人工神经网络是对人脑功能的某些程度的反映,具有自适应和自学习的能力,可通过对模式样本的自学习,从中获取特征,并能将学习获得的知识应用到图像、文字等识别中。本文采用Hopfield神经网络进行英文字母识别,仿真结果表明,该方法能有效地对含噪声的英文字母进行识别。在相同的白噪声模型下,该方法在噪声均方差稍小时,其容错能力比起 Back Propagation网络方法有一定的增强。 Artificial neural network is to some extent reflect the human brain function, with the ability to adapt and self-learning, self-learning through the pattern of samples to obtain features, and can apply the learning of knowledge acquired to images, text, etc. Recognized. In this paper, Hopfield neural network is used to identify English letters. The simulation results show that this method can effectively identify the noisy English letters. Under the same white noise model, the proposed method has a better performance in fault tolerance than the Back Propagation network method when the mean square error of noise is small.