基于自适应加权和D-S证据理论的风电机组故障诊断  

Wind Turbine Fault Diagnosis Based on Adaptive Weighting Algorithm and D-S Evidence Theory

作　　者：神显豪[1] 张祁[1,2] 

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541004 [2]熊本大学,日本熊本860-8555

出　　处：《机床与液压》2014年第7期148-151,140,共5页Machine Tool & Hydraulics

基　　金：国家自然科学基金资助项目(51167004);广西教育厅科研项目(200911LX131)

摘　　要：由于风电机组系统相当复杂,故障原因及其现象不成简单或线性对应关系,单一检测不能够满足诊断需要。针对这一问题,将无线传感器网络(Wireless Sensor Network)中信息融合的理论和方法应用于风电机组状态监测和故障诊断中,使采集到的海量数据分别进行信号层与特征层两个层次的信息融合,运用自适应加权融合算法降低网络的数据冗余和传输能量消耗,利用高斯隶属度函数获得基本概率的赋值,提高了D-S证据理论数据的可靠性,改进的证据组合方法提高了故障识别能力。最后,对风电机组齿轮箱的故障诊断进行仿真实验,实验结果验证了该方法具有较高的诊断精度,明显提高诊断的可信度。Since the system of wind turbine is quite complex, the relationship between faults and phenomena is not simple or lin- ear, the diagnostic requirements could not be met by single detection. Aimed at this problem, the information fusion theory of Wire- less Sensor Network was applied in wind turbine＇s state monitoring and fault diagnosis, which made information fusion separately in two levels of signal level and characteristics level of a large amount of collected data. By using self-adaptive weighting fusion algorithm, the data redundancy and transmission energy consumption of network were reduced, and using the Gauss membership function, the basic probability assignment was obtained, which enhanced the D-S evidence theory data reliability and improve the ability of fault i- dentification. Finally, a simulation experiment of fault diagnosis was held on gearbox of wind turbine. The experimental results prove that the method has a high diagnostic accuracy, and obviously improves diagnostic reliability.

关 键 词：无线传感器 信息融合 自适应加权 D-S证据理论 基本概率赋值 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]