Application of CNN-LSTM in Predicting the Status of Control Drive Plug-Ins

In the third generation nuclear power system of China, the control drive plug-in is the key component to realize the operation of the control rod. The status prediction of the control drive plug-in can improve the operation safety of the nuclear reactor. Based on convolutional neural network (CNN) and long short term memory (LSTM) network, the status prediction model of control drive plug-in is established, which consists of 2 convolution layers, 2 pooling layers, 2 LSTM layers and 1 fully connected layer. After data processing, the main input and output analog quantity of control drive plug-ins collected under different working conditions are divided into training set data and test set data, which are used as input data of training model and test model respectively to conduct CNN-LSTM model training and test. The experimental results show that the CNN-LSTM model not only has higher accuracy, but also can better deal with long sequence dependence problems compared with the LSTM model, enhance the expression ability of the model, and have strong advantages in feature extraction and processing of long sequence dependence problems, and can better solve some complex problems.