段炼达,刘晓波.基于摆度大数据的水轮发电机故障预测方法研究[J].中国水利水电科学研究院学报,2017,15(6):439-443
基于摆度大数据的水轮发电机故障预测方法研究
Study on failure prediction for hydro-generator based on big data of run-out value
投稿时间:2016-12-27  
DOI:10.13244/j.cnki.jiwhr.2017.06.005
中文关键词:  大数据  水电厂  机器学习  蒙特卡洛方法  故障预测  摆度
英文关键词:big data  hydropower plant  machine learning  Monte Carlo method  failure prediction  run-out value
基金项目:
作者单位
段炼达 中国水利水电科学研究院 北京中水科水电科技开发有限公司, 北京 100038 
刘晓波 中国水利水电科学研究院 北京中水科水电科技开发有限公司, 北京 100038 
摘要点击次数: 323
全文下载次数: 374
中文摘要:
      水电厂积累的海量生产数据对发电机故障预测具有重要价值。本文采用蒙特卡洛方法对摆度大数据分析处理以进行故障预测。选取了某水电厂水轮发电机下导轴承摆度X峰峰值作为研究对象,在处理453 601组摆度数据时,选取3种典型工况进行研究。引入归一化加权平均值作为统计量,使用蒙特卡洛方法进行分析,发现该统计量服从正态分布。通过监测该统计量的实时值,依据正态分布3σ准则对水轮发电机工作状态做出定性评价,依据计算出的概率P值对水轮发电机工作状态定量评价,利用这些评价可以进行故障预测。
英文摘要:
      The hydropower plants have accumulated mass production data, which is of great value to failure prediction. In this paper, the Monte Carlo method is applied to analyze the big data of run-out value to predict the failure. The X peak-peak values of the run-out of the lower guide bearing of a hydropower plant are taken as the investigation object. In the treatment of 453,601 sets of the run-out value, three typical operating conditions are selected to study. The normalized weighted average is used as the statistic, and the Monte Carlo method is adopted to analyze it. It is found that the statistic obeys normal distribution. By monitoring the real-time value of the statistic,the working status of the hydro-generator is evaluated qualitatively according to the normal distribution 3σ criterion, and evaluated quantitatively according to the calculated probability P-value. Failure can be predicted by using these evaluations.
查看全文  查看/发表评论  下载PDF阅读器
关闭