基于模糊逻辑的转子振动故障诊断决策树C4.5算法模型研究

发布时间：2018-04-25 07:09

本文选题：统计特征参数 + 决策树　；参考：《华北电力大学（北京）》2011年硕士论文

【摘要】：本文采用基于决策树(C4.5算法)和模糊逻辑系统(FIS)相结合的智能故障诊断方法,利用汽轮机实验室中的转子实验台进行试验研究。结合实验设备,拟设计转子正常运转(GOOD);质量不平衡(RMUB);动静碰磨(MSCF);油膜振荡(RWOW)四种振动故障实验,并分别设计三种工作转速(2000rpm,4000rpm,6000rpm)的工作状态下运行。每种故障实验进行50组,其振动信号可从电涡流传感器结合DASP采集分析软件获取,然后对各组实验样本数据计算出统计特征参数,本文利用决策树方法中C4.5算法作为故障特征选择算法从数据样本集中提取相关特征,再利用统计特征参数构建决策树,决策树输出结果用来产生清晰、简洁的if-then规则。然后利用Matlab模糊逻辑工具箱来产生的隶属函数集并确定模糊推理分类结构。通过测试样本对决策树分类和模糊逻辑规则集的准确性进行检验,并将产生的模糊推理机利用Matlab中Simulink工具箱得到最终的简化模型。本文以4000rpm的样本数据为例计算统计参数并构造决策树,得出if-then语句规则和I隶属函数集,最后再用基于转子实验台获得的转子振动信号的测试样本来评估本文使用的C4.5-FIS模型故障诊断方法,再分别评估对于2000rpm和6000rpm两种不同转速条件下这该诊断模型的准确性,结果表明其准确度都很高.故该C4.5-FIS模糊推理模型对于汽轮发电机组故障诊断具有很好的适用性。
[Abstract]:In this paper, an intelligent fault diagnosis method based on decision tree C4.5 algorithm and fuzzy logic system (fish) is proposed. Combined with the experimental equipment, it is proposed to design four kinds of vibration failure tests, such as normal operation of rotor, mass imbalance, dynamic and static impact grinding, oil film oscillation and RWOW, and design three kinds of rotating speed of 2000rpm 4000rpmC6000rpm respectively. The vibration signals can be obtained from eddy current sensor and DASP software, and then the statistical characteristic parameters are calculated for each group of experimental samples. In this paper, C4.5 algorithm in decision tree method is used as fault feature selection algorithm to extract relevant features from data sample set, and then statistical feature parameters are used to construct decision tree. The output result of decision tree is used to produce clear and concise if-then rules. Then the membership function set is generated by Matlab fuzzy logic toolbox and the classification structure of fuzzy reasoning is determined. The accuracy of decision tree classification and fuzzy logic rule set is verified by test samples, and the resulting fuzzy inference engine is obtained by using Simulink toolbox in Matlab to obtain the final simplified model. In this paper, taking the sample data of 4000rpm as an example, the statistical parameters are calculated and the decision tree is constructed. The rules of if-then statements and the set of I membership functions are obtained. Finally, the C4.5-FIS model fault diagnosis method used in this paper is evaluated with the test samples obtained from the rotor vibration signal obtained from the rotor test bench. The accuracy of the model is evaluated for the two different rotational speeds of 2000rpm and 6000rpm, respectively. The results show that the accuracy is very high. Therefore, the C4.5-FIS fuzzy inference model is suitable for fault diagnosis of turbogenerator.
【学位授予单位】：华北电力大学（北京）
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【参考文献】