Maximum Possibility Estimation Theory With Applications In Survey Data Processing

Traditional parameter estimation theories are based on the probability theory, which assumes that the survey data are all stochastic variables. Actually, the uncertainty of survey data composes a lot of factors, not only stochastic variable. So traditional parameter estimation theories are not strict. To overcome their shortcoming, Prof. Wang Xinzhou raised the maximum possibility estimation theory, which avoids the assumption and processes the uncertainty of survey data precisely. The maximum possibility estimation theory was raised recently and still have some problems. To solve these problems, the author made some researches on this theory in-depth, supervised by Prof. Wang Xinzhou.Firstly, this paper introduced the possibility theory and fuzzy number, and devised a new fuzzy number—logarithm fuzzy number. Secondly, the main principium and method of the maximum possibility estimation theory were discussed. And then the nonlinear programming model of the maximum possibility estimation theory was devised. The author used the mix intelligent optimization algorithms of Genetic Algorithm and Simulated Annealing to compute the nonlinear programming of the maximum possibility estimation theory. The survey data qualities were calculated also.Finally, the author use the maximum possibility estimation theory to solve survey data-processing problems, and validated it's rationality. The abilities of detect and restrain gross errors of that theory were discussed also.It shows that the maximum possibility estimation theory can avoid the shortcoming of traditional parameter estimation theories, and it also enriched the theories of survey data processing.