Study on Long-period Time Series Data Prediction Method for Qins-han Nuclear Power Accident Management
This article introduced a deep neural network method to predict the status of a nuclear power plant un-der accidental conditions,which is a proactive analytical method for the NPP safety analysis and accident man-agement.It was motivated by the data-driven method in deep learning,adept at discovering patterns from data and making inferential predictions.Through the method,we estimated the statistic and long-term connections between inputs and outputs,which was able to forecast the progression of nuclear accident with limited histori-cal signals.Finally,the method validations and applications on Fangjiashan Units 1 and 2 were discussed,and the results proved that the method was quite efficient for predicting the operating states of the NPP under acci-dental conditions,and assisting the operators on the accident management.
neural networksevere accidenttime series prediction
NETL
NSTL
万方数据
