中文摘要：

针对餐饮客流量自身特点,提出了以实验设计方法的样本分类为基础、综合利用BP（back propagation）神经网络与灰色-马尔科夫链组合的预测方法,并以某餐饮企业为例证明了方法的有效性,客流量预测误差从企业原来的士20％下降到±5％左右.

英文摘要：

According to characteristics of the catering passenger flow itself, the method by combining the back propagation （BP） neural network method and Grey-Markov chain was proposed, in which the design of experiment （DOE） was used for classifying the sample. At last, a case study proved that the method was effective, and the errors decreased from±20% to ±5%.