为准确预测蔬菜市场价格走势,现选取海南省儋州市2012—2015年117组青椒旬零售价格及相关因素的旬价格为样本数据,其中100组作为训练数据,17组数据作为测试数据,分别建立基于粒子群算法优化BP神经网络的蔬菜价格预测模型和基于RBF神经网络的蔬菜价格预测模型,并在这2种模型的基础上建立蔬菜价格的线性组合预测模型。结果表明:2种单项预测模型在蔬菜价格预测上的应用效果都较好,且在不同评价指标上各有优势。将这2种模型的预测结果进行线性组合,可以使各单项模型优势互补,拟合效果明显优于各单项预测模型。 In order to accurately predict the vegetable market price trend, the ten-day price of 117 sets of green pepper in 2012-2015 in Danzhou City of Hainan Province is selected as the sample data, of which 100 are used as training data and 17 sets of data are used as the test data respectively The vegetable price forecasting model based on Particle Swarm Optimization (BPO) neural network and vegetable price forecasting model based on RBF neural network are established. Based on these two models, a linear combination forecasting model of vegetable prices is established. The results show that the two single-item forecasting models have good effect on vegetable price forecasting, and each has its own advantages in different evaluation indexes. By linear combination of the prediction results of these two models, the advantages of each individual model can be complemented, and the fitting effect is obviously better than the single prediction models.