Small modular reactor (SMR) is a complex, time-varying

and strongly coupled nonlinear system, which adopts integrated

design with short pipeline and small coolant inventory. sensor

failure of SMR under frequent load-following conditions

threatens the safety of reactor. It is an effective way to realize

sensor fault tolerance by abnormal sensing value

reconstruction with machine learning method. In this paper, a

sensing value prediction model combining hybrid dilated

convolution neural network and long-term memory (LSTM)

neural network is proposed. The dilated convolution neural

network is used to extract the feature information in the data

set as the input of LSTM. LSTM is trained through training set

to obtain the predicted sensor parameter time series. The

training set is obtained from the SMR simulation model

established by RELAP5 code. Multiple sensing values are input

to each other's prediction model, in order to avoid the one

abnormal sensing value affecting the prediction behavior of

other sensing value prediction models, the abnormal sensed

value is replaced by its predicted value as the input of other

models. The test set completely different from the training set is

used to verify the behavior of the prediction model. The results

show that the prediction model proposed in this paper fits the

real data well, which has good prediction behavior on sensing

value reconstruction.