一个基于灰色神经网络组合的交通量预测模型应用研究

GM模型在预测中对历史数据作不同取舍时,其预测值并不相同,即这种预测结果将是一个预测值的区间,这就给预测人员的取舍带来一定困难。利用GM模型少数据建模和人工神经网络非线性逼近的优点把两种模型结合起来,用对历史数据作不同取舍的GM模型的预测值和纯神经网络的预测值作为组合神经网络的输入,由人工神经网络确定这些不同GM模型和纯BP网络的组合,实例验证得出更为准确的预测值,从而证明这一模型的可行性和有效性。

The application study of traffic flow forecast model based on grey neural network

In the grey forecasting,after differently accepting or rejecting historical data and through accumulation and generation,different models are established,which forecast results are different that bring great difficulty to forecaster because of the uncertain interzone forecast results.Based on the combination of grey forecast and artificial neural network,a new model for traffic flow forecast is put forward in the paper.For this new combined model its inputs are the different forecast results of different grey models and pure artificial neural network,which ascertain the combination of different grey models and pure artificial neural network.The presented model synthesizes the advantages of GM forecasting method,which is simple and the needs less original data,and neural network which possesses the characteristic of nonlinear fitting,therefore the more accurate certain results we can get.Calculation examples show that the presented method is feasible and effective.