基于BP神经网络的货币识别算法设计与实现

神经网络模式识别方法是近几年兴起的模式识别领域的一个新的研究方向,而BP算法是神经网络众多算法中应用最为广泛的一种.BP神经网络已应用于非线性建摸、函数逼近、模式分类等方面,但对解决具体问题而言,其模型结构的确定、每层神经元个数的选择、训练参数的设定等都无现成的规律可供遵循,必需由实验确定,由此,在分析BP算法和收敛性的基础上,设计了基于BP神经网络的货币识别算法,测试证明,其具有良好的识别率,具有一定的理论和实用意义.

Design and Implementation of Algorithm of Currency Recognition Based on BP Nerve Network

Pattern--recognition methods of nerve network are a new research direction which emerges in recent years in pattern--recog- nition domain and BP algorithm is the most widespread one in the nerve network algorithms. The BP nerve network has widely applied on the non--linearity constructing--model, the function approaches, pattern classification, and so on.. But, speaking of solving the concrete problem, determination of its model structure, choice of neuron number of each layer, enactment of the training parameter, which do not have the ready--made rule to supply to follow, must be determined by the experiment. So, by analyzing BP algorithm and astringency, Algorithm of Currency Recognition Based on BP Nerve Network is designed. It is good for recognition-- efficiency by proof on test and has the theory significance and the practical significance.