Abstract: Affected by many complications, the wave is too complex to be described by only one precise model. Firstly, the exchange rate fluctuating trends are expressed by fuzzy orientation codes; Secondly, artificial neural network is used in forecasting. In the end, experimental results are presented, which prove the higher validity and the feasibility of the method and can provide higher paper profits

Key words: Neural network; Exchange rate; Fuzzy-orientation-codes

摘要： 汇率的波动受到多种不确定因素的影响，很难找到一种模型精确有效地刻画汇率波动与其影响因素之间存在的联系。该文提出了采用改进的模糊方位编码方法表示汇率走势，然后采用神经网络预测汇率走势的方法。通过实际系统运行，结果表明，该方法具有较高的预测准确度，并且能够得到客户更关心的较高的理论获益度。

关键词: 神经网络；外汇汇率；模糊方位编码