基于水电机组运行工况的水轮机压力脉动诊断策略

摘要

The pressure pulsation during operations of a hydroelectric generating unit (HGU )is an inevitable phenomenon.Diagnosing and assessing accurately its state are of particular importance.As the pressure pulsation was closely related to the operating state of a HGU,a novel diagnosis strategy based on the HGU's working condition was proposed here.Firstly,the contribution rates of condition parameters based on the mutual relation analysis were computed to extract the superior condition parameters.The superior condition parameters and the time-frequency featares of pulsation signals were fused,from the fusion information the eigenvectors of pressure pulsation were extracted.Then,the support vector machine (SVM)and the extreme learning machine (ELM)were used to diagnose the pulsation state.Finally,in order to achieve a quantitative diagnosis for pressure pulsation,the fuzzy evaluation function for pressure pulsation versus the unit degradation level was proposed with the fuzzy evaluation theory.The results of a real example showed that this diagnosis strategy is better than the traditional time-frequency diagnosis strategy,and it is of practical guiding significance for safety and stable operation of a HGU.%水轮机压力脉动是水电机组运行过程中不可避免的现象，准确地识别和定量诊断脉动状态对机组高效稳定运行尤为重要。为此，提出了基于水电机组运行工况的水轮机压力脉动诊断策略，以水电机组实际运行工况为切入点，通过分析工况参数与压力脉动的非线性相关关系，得到影响压力脉动的主要相关工况参数，提取了融合机组运行工况参数与脉动幅值特性的特征向量，并利用支持向量机（SVM）与极限学习机（ELM）两种诊断方法进行脉动状态定性诊断。研究压力脉动幅值历史统计规律，提出了脉动状态对机组劣化程度的模糊评估函数，反演了定性诊断结果与机组健康状态的映射关系，实现压力脉动的定量诊断。实例验证表明，相对于仅基于脉动幅值的诊断策略而言，该方法诊断准确率更高，定量诊断指标可靠有效。这为水电机组安全稳定运行提供技术保障。