论文部分内容阅读
为了进一步提高短时交通参数多步预测的效果,以自适应指数平滑法、BP神经网络法和小波分析理论作为基础模型,利用前一时刻预测误差确定基础模型在组合模型中所占权重,提出了一种交通参数一步预测组合模型;通过分析交通参数合成和分解机理,在分别提出多时间尺度交通参数合成方法和交通参数分解方法的基础上,设计了一种基于多时间尺度一步外推的短时交通参数多步预测方法,采用某大城市感应线圈1 min时间尺度的交通参数数据进行了验证和对比分析。验证结果表明,交通参数一步预测组合模型的预测效果明显优于任一基础模型,且该方法的多步预测效果明显优于循环一步外推短时交通参数多步预测方法。
In order to further improve the effect of multi-step forecasting of short-term traffic parameters, adaptive exponential smoothing, BP neural network and wavelet analysis are used as the basic models, and the weight of the basic model in the combined model is determined by using the prediction error of the previous moment. A combined model of one-step prediction of traffic parameters is proposed. By analyzing the mechanism of traffic parameter synthesis and decomposition, a method of traffic parameter synthesis and traffic parameter decomposition is proposed respectively based on multi-time scales. The multi-step short-term traffic parameters prediction method, using a large city induction coil 1 min time scale traffic parameters data validation and comparative analysis. The results of the verification show that the forecasting effect of the combined model with one-step traffic parameter prediction is obviously better than any of the basic models, and the multi-step prediction effect of the method is obviously better than the one-step extrapolation method of short-term traffic parameter multi-step prediction.