论文部分内容阅读
在某停车场泊位变化实测数据分析的基础上,建立了一个停车场有效泊位占有率短时预测模型.南京某停车场实测泊位数据分析表明,在不同的观测尺度上,停车场有效泊位占有率具有很强的周期性和相似性,但观测尺度越小,随机性越强.基于有效泊位占有率的这种短时变化特性,提出采用小波分析和加权马尔可夫组合模型对有效泊位占有率进行短时预测.首先,通过选择合适的小波函数对有效泊位占有率时间序列进行多分辨率的小波分解,并对低频信号与高频干扰信号分别进行重构,然后对重构后的基本信号和不同分辨率的干扰信号分别建立加权马尔可夫预测模型,最后对各自外推的预测结果进行合成,得到最终预测结果.实例分析表明,所提出的预测模型对有效泊位占有率的短时预测结果是有效的,但模型的预测精度依赖于有效泊位占有率数据库的实时更新.
Based on the analysis of measured data of berth changes in a parking lot, a short-term forecasting model of effective berth occupancy in a parking lot is established.Analysis of measured berth data of a parking lot in Nanjing shows that at different observation scales, the effective berth occupancy rate However, the smaller the observation scale, the stronger the randomness.According to the short-term variation of the effective berth share, this paper proposes that the combination of the wavelet analysis and the weighted Markov model to calculate the effective berth occupancy For short-term prediction.Firstly, multi-resolution wavelet decomposition of the effective time series of berths is carried out by selecting the appropriate wavelet function and the low-frequency signals and high-frequency interference signals are respectively reconstructed, and then the reconstructed basic signals And the interference signals with different resolutions are used to establish the weighted Markov forecasting models respectively, and finally the prediction results of each extrapolation are synthesized to obtain the final prediction results.Example analysis shows that the proposed prediction model can predict short-term effective berth occupancy The results are valid, but the prediction accuracy of the model depends on the real-time update of the effective berth occupancy database.