短时公交客流小波预测方法研究

短期客流表现出不同于中长期客流的特性，本文在研究短期客流序列特性的基础上建立预测方法．采用离散傅里叶变换研究短时公交客流序列的频域特性；基于混沌理论，通过计算Lyapunov指数判断短时客流序列的混沌特性；最终建立短期客流序列的小渡预测方法．研究结果表明短时公交客流序列信号包含高频成份与低频成份，低频成份构成信号主体，高频成份导致信号波动，且短时公交客流序列具有混沌特性，这些特性导致单一方法短时客流预测精度较低；小渡预测方法通过信号分解重构在保留全部客流信息的基础上有效降低了客流信号的波动性，实例分析证明该方法可以有效提高客流预测精度．

Study on Wavelet Forecast Method for Short-term Passenger Flow

Short-term passenger flow exhibits unique characteristics different from the long term passenger flow. This paper studied a short-term passenger flow forecast method based on the short-term passenger flow characteristics. Having summarized the existed research results, DFT was proposed to carry on the frequency domain analysis of the short-term passenger flow. Lyapunov index was calculated to discriminate the chaotic characteristics with the chaos theory. Then, a wavelet forecast method was proposed, which was formed by three steps, namely wavelet decomposition and reconstruction, single branch forecast, and prediction results synthesis. Result of this paper showed that the short-term passenger flow sequence contains low-frequency components as the main composition and high-frequency components resulted in interference signal. In addition, the short-term passenger flow has chaos characteristics; The wavelet prediction method reduced the signal volatility while retained all the passenger flow information. An example analysis proved that wavelet prediction method raised the forecast accuracy effectively.