基于灰色—小波神经网络的有效泊位预测

针对停车场有效停车泊位的变化特征,提出了基于灰色—小波神经网络的组合模型.先通过灰色单因素预测模型对有效停车泊位时间序列进行修正处理,再基于分步式小波神经网络模型对修正预测值进行运算,并通过马克科夫链预测模型得到更精确的预测区间,并利用实际案例分析,对模型的预测精度、稳定性、拟合度和训练时间进行了评价.研究表明,灰色—小波神经网络预测模型可降低初始数据波动性的干扰,与传统神经网络相比,预测结果误差波动性降低了10%~19%,稳定性提高了27%~33%,拟合度提高了10%~15%,精确度明显提高.

Forecasting of Effective Parking Space Based on Grey-distributed Wavelet Neural Network Model

Aiming the variation characteristics of the effective parking space, a combination model based on the grey single factor forecast model and the wavelet neural network model is proposed to forecast the effective parking space. Firstly, the effective parking space time series are revised through the grey single factor forecast model in order to reduce the random fluctuation. Then the data is input in the distributed wavelet neural network model and corrected though the Markov Chain forecast model to get the finial forecast results. The prediction accuracy, stability, fitting degree, and training time of the proposed model are evaluated by a case study. Results show that the proposed model can reduce the variation interference of the initial data. Comparing with the traditional neural network, the proposed model could reduce the error fluctuation 10%~19%, increase the stability 27%~33%, improve the fitting degree 10%~15% improved, and has higher accuracy.