【摘要】 目前,设备健康预测问题的研究大都在样本数据准确下进行,而在样本数据不确定下的研究却很少.因此,针对不确定样本数据下设备健康预测问题,提出了集成Dempster-Shafe(DS)证据理论与马尔可夫模型(MM)的联合优化模型.首先,基于马尔可夫模型,利用DS证据理论建立状态识别框架.其次,用区间数表示不确定的数据,并利用区间数之间的距离和相似度作为产生基本概率赋值(BPA)的证据,为了使预测结果更加可靠,采用Pignistic概率转换将BPA转化为基础状态的概率分布.最后,通过案例分析对模型进行评价和验证.结果表明,提出的方法能够有效解决数据不确定下的设备健康预测问题.更多还原

【Abstract】 At present,most of the researches on equipment health prognostic information executed under the certain sample data,but there are fewstudies under the uncertain sample data. Therefore,this paper develops a joint optimization model of Dempster-Shafe evidence theory and Markov model for the problem of equipment health prediction under the uncertain sample data. First,based on Markov model,DS evidence theory is used to build the state recognition structure. Secondly,the uncertain data is showed by interval number and basic probability assignments( BPA) are generated based on the distance and similarity between interval numbers. To make the results more reliable,BPA is transformed into the probability distribution of basic states by Pignistic probability transform.Finally,a case study is used evaluated the performance of the model. The results showthat the proposed method could effectively solve the problem of equipment health prognostic under the uncertain data.更多还原