基于天牛须小波神经网络的公交到站时间预测

Bus arrival time prediction based on wavelet neural network optimized by Beetle Antennae Search

通过对公交运行环境的实际分析,考虑工作日和非工作日的运行特性差异,提出了一种基于天牛须搜索算法的小波神经网络(BAS-WNN)公交到站时间预测模型。该模型利用寻优性能更强的天牛须搜索算法优化WNN的初始参数,使得WNN对时间序列的预测具有更好的性能。最后,利用行车历史数据对神经网络进行训练和建模来实现到站时间的准确预测,将该优化算法与传统的WNN算法和Elman神经网络算法用MATLAB分别仿真测试,对比结果显示,无论工作日还是非工作日,BAS-WNN预测模型对公交到站时间的预测均具有更高的准确性且结果更加稳定。

Through analysis of the real bus operating environment,considering the difference in operating characteristics between working days and non-working days,a new bus arrival time predicting method was proposed based on the beetle Wavelet neural network(BAS-WNN)prediction model of search algorithm.This model uses the beetle whisker search algorithm with stronger optimization performance to optimize the initial parameters of the WNN so that the WNN has better performance in the prediction of time series.Finally,the historical driving data was used to train and model the neural network to achieve accurate prediction of the arrival time.The optimization algorithm is simulated with the traditional WNN algorithm and the Elman neural network algorithm in MATLAB.The comparison results show that no matter the working day or on non-working days,the BAS-WNN prediction model has a higher accuracy in predicting bus arrival time and the results are more stable.