通过Matlab基于BP神经网络实现数字0~9的识别。首先创建50个训练样本供网络学习，根据训练样本的特点确定输入层、输出层神经元个数，并确定隐含层神经元个数，完成对BP神经网络的设计；然后将训练样本输入BP网络中，完成对网络的训练；最后通过20个测试样本测试训练完成的网络性能，并显示识别结果。-Digital  recognition based on BP neural network.First according to the  characteristics of the training samples create the  50 training samples  to determine the input  layer and output layer  neuron number,and ,  and determine  the number of hidden layer neurons:complete the design  of the BP neural network   input the training sample  in the BP network,  train the network 　  finally, 20  test  samples  training  test the performance of BP neural network ,  and display identification results .