基于BP神经网络的天平校准数据处理方法研究

针对天平校准公式拟合法方,提出了基于BP神经网络的公式拟合方法。分析了典型的3层BP神经网络的结构,论述了其应用于天平公式拟合的可行性。讨论了常规多项式公式拟合方法以及BP神经网络的公式拟合方法,并给出了BP神经网络的训练过程和使用方法。最后通过一组天平校准数据对BP神经网络进行训练,并给出拟合结果。

Research on data processing method of balance calibration based on BP neural network

This paper aims at fitting the calibration formula of the balance, a formula fitting method based on BP neural network is presented. Analyses the structure of typical 3-layer BP neural network, discusses the feasibility of the application of balance formula fitting. Discusses the fitting methods for general polynomial and BP neural network, and the training process and application method of BP neural network are given. Finally, the BP neural network is trained by a set of balance calibration data, and the fitting results are given.