基于BP神经网络的钢琴基波检测方法

Approach to Piano Fundamental Detection Based on BP Neural Network

钢琴基波的检测会影响钢琴质量和弹奏效果。介绍了一种基于BP神经网络实现基波检测的方法,充分利用了神经网络的逼近非线性映射的特性。根据推导得出的钢琴弦振动方程,分析基波与谐波波形的特点组合生成训练样本,采用合适参数进行网络训练。计算机仿真的实验结果良好,有效地检测出了钢琴的基波信号。并用未经训练的含有更高次谐波的样本试验,得到了令人满意的检测结果,表明该方法具有一定的泛化和外推能力,有助于今后对钢琴基波的进一步研究工作。

FundamentaI detection wiI be criticaI for its quaIity and pIaying effect.This paper describes a new fundamentaI detec-tion method based on BP neuraI network,using its abiIity of approximation to nonIinear mapping.According to the equation from deriving and considering the characteristics of signaIs in piano vibration system,the paper chooses training sampIes seriousIy to make sure they cover aImost aI the conditions without redundancy.And training the network with proper param-eters is another work to reaIize perfect resuIt.By means of computer simuIation,this method receives good resuIt.