基于改进MSET的循环水泵温度监测模型研究

Research on circulating water pump temperature monitoring model based on improved MSET

核电厂循环水泵重要部件的损伤和失效都会直接或间接导致核电机组降功率甚至停机,从而导致计划外的更换和维护成本。利用遗传算法GA(Genetic Algorithm)改进的MSET(Multivariate State Estimation Technique)方法对核电厂循环水泵进行状态估计,然后利用滑动窗口残差统计方法对测点的估计值和实际值的残差进行分析,当残差值曲线超出设定的动态阈值区间时,构建残差统计窗口进行设备状态分析。经验证表明基于MSET方法的核电厂循环水泵预警系统能够实时监测机组循环水泵的运行状态。

Damage and failure of critical components of circulating water pumps in nuclear power plants can directly or indirectly lead to power reduction or even shutdown of nuclear power units, resulting in unplanned replacement and maintenance costs. The MSET method modified by genetic algorithm is used to estimate the status of circulating water pumps in nuclear power plants, and then the residuals between the estimated and actual values at the measurement points are analyzed by using the sliding window residual statistics method, and when the curve of residual values exceeds the set dynamic threshold interval, a residual statistics window is constructed for equipment status analysis. It is verified that the circulating water pump early warning system of nuclear power plant based on MSET method can monitor the operation status of circulating water pump of the unit in real time.