基于NARX神经网络的核电汽轮机超速保护系统可靠性实时预测

Real-time Predicting Reliability of Nuclear Turbine Overspeed Protection System Based on NARX Neural Network

汽轮机超速保护系统可靠性预测是保障机组安全运行的重要方法之一。针对静态可靠性分析方法不适于系统有缓慢渐变故障的可靠性预测,提出一种GO法和NARX神经网络相结合的超速保护系统可靠性实时预测方法。该方法采用GO法计算系统可靠度,然后结合系统可靠度和状态参数,利用NARX神经网络的时间关联性,进行汽轮机超速保护系统可靠度实时预测。以某核电机组汽轮机超速保护系统的一些设备渐变故障引起油压状态变化为例,通过仿真验证了该方法能实时有效预测系统可靠性。

The reliability prediction of turbine overspeed protection system is one of the important methods to ensure the safe operation of the unit.Since the static reliability analysis method is not suitable for the reliability prediction of the system with slow gradual failure,the paper proposes a real-time reliability prediction method of the overspeed protection system based on the combination of GO method and NARX neural network.GO method is used to calculate the reliability of the system,and NARX neural network is used in combination with the system reliability and state parameters to predict the reliability of the steam turbine overspeed protection system in real time.This paper conducted simulation results by taking the state change of oil pressure caused by the gradual failure of some equipment in the turbine overspeed protection system of a nuclear power unit as an example.The simulation results show that this method can effectively predict the reliability of the system in real time.