Research on Fault Prediction of Secondary Processing Circuit of Temperature Measuring Instrument in Nuclear Power Plant Based on GWO-LSTM

As an indispensable part of nuclear power plant instrumentation and control system, if the circuit has a failure problem, it may lead to the whole instrumentation and control system can not operate normally, which will cause incalculable losses. Nuclear power plant circuit of each module as an indispensable component of electronic equipment, but also faultprone links, the circuit failure to accurately predict the situation has become very important. Based on Multisim software, the simulation circuit diagram of the temperature measurement circuit of the nuclear power plant temperature measurement instrument is constructed, suitable test points are selected and Monte Carlo sampling is performed on the output to obtain the simulation data. And combined with the improvement of Dynamic Time Warping (DTW) distance and Manhattan distance (City Blo ck distance) as the circuit fault indication, and at the same time reflect the data distribution characteristics and morphological fluctuation characteristics, a better measure of the similarity distance of the circuit fault sequence, reflecting the degradation characteristics of the circuit. Finally, Grey Wolf Optimizer (GWO) is used to optimize the parameters of Long Short-Term Memory (LSTM) algorithm to get the optimal parameter combinations of LSTM algorithm, and the fault indication data and optimal parameters of LSTM are read into the LSTM network model for training. After simulation verification, the trained model can better predict the future trend of the circuit fault indication data, and thus can predict the operation of the circuit in the future moments.