灰色神经网络交通事故预测比较

针对BP神经网络多变量输入难以确定的缺点,提出了采用灰色关联分析法确定主要影响因子输入的多因子灰色关联分析神经网络预测模型,实例证明,该方法预测精度优于全输入BP神经网络预测。进一步提出了应用选优BP神经网络输入预测和GM(1,N)组合预测的模型,它结合了灰预测利用少数据累加生成建模,容易找出数据变换规律的特点和神经网络能很好地非线性逼近,又需要较全数据的特点。实证研究结果表明,该组和网络模型获得了更准确的预测值,模型新颖,具有更好的预测精度,可广泛应用于各种预测研究,有较高的推广价值。

The compare study of grey neural network forecast for traffic accidents

Aiming at the difficulties in deciding the variables of BP artificial neural net-work,by means of grey relational analysis to decide the input variables,we put forward a prediction model basing on grey relational analysis BP artificial neural network.The results show that the prediction precision of the model is higher than that of all-input BP artificial neural network.Based on the combination of GM(1,N) forecast and optimized inputs of BP artificial neural network,a new model for traffic(accidents) forecast is put forward here.The new model synthesizes the advantages of both GM(perdiction) which is simple and needs less original data to discover the rule and model and BP neural(network) which possesses the characteristic of nonlinear fitting and needs all-sided original data at the same time.Examples show that the new model gets more accurate results and higher perdiction(precision).It can be widely applied in many kinds of prediction researches and has a bright application future.