-无偏直接PGM(1-1)模型及其优化.pdfVIP

PGM (1, 1) 陈鹏宇　段新胜 (中国地质大学(武汉)工程学院, 武汉　430074) :无偏直接GM (1, 1)模型在原始数据为纯指数序列时能保持无偏性, 但在原始数据为近似指 数序列时, 不是最佳无偏直接GM (1, 1)模型.提出了 种无偏直接PGM (1, 1)模型, 并以模式搜索 法求解最佳权值p , 同时由于无偏直接GM (1, 1)模型默认经过了初始点, 因此对初始值进行了修 正, 结合具体数据将无偏直接GM (1, 1)模型与无偏直接PGM (1, 1)模型进行了比较, 结果表明无 偏直接PGM (1, 1)模型优于无偏直接GM (1, 1)模型. :无偏直接GM (1, 1)模型;　无偏直接PGM (1, 1)模型;　模式搜索 :N941.5　　　:A 　　　:1672-948X(2009)05-0066-03 Unbiased Direct PGM (1, 1)and Its Optimization Chen Pengyu　 uan Xinsheng (Faculty of Engineering, China University of Geosciences, Wuhan 430074, China) Abstract 　Unbiased direct GM (1, 1)can keep theunbiased characteristicwhen original data is an exponential sequence, but it is not the best unbiased direct grey model when original data is close to an exponential se- quence.An unbiased direct PGM (1, 1)is proposed.The best weight p is worked out by using pattern search method.And becauseunbiased direct PGM (1, 1)pretermits to pass the initial point, so the initial value is a- mended.The comparison between unbiased direct PGM (1, 1)and unbiased direct GM (1, 1)is made.The results demonstrate that the proposed model is superior to unbiased direct GM (1, 1). Keywords 　unbiased direct GM (1, 1);　unbiased direct PGM (1, 1);　pattern search 　　, , GM (1, 1) PGM (1, 1), [1-3] , . PGM (1, 1)GM (1, 1). [4] GM (1, 1) , [5] GM (1, 1) , 1 　GM (1, 1) [6-8] , . x (0) [9] (0) (0) (0) (0) GM (1, 1) x =(x (1), x (2), …, x (n