CAJ | 학술논문

混沌时间序列在许多自然经济现象中都存在。目前对其进行预测的常用方法是加权一阶局域法和基于最大Lyapunov指数预测法，加权一阶局域法在计算的过程中，可能导致预测结果呈现出平滑趋势，而在基于最大Lyapunov指数的预测的计算中，则可能出现预测的结果剧烈变动的趋势。因此这两种方法的适用范围不同，在对道路日交通量时间序列进行预测时，由于道路交通量总体来说变化平稳，但有蕴含丰富的波动特征。因此，提出将两种方法有机结合起来，形成组合预测模型，并很好地解决了嵌入维的确定问题，实证表明，预测效果很好。
혼돈시간서렬재허다자연경제현상중도존재。목전대기진행예측적상용방법시가권일계국역법화기우최대Lyapunov지수예측법，가권일계국역법재계산적과정중，가능도치예측결과정현출평활추세，이재기우최대Lyapunov지수적예측적계산중，칙가능출현예측적결과극렬변동적추세。인차저량충방법적괄용범위불동，재대도로일교통량시간서렬진행예측시，유우도로교통량총체래설변화평은，단유온함봉부적파동특정。인차，제출장량충방법유궤결합기래，형성조합예측모형，병흔호지해결료감입유적학정문제，실증표명，예측효과흔호。
Chaotic time series exists in many phenomena. Currently, a common way to predict it is adding - weight one - rank local - region method and index prediction based on largest Lyapunov. Adding - weight one -rank local - region method is likely to cause smoothing trend, and largest - Lyapunov - based prediction possibly produce violent fluctuated results, so they have different scope of application. When predicting time series of road daily traffic volume, we put forward to combine these two methods to form a compounded prediction mode, so as to solve the determination of embedding dimension. Practice showed effect of prediction was excellent.