用改进的小脑模型神经网络实现光栅信号连续细分

光栅测量系统常用硬件进行信号细分，存在细分数不高、硬件复杂等问题。为此，研究了一种神经网络细分方法，即利用一种改进的小脑模型神经网络的泛化能力实现光栅信号的连续细分。该小脑模型神经网络采用直接权地址映射技术，将光栅样本信号直接映射到联想存储器，对非样本信号经联想泛化即可实现连续细分。仿真实验结果表明，仅用少量样本即可达到很高的细分精度。

Research on the Subdividing of Grating Signals Using an Improved CMAC Neural Network

It is complex and lower accuracy that grating signal is subdivided by the hardware. Therefore, a method for subdividing the grating signals is proposed using an improved Cerebellar Model Articulation Controller (CMAC) neural networks, which direct weight address mapping techniques are employed in the algorithm of the model, and the sample of grating signals are direct mapped to the associative memory as the head address of C weights. After training with the samples, any input is taken as the head address of C weights between the two similar samples. The subsequent accurate outputs are obtained by associative algorithm. Simulation experiments show that the accuracy of subdividing the grating signals is very high so long as using a few training samples.