摘要
针对故障诊断中计算量大,模式分类复杂的问题,提出了一种基于粗糙集的分形容错故障诊断方法。首先对可能的诊断属性用粗糙集约简的方法进行故障特征提取;然后计算所采集的故障数据的分形维数,并用回归辨识方法得到维数序列的数学模型;利用所建立的数学模型可完成对故障的分类和故障程度的辨识。仿真结果表明了该方法的有效性。该方法解决了单独利用分形几何方法无法对故障程度进行辨识的问题,简化了计算,并为高可靠性设备的故障诊断提供了新的思路。
To solve the problem of complex computing and classifying of patterns in fault diagnosis, a fractal fault-tolerant diagnosis method based on rough reduction is presented.First, using rough reduction, the fault features is extracted according to the possible diagnosis attributes.Then, compute a series of fractal dimensions of fault states data,and constitute mathematic models of these dimensions by recursive identification method. the degree of the faults can be classified and identified by the mathematical models.The results of simulations indicate the validity of this method.Using this method can resolve the problem that can not be identified the degree of the faults by fractal only,predigest computation,and a new idea is provided to diagnosis of high reliability equipments.
出处
《计算机测量与控制》
CSCD
2005年第6期511-513,共3页
Computer Measurement &Control
基金
国家自然科学基金重点项目(60234010)
航空科学基金项目及国防基础科研资助项目(02E52025)
国防基础科研项目(K1603060318)。
