为提升电站设备的运行可靠性水平,以某660MW超临界火电机组一次风机为研究对象,采用多元状态估计技术(MSET)对设备的状态估计与故障预警方法进行了研究。提出了一种基于多重特征参数的动态记忆矩阵构建方法,可在确保MSET算法计算结果精度的同时大幅度减小记忆矩阵的规模。在非线性运算符中引入权重系数构造了改进MSET算法,并应用于对一次风机的异常工况进行仿真,结果表明：改进MSET算法可有效提高异常工况下各参数计算结果精度,通过对各参数估计残差的监测,不仅能够实现故障的提前预警,还可确定故障参数和各参数偏离应达值的程度。

To improve the operating reliability of power plant equipment, multivariate state estimation technique (MSET) is employed to investigate the method of state estimation and fault early warning taking the primary air fan of a supercritical 660MW thermal power unit as research object. A method to construct dynamic memory matrix based on multiple characteristic parameters is proposed, which can reduce the size of memory matrix greatly while ensuring the accuracy of calculated results. A modified MSET is constructed by introducing weighted coefficient to modify the nonlinear operator and is applied to simulate abnormal conditions of primary air fan. The results show that the modified MSET can improve the accuracy of calculated results of each parameter under abnormal conditions effectively. The fault early warning and determination of the fault parameters and the deviation between measured value and target value can be realized through monitoring the estimated residual of each parameter.

TP277

国家重点研发计划资助(2022YFB4100700)