视觉导航自动车辆用BP神经网络数字识别方法的研究

为了满足视觉导航自动车辆AGV对阿拉伯数字准确、快速识别的要求,该文提出了基于BP神经网络的数字识别方法。为满足对数字识别实时性、准确性的要求,研究改进了数字识别的算法,考察了隐层节点数对收敛速度、数字识别效果的影响以及学习率η对收敛速度的影响,并选取了较好的学习率对阿拉伯数字进行了网络训练,得到了网络收敛后的权值和阈值矩阵。在有噪声的条件下,进行了实验检验,实验结果表明该方法能够实现对阿拉伯数字的准确、快速识别。

Study on Numerical Recognition Method of Vision Navigate AGV Based on BP Neural Network

In order to meet the exact and fast recognition of Arab numbers of AGV(Automatic Guided Vehicle)on vi-sion navigation,a new control method—— — BP neural network method has been successfully developed.For fulfilling the real time and veracity demands of numerical recognition,the improved arithmetic of numerical recognition is researched and the influence to convergent speed and numerical recognition effect of both concealed layer node count and the learning rate is also considered.The network weight and the matrix threshold are obtained by training the neural net-work of Arab numbers through a better learning rate.Practical examples indicate that the method works well in recogniz-ing the Arab numbers veracious and quickly.