基于有机灰色神经网络模型的空气污染指数预测

针对灰色预测对波动较强的序列只能预测大致变化趋势的缺陷，结合灰理论中的GM(1，1)、无偏GM(1，1)、非等时距GM(1，1)、pGM(1，1)和BP神经网络的特点，提出有机灰色神经网络预测模型，将一维序列通过三个灰色模型得到三组值作为神经网络的输入，原始序列作为神经网络的输出，训练得到最佳神经网络结构．以哈尔滨市近三年内空气污染指数为例，结合其变化规律，建立哈尔滨市月平均空气污染指数的有机灰色神经网络预测模型，结果表明，该模型拟合误差小，预测精度高．

Prediction for air pollution index based on organic grey neural network model

For the violent fluctuating sequence, grey prediction method can only forecast the tendency approximately. By combining the advantages of GM(1,1), unbiased GM(1,1), unequal time difference GM(1,1), pGM(1,1) and BP neural network, a new organic grey neural network model was proposed, in which the predicting results of three grey prediction models were used as the neural network's inputs, and the original sequence was used as the output of the neural network. The neural network was trained to get the optimal structure, weights and thresholds. Dynamic laws of the recent three years' average month air pollution index of Harbin for example, the average month air pollution index's organic grey neural network model of Harbin was built. Experimental results showed that the model had highly fitting and predicting precision advantages over the other models.