Analysis Methods And Applications Research On Umpty Uncertainty Problems In Large Dam Monitoring And Control

The uncertainty problems in dam safety monitoring were studied in this dissertation. Based on regarding dam and its bedrock as an uncertainty system, the international actuality of correlative research was reviewed, and then closely combining with engineering practice, umpty analyses methods and their applications for uncertainty problems in dam safety monitoring were studied. The research contents and main results acquired are as follows:(1) The research thoutht was put forward for the uncertainty problems studying in dam safety monitoring, and a basic frame system for uncertainty problems research was set up, which takes the uncertainty dam monitoring system as study object, mainly utilizes uncertainty mathmatics as estimation methods, and regards the model simulation and forcast of uncertainty information as the core of studying. At the same time, the basic principles, theories, modeling methods and steps, model precision and its evaluating standards were studied.(2)Based on uncertainty inverse analyses on material parameters, the studies on fiducial monitoring state correction were carried out. A Bayesian uncertainty inverse analysis method based on POME (Principle of Maximum Entropy) was advanced for calculating optimization material parameters of dam concrete and the bedrock. Then by utilizing the correcting methods of fuzzy clustering and analogy proposed, the optimization material parameters gained were used to correct the fiducial monitoring states, which effectively overcomes the deficiency of present method only according to engineers experience and settles the base for dam safety analysis and evaluation.(3) Aimed at the multicollinearity among independent variable factors and the relevant uncertainty problems of dam safety monitoring and control model, the partial least squares regression-uncertainty analysis (PLSR-UA) method for synthetical analysis was proposed to simulate and analyze observation data. The PLSR model simulation method was integrated with uncertainty analysis of data connotation in dam safety monitoring, which solves the prevalent problem of accuracy of the least-squares regression model always being affected by independent variables multicollinearity.(4) Researches on distemper dam failure risk and life loss forecast for dam uncertainty system were carried out. An analysis method of grey-stochastic risk probability was put forward to estimate dam failure risk. The risk probability obtained falls in a grey space, which satisfactorily reflects the uncertainty in dam failure risk. Furthermore, based on systematical analyses on the uncertainty factors which affect the life loss forecast in dam failure, the life loss predicting method according to the estimation of probability values of life loss were proposed, and the suggestion values for 15 kinds of flood scenes caused by dam failure were gived, which overcomes the forecast shortage simply utilizing mathematics model.