基于支持向量机的风力发电机组故障诊断预警模型
A Support Vector Machine-based Fault Diagnosis and Prognosis Model for Wind Turbines
吉思良 1张峰 1孙海星 1王家宝 1孙也棋1
作者信息
- 1. 国家电投集团河北电力有限公司张家口分公司,河北 张家口 075000
- 折叠
摘要
当前的风力发电机组故障诊断预警模型设置的矩阵一般为单元式的,预警的范围受到限制,导致预警响应时间延长,为此提出对基于支持向量机的风力发电机组故障诊断预警模型的设计与验证分析.根据当前的测定需求及标准,提取故障诊断预警特征量,采用多目标的方式,打破预警范围的限制,设计多目标诊断预警矩阵,以此为基础,构建支持向量机故障诊断预警结构,采用阶次分析实现故障诊断预警处理.测试结果表明:对比传统机器学习风力发电机组故障诊断预警模型测试组、传统 MSK-CNN和多源机电信息融合风力发电机组故障诊断预警模型测试组,此次所设计的支持向量机风力发电机组故障诊断预警模型测试组最终得出的预警响应时间被较好地控制在 0.25s 以下,说明在支持向量机的辅助下,所设计模型对故障诊断预警的效率较高,针对性更强,具有实际的应用价值.
Abstract
Current matrix for fault diagnosis and prognosis models of wind turbines is generally unit based,and the scope of warning is limited,resulting in an extension of warning response time.Therefore this paper proposes design and valida-tion analysis of a support vector machine-based wind turbine fault diagnosis and prognosis model.Based on current meas-urement requirements and standards,fault diagnosis and warning feature quantities are extracted.By adopting a multi-ob-jective approach and breaking the limitations of warning range,a multi-objective diagnosis and warning matrix are de-signed.And based on this,and SVM fault diagnosis and prognosis structure is constructed and order analysis is used to a-chieve fault diagnosis and prognosis processing.The test results show that compared with conventional machine learning model,conventional MSK-CNN,and multi-source electromechanical information fusion,the designed model has a better warning response time controlled below 0.25 second,indicating that the assistance of support vector machine facilitate high efficiency of fault diagnosis and prognosis,with stronger specificity and practical application value.
关键词
支持向量机/风力发电/发电机组/故障诊断/预警模型/诊断识别Key words
support vector machine/wind generation/generator set/fault diagnosis/prognosis model/diagnostic identi-fication引用本文复制引用
出版年
2024
NETL
NSTL
万方数据