分形-混沌混合预测模型在大坝安全监测中的应用

如果能针对大坝监测效应量测值序列建立一种不依赖于环境因子,而又综合考虑序列中蕴含的确定性分量和混沌分量的模型,则可以解决常规统计模型由于模型因子选择不当和环境量观测误差引起的模型失真问题.在对确定性分量、混沌分量和随机性分量可预测性分析的基础上,首先利用分形插值算法建立效应量确定性分量预测模型,然后对实测数据和确定性分量预测结果间的误差序列通过相空间重构建立混沌分量预报模型,再以二者叠加组成最终混合预测模型.计算实例表明,在不依赖于环境因子的情况下,该模型比常规统计模型有更高的预测精度.

Application of fractal and chaotic prediction model to dam safety monitoring

The observed data sequence in dam safety monitoring is usually composed of certainty component,chaotic component and random component.Though the statistical model can make good prediction,it can not reflect the chaotic component of the effect variables.Many environmental factors must be determined before the statistical model could be used.The precision of the model will be affected if the environmental factors are not determined properly or the environmental observed data error is kind of large.Through the analysis of the certainty component,chaotic component and random component of the observed data sequence,fractal forecasting model and chaotic forecasting model are respectively established for the prediction of the certainty component and chaotic component based on fractal and chaotic theories;and the hybrid forecast model is established by the superposition of these two models.The application of the hybrid model indicates that it could get same or even better forecasting result than the traditional statistical model;and it avoids the using of environmental factors.