Ill-conditioned problems of dam safety monitoring models and their processing methods

Abstract

The focus of this paper is the ill-conditioned problems in the dam safety monitoring model. The reasons to give rise to the ill-conditioned problems in statistical models, deterministic models and hybrid models are analyzed in detail, and the criterions for ill-conditioned models are investigated. It is shown that safety monitoring models are not easy to be ill-conditioned if the number of influence factors is less than seven. Moreover, the models have a high accuracy and can meet the engineering requirements. Another frequently encountered problem in establishing a safety monitoring model is the existence of inflection points, which are often present in the mathematical model for the hydraulic components in deterministic models and hybrid models. The conditions for inflection points are studied and their treatments are suggested. Numerical example indicates that the treatments proposed in this paper are effective in removing the ill-conditioned problems.