【作者】 薛荆岩； 沈永良； 孙来军； 涂磊；

【Author】 Xue Jingyan Shen Yongliang Sun Laijun Tu Lei (Key Laboratory of Electronics Engineering, Collage of Heilongjiang Province (Heilongjiang University) Harbin 150080, China)

【机构】 黑龙江省电子工程高校重点实验室(黑龙江大学)；

【摘要】 微波谐振腔含水率传感器测量过程中输入谐振频率、品质因数和环境温度与含水率之间为多元非线性函数关系,而多元非线性回归算法的优劣是决定测量精度的主要因素。针对这一问题,提出使用径向基函数神经网络对测量过程的输入输出进行多元非线性回归。该方法避免了经典BP算法容易陷入局部极小,训练效率低的缺点,保持了较高的泛化精度,且收敛速度快,进而提高了测量精度。实验表明,应用本文径向基神经网络多元非线性回归方法的含水率测量值与真实值间相对误差均方差为0.076,相对误差均值为0.103,相对误差最大值为0.154。更多还原

【Abstract】 The relationship between inputs of resonant frequency,quality factor and environment temperature and output of moisture content is multi-nonlinear.Hence the performance of multi-nonlinear regression algorithm can influence remarkably the measurement accuracy.A regression is put forward based on a RBF network algorithm to modify the measurement result.The RBF neural network regression algorithm avoids getting into infinitesimal locally effectively and keeps the merits of high prediction precision and rapid convergence so as to enhance the measurement accuracy.The results show that the root mean-square relatively error is 0.076,the mean absolute relatively error is 0.103,and the maximize absolute relatively error is 0.154 between the measurement value and the real one.更多还原