多元线性回归分析在换流站内冷水漏水检测中的应用

Application of Multiple Linear Regression Analysis in Converter Station Leak Detection of Internal Cooling Water

为快速检测换流站内冷水是否存在漏水情况,确保设备以及电网安全,在确定与高位水箱液位相关的数个因素后,基于逐步回归分析法可找出高位水箱液位与数个因素中显著因素的数学表达式。以牛寨换流站牛从乙直流极2为例,利用均匀设计软件V3版本求出高位水箱水位与进阀温度及直流负荷之间的数学表达式,并对表达式的可信度进行了校验。当表达式中的自变量发生变化时可由表达式预测高位水箱液位,若实际值与预测值之差超过设定值时,则极控系统可认为内冷水存在漏水现象,应发出告警信号。此外,该保护可与当前VHA漏水检测构成两种不同原理的保护。并为今后高压直流系统漏水检测方法提供参考。

In order to rapid detect whether there is leakage situation in internal cooling water of converter station, and to ensure equipment and the safety of the power grid. After determined some factors associated with the high tank. This massage finds out the mathematical expressions of high tank level and significant factors in several factors based on stepwise regression method. Take the pole 2 of Niu-Cong-Yi in Niu Zhai converter station for example. The mathematical expression between the high water tank level and the temperature of the inlet valve and the DC load was obtained by using uniform design version v3, and checked the credibility of the expression. It can predict the high water tank level by the expression when the independent variable changes in the expression. If there is difference between the actual value and the predicted value exceeds the set value. The pole control system should judge there exist leakage phenomenon in internal cooling water and should issue alarm signals. Besides, this protection can constitute two different protection principles with the current VHA Leakage detection. And it can provide reference for the future water leakage detection method in HVDC system.