科学技术与工程
科學技術與工程
과학기술여공정
Science Technology and Engineering
2015年
29期
21-28
,共8页
极大似然函数(MLE)%Fisher信息矩阵加速因子%Wiener过程%Bayes理论%先验分布%后验分布
極大似然函數(MLE)%Fisher信息矩陣加速因子%Wiener過程%Bayes理論%先驗分佈%後驗分佈
겁대사연함수(MLE)%Fisher신식구진가속인자%Wiener과정%Bayes이론%선험분포%후험분포
maximum likelihood function (MLE)%Fisher information matrix%acceleration factor%Wiener process%Bayes theory%prior distribution%posterior distribution
针对新研产品加速因子难以确定的问题，提出基于随机维纳过程的产品加速因子分布确定方法。首先根据相似产品信息，利用整体极大似然函数（ MLE）和Fisher信息矩阵确定相似产品加速因子先验分布；其次根据专家经验信息给出新研产品加速因子先验分布；再其次通过加权融合思想，将相似产品加速因子先验分布和专家经验加速因子先验分布融合，给出新研产品加速因子最终的先验分布；然后根据新研产品内场试验信息，给出Wiener过程的参数估计；最后利用Bayes理论，充分利用产品低层试验信息对加速因子分布进行更新得到后验分布。以某型新研加速度计为实例，验证了所提出的方法适用性和有效性。
針對新研產品加速因子難以確定的問題，提齣基于隨機維納過程的產品加速因子分佈確定方法。首先根據相似產品信息，利用整體極大似然函數（ MLE）和Fisher信息矩陣確定相似產品加速因子先驗分佈；其次根據專傢經驗信息給齣新研產品加速因子先驗分佈；再其次通過加權融閤思想，將相似產品加速因子先驗分佈和專傢經驗加速因子先驗分佈融閤，給齣新研產品加速因子最終的先驗分佈；然後根據新研產品內場試驗信息，給齣Wiener過程的參數估計；最後利用Bayes理論，充分利用產品低層試驗信息對加速因子分佈進行更新得到後驗分佈。以某型新研加速度計為實例，驗證瞭所提齣的方法適用性和有效性。
침대신연산품가속인자난이학정적문제，제출기우수궤유납과정적산품가속인자분포학정방법。수선근거상사산품신식，이용정체겁대사연함수（ MLE）화Fisher신식구진학정상사산품가속인자선험분포；기차근거전가경험신식급출신연산품가속인자선험분포；재기차통과가권융합사상，장상사산품가속인자선험분포화전가경험가속인자선험분포융합，급출신연산품가속인자최종적선험분포；연후근거신연산품내장시험신식，급출Wiener과정적삼수고계；최후이용Bayes이론，충분이용산품저층시험신식대가속인자분포진행경신득도후험분포。이모형신연가속도계위실례，험증료소제출적방법괄용성화유효성。
Aiming at the problem that the acceleration factor of the new product is difficult to determine , the distribution of product acceleration factor based on stochastic Wiener process is proposed .First, according to the similar product information , determine the similar products to accelerate factor prior distribution using the overall maximum likelihood function (MLE) and the Fisher information matrix;Secondly, according to the expert experi-ence information gives new research products acceleration factor prior distribution ;Again followed by weighted fu-sion theory , fusioning accelerated factor prior distribution of the similar products and acceleration factor prior distri -bution based on expert experience gives a accelerating factor final prior distribution of new research products ;then, according to new research products infield test information , given the Wiener process parameter estimation;Finally, by the Bayes theory , make full use of products low layer test information to update acceleration factor distribution to get acceleration factor posterior distribution .Taking a case study of the accelerometer , the suitable and validity of the proposed method has been proved .
