A Combination Forecasting Model Based on IOWA Operator for Dam Safety Monitoring

Abstract

In prediction of dam safety monitoring, single models such as mathematical model and neural network model are established according to different modeling mechanism. The effective combination of these single models can make full use of all useful information and improve the fitting and forecasting precision. The most commonly used combination method is weighted arithmetic averaging combination forecasting method, in which each single forecasting model is endowed with the same weighted averaging coefficient at various moments. But in fact, each single forecasting model has different performance at various moments, namely, exhibiting higher prediction accuracy at one moment and lower prediction accuracy at another moment. Thus, this combination forecasting method is not absolutely reasonable. Therefore, the concept of induced ordered weighted averaging (IOWA) operator is introduced here, and then a new dam safety monitoring combination forecasting model based on IOWA operator is put forward. In this combination model, each single forecasting model is endowed with weights orderly according to its prediction accuracy at each moment. Case study shows that the proposed combination forecasting model is superior to various single forecasting models. It has high fitting and forecasting accuracy and is applicable to prediction of various dam safety monitoring indexes.