抽水蓄能电站机组异常状态检测模型研究

针对当前水电机组故障样本少,难以对其开展有效诊断的难题,提出了一种综合考虑有功功率、工作水头等工况参数的基于最小二乘曲面的抽水蓄能电站机组异常状态检测模型,即在深入分析有功功率、工作水头对机组运行状态影响的基础上,确定了机组的健康标准状态,根据机组运行状态在不同功率和不同水头下的特性,划分了不同单元,在不同单元内选取能反映机组运行状态的敏感特征参数,分别建立了基于最小二乘曲面的分布式健康模型,将功率、水头等实时在线数据代入分布式健康模型,通过计算机组健康度建立最终的异常状态检测模型。实例应用结果表明,该模型能有效地挖掘机组海量状态数据和真实可靠地进行在线状态评估,从而实现机组异常状态的早期预警。

Abnormal Condition Detection Model for Pumped Storage Unit

An abnormal condition detection model of pumped storage unit is proposed in this paper. This model considers comprehensively active power and working head by using least squares surface. Firstly, the impact of active power and working head to unit operating conditions is analyzed. The standard health conditions of unit are determined. Then, the unit operating conditions are divided into condition elements, according to the characteristics of unit conditions in different power and working head. The sensitive characteristic parameters of operating conditions are selected. The health models based on least squares surface are founded in different condition elements. Finally, the health degree of pumped storage unit is obtained when real-time power and working head are input into distributed health model. And the abnormal condition detection model is founded ultimately. The proposed model is applied to the abnormal vibration detection for pumped storage unit. The results show that the model can effectively exploit unit massive condition data. It can truly and reliably assess unit on-line condition, realize early warning of unit abnormal or failure condition.