基于gbm模型的复杂机电产品故障部位频率动态分析 张根保,喻春 .docVIP

基于GBM模型的复杂机电产品故障部位频率动态分析 张根保,喻春 （重庆大学　机械工程学院,重庆　400044） 摘要：为了更充分有效地利用故障模式与影响分析(FMEA)中产品故障部位频率的信息,使故障部位频率变化的趋势及时反馈到产品设计和生产改进环节,提出基于广义信度马尔科夫（GBM）模型的故障部位频率动态分析方法。首先分析马尔科夫链描述产品故障部位频率动态变化的可行性,然后引入广义证据理论解决经典马尔科夫模型中状态空间数值的微变导致预测结果发生跳跃的问题,建立了适用于开放世界的广义信度马尔科夫模型。以某机床企业所生产的数控车床为对象,应用该方法展开了故障部位频率动态分析研究,最后针对各个故障部位频率的变化情况提出了相应的应对措施,证明了该方法的可行性及有效性。 关键词：复杂机电产品；故障部位频率；广义信度马尔科夫模型；动态分析 中图分类号：TB114.3 Dynamic analysis of fault part frequency for complex electromechanical products based on GBM model ZHANG Gen-bao, YU Chun (College of Mechanical Engineering, Chongqing University, Chongqing 400044) Abstract: In order to utilize information of fault part frequency(FPF) in fault modes and effects analysis (FMEA) more fully and effectively, and to feed back FPF’ trend timely into design and production segments, a method of FPF dynamic analysis based on generalized belief Markov (GBM) model was proposed. Firstly,the feasibility of Markov chain describing FPF dynamic change was analyzed and then a Generalized Evidence Theory(GET) was introduced to solve the prediction jump problem leaded by slight variation of state value in classical Markov chain model ,at last a GBM model which apply to open world was constructed. Taking the CNC lathes manufactured by a machinery enterprise as an example, the GBM model was applied to dynamic analysis of FPF and then corresponding measures were put forward for FPF’s changes, which prove the feasibility and effectiveness of this method . Key words: complex electromechanical products; fault part frequency; generalized belief Markov(GBM); dynamic analysis 0 引言 复杂机电产品是由机械结构、电器设备、控制装置、检测装置等有机组合在一起的复杂系统[1][4-5],特别是在预测技术中有着深入的应用。传统的马尔科夫链对易于区分的离散状态有着很好的预测效果。,当状态空间是近似连续的数值时,其预测结果可能会随着数据的微小变动产生状态的飞跃。 ,当状态发生转移时,可能无法确定马尔科夫链在下一步处于何种唯一的状态,即状态的描述无法确切地给出。,主要表现为研究对象的状态无法明确的划分。所处状态已知的情况下,过程在时刻所处状态的条件分布与过程在时刻之前所处的状态无关。而参数集与状态集均为离散的马尔科夫过程称为马尔科夫链[10]。下面就马尔科夫链在描述产品故障部位频率动态变化的适用性及可行性进行简要说明。 记表示同型号的一批产品第i个部位发生故障的频率在时间周期t内的统计值,则表示第i个故障部位频率动态变化这一随机过程,为状态空间,由于故障部位频率的统计值在产品使用