考虑故障关联不确定性的海上风电机组预防性维护策略优化方法  

Preventive Maintenance Decision-Making for Offshore Wind Turbine Considering Failure Correlation Uncertainty

作　　者：王昊[1] 刘璐洁 陈龙[1] 符杨 WANG Hao;LIU Lujie;CHEN Long;FU Yang(Engineering Research Center of Offshore Wind Technology Ministry of Education(Shanghai University of Electric Power),Shanghai 200090,China)

机构地区：[1]教育部海上风电技术工程研究中心(上海电力大学),上海200090

出　　处：《上海电力大学学报》2024年第4期331-339,共9页Journal of Shanghai University of Electric Power

基　　金：上海市教育委员会科研创新计划项目(2021-01-07-00-07-E00122)。

摘　　要：针对海上风电机组部件故障关联作用的不确定性量化及维护策略优化问题,提出了一种考虑故障关联不确定性的海上风电机组预防性维护策略优化方法。首先,采用区间分析方法对故障关联不确定性进行量化,引入区间故障关联系数描述故障关联的不确定性对机组部件可靠度的影响。其次,根据部件运行状态信息,利用神经网络捕捉故障关联图的时空特征,构建基于图注意力-门控循环单元神经网络(GAT-GRU)的故障关联预测模型,形成区间故障关联系数矩阵。最后,以维护费用最低为目标函数,部件可靠性指标为约束条件,构建了考虑故障关联不确定性的联合维护模型,优化各部件的预防性维护时间区间。算例仿真结果验证了所提方法的有效性和经济性。Aiming at the problem of uncertainty quantification and maintenance decisionmaking of the failure correlation effects of offshore wind turbine components,an optimisation method of preventive maintenance strategy for offshore wind turbines considering the failure correlation uncertainty is proposed.Firstly,the interval analysis method is used to quantify the uncertainty of failure correlation,and the failure correlation interval model is constructed to obtain the reliability interval of components under the influence of uncertainty.Secondly,according to the running state information of the components,the neural network is used to capture the spatiotemporal characteristics of the fault correlation graph,and a fault correlation prediction model based on GAT-GRU is constructed to form an interval fault correlation coefficient matrix.Finally,taking the minimum maintenance cost as the goal and the component reliability index as the constraint,a preventive maintenance model considering the uncertainty of fault correlation is constructed to optimize the preventive maintenance time interval of each component.The simulation results verify the effectiveness and economy of the proposed method.

关 键 词：海上风电机组 故障关联 不确定性 区间分析法 神经网络 预防性维护 

分 类 号：TM315[电气工程—电机]