【摘要】 介绍了模拟神经网络 VLSI脉冲流技术实现神经网络模式识别硬件电路的方法 ,并且直接将故障分类。提出利用包含有故障信息的原始模拟噪声信号 ,经过前置信号处理和神经网络运算 ,得出 VLSI电路输出端电容的电压值——代表待识别信号与模板故障信号的“欧氏距离”,以实现噪声故障信号的实时硬件在线识别更多还原

【Abstract】 A analog neural network VLSI & pulse stream technique is presented. The technology can realize the hardware circuit of neural network model recognition and classify the fault samples directly. It makes original analog noise signal including fault information go through the signal processing and neural network computation. Then the voltage value of output capacitor in VLSI circuit is obtained, which represents Euclid distance between template fault signal and the signal needed to be recognized. By above methods, the real time & online hardware recognition of noise fault signals is realized.更多还原