灰色GM(1，1)模型预测沉降的局限性分析

Limitation analysis of GM(1,1) gray model in settlement prediction

考虑现场沉降监测数据的不等时间间隔性及数据的不断更新性,建立了不等时距等维新息GM(1,1)沉降预测模型并研发了相应的预测程序RIID,将其应用于实际工程的沉降预测,验证了预测模型合理性和程序的可行性。分析了实测数据时间间隔和预测步数对GM(1,1)模型预测精度的影响。结果表明:数据时间间隔相差太大,将导致模型失真;GM(1,1)模型只能进行短期预测,若要预测未来较长时间内的沉降,必须有新增数据,这就使得该模型在实际工程中的应用受到限制。

The recent information of identical dimension GM（ 1,1 ） grey model for unequal time-interval data was established to predict settlement and the program RIID was developed, in which these characteristics of in - situ moni-tored data such as time-intervals of monitored were unequal and the data are uninterrupted updated were taken into account. The model was performed to predict settlement of practical project, the rationality of model and the feasibility of program were verified. The effect of test interval and of predicted steps on prediction precision of GM（1,1） model was investigated. The results indicate that too large difference of time- interval will induce the model distortion. GM （ 1,1） model is available only for short term settlement prediction and the new data is absolute necessary for predicting settlement in longer term, which limit the application of this model in practical project.