Abstract: In order to forecast the demand of agricultural products cold chain logistics caused by the consumption，this paper constructs an impact index system from six aspects of agricultural products supply，social economy，cold chain development，humanities，logistics demand scale，and uses the advantages of BP neural network and RBF neural network in demand forecasting，establishes a demand forecasting model based on principal component analysis and neural network combination model. This paper takes Bei Jing as an example，and compares and analyzes the forecasting results of the two models. The results show that：through the grey relational analysis of various impact indicators，it is found that the number of urban population，the added value of the primary industry， the proportion of the tertiary industry to GDP and other factors have the greatest impact on the demand of agricultural products cold chain logistics；Forecast by 2020 Beijing urban residents agricultural products cold chain logistics demand will reach 6.642 million tons；The model established in this paper has high precision and application value in the non-linear relationship between cold chain logistics demand and its influencing factors，and can provide quantitative decision-making basis for agricultural cold chain logistics planners and governments.