Bayesian Model Of Ice Thickness And Its Applied Research Based On The Error Correction

Forecast of ice coating on electric wires is an important design criterion and indispensable indicator of the assessment in power industry under snow weather disasters. Currently, there are many ice thickness models, most of which conduct mechanism research on ice coating from perspectives of meteorology, fluid mechanics and thermodynamics, which can not fully predict the disaster when disasters occur frequently. The Bayesian model of wire ice thickness based on error correction, which fully considers the prior information, can well predict the ice thickness values only through the ice thickness data.The thesis gives a detailed description of the time series model, the maximum likelihood estimation and the Bayesian estimation, and focuses on the choice of prior distribution of the Bayesian method and the posterior distribution of the MCMC method. On this basis, a Bayesian inference is conducted on the constructed time series model of wire ice coating, Bayesian inference theory is analyzed under the normal-Weibull prior distribution and non-information prior distribution, and an error correction is done on the time series model. Through the establishment of the error-correcting ice thickness Bayesian model, the predictive analysis on the ice coating data of Hunan Chenzhou 57972 site and the extreme ice coating data of Nanyue 57776 site during the ice storm in 2008, and the comparison between the predictions together with the maximum likelihood estimation of the error correcting model and the non-correcting results of the Bayesian estimation, the superiority of the error-correcting Bayesian model is verified. The author hopes to provide an empirical basis and decision support for the safe operation as well as the disaster prevention and mitigation of the power grid with the prediction and results of the model.