一个模糊时间序列预测模型的集合

A set of fuzzy time series prediction models

提出一个模糊时间序列预测模型的集合(SPMFTS)。证明了对于任意时间序列S,应用SPMFTS的自动寻优搜索建模法可以筛选出集合的一个模型FMC_j^1(x,f,i),使得预测值C_j(x,f,i)的相对误差均达到0.000 0%。例如预测美国2010—2016年的跑道侵入问题时,建模FMC可以实现的预测值的相对误差均为0.000 0%;对于预测未来的(美国,2017年)跑道侵入数据时,建模FMC实现的预测值也可达到相对误差为0.000 0%。这些预测模型都比灰色模型GM(1,1)和马尔科夫模型的预测精确度高。因此SPMFTS的建模FMC比灰色马尔科夫模型更有优势。

Put forward a set of predict model of fuzzy time series (SPMFTS). It can prove that: for any time series S, the model FMC 1 j(x,f,i ) in the set SPMFTS, can definitely be selected by applying the automatic optimization search modeling method of SPMFTS, makes the predictive value C j(x,f,i) relative error be 0.000 0%. For example, when simulating and predicting runway invasion problems in the United States from 2010 to 2016, FMC can realize predictive value of relative error is 0.000 0%;For the prediction of future (US,2017) runway invasion data, the relative error of the predictive value achieved by the modeling FMC can also reach 0.000 0%. These prediction models are more accurate than the gray model GM(1,1) and markov model. Therefore, the modeling FMC of SPMFTS is more advantageous than the gray markov model.