摘要
介绍了置信规则库转换(BRB-transformation,BRB-T)模型以及运用置信规则库(belief rule base,BRB)推理进行寿命评估的过程。针对当前的置信规则库学习方法还存在依赖初始解、规则条数偏多、系统结构复杂的问题,结合装备贮存寿命评估的BRB-T模型,提出了基于差分进化算法(DE)对该模型的BRB参数进行学习的优化方法。最后,通过某航天设备的失效数据进行了验证。结果表明,该优化方法能够有效地对模型的BRB参数进行优化;并且优化后的模型可以准确地对产品的寿命进行评估。
Belief rule base transformation (BRB-T) model and the process of BRB reasoning is proposed.In order to obtain a more accurate belief rule base (BRB) system,several parameter learning methods are proposed for training the BRB-T model.These learning methods essentially rely on the initial set of the BRB parameters and have too more rules,which leads to a more complex structure.A parameter optimization approach based on the differential evolutionary (DE) algorithm is proposed for training the BRB-T model.The main feature of the proposed method is that it is not necessary to set initial parameters and can reduce the number of belief rule.Finally,a practical space product life assessment with failure data is studied to validate the efficiency and accuracy of the BRB system applied proposed method.
出处
《科学技术与工程》
北大核心
2015年第1期245-249,284,共6页
Science Technology and Engineering
关键词
置信规则库
证据推理
寿命评估
差分进化算法
belief rule base
evidential reasoning
life assessment
differential evolutionary algorithm