为提高船舶交通量的预测精度,在BP神经网络的基础上结合马尔科夫预测模型建立一个新的预测模型.采用通过长江九江大桥的月度船舶交通量数据进行模型训练、验证和预测,求出相对残差值,将相对残差的前8项归一化后划分为3个状态,利用马尔科夫预测模型修正BP神经网络的预测值.该新模型将BP神经网络的相对残差值区间从[-12.9%,12.3%]降低至[-9.9%,5.4%].该模型能提高船舶交通量的预测精度,用于预测船舶交通量是可行的.%To improve prediction accuracy of the ship traffic flow, a new prediction model is established based on BP neural network combined with Markov prediction model.The month data of ship traffic flow crossing the Jiujiang Yangtze River Bridge are used to do the model training, verification and prediction.The relative residuals are calculated, the top 8 relative residuals are divided into 3 states, and Markov prediction model is used to correct the prediction values by BP neural network.The relative residual interval of BP neural network is improved from [-12.9%, 12.3%] to [-9.9%, 5.4%] by the new model, which shows that the new model can improve the prediction accuracy, and is feasible for the prediction of the ship traffic flow.
展开▼