光与粒子束》
EI
CAS
CSCD
北大核心
2011年第5期1324-1328,共5页
High Power Laser and Particle Beams
基金
国家高技术发展计划项目
关键词
小样本
偏最小二乘回归
微波效应
自适应神经模糊推理网络
small sample size
partial least-square regression
microwave effects
adaptive neuro-fuzzy inference system
