一种使用前向神经网模型的A／D转换器

An A / D Converter Using Feedforward Neural Network Model

首先指出了Ｔａｎｋ－ＨｏｐｆｉｅｌｄＡＤＣ由于“磁滞”现象的存在，引起模拟量的不正确二进制表示；其次给出一种使用前向神经网模型的ＡＤＣ，避免了上述问题，最后对转换时间作了讨论。

The contribution of the paper is mainly two-fold. Firstly, it suggests Tank-Hopfield ADC exhibits 'hysteresis' in the sense that the state of the network sometimes converges towards a binary vector which is not the correct representation of the input current. Secondly, an ADC using feedforward neural network model is proposed. Finally the time of conversion is discussed.