Review of nuclear power plant control research: Neural network-based methods

Abstract

The control systems of nuclear power plants (NPPs) implement the controls of a nuclear reactor and its power system, equipment, process, and parameters. The performance of the control system directly affects the safety, reliability, and economy of NPPs. To improve the performance of the control systems of NPPs, the application of new control theories, technologies, and methods to control NPPs have been studied and new control methods have been explored extensively. With the development of intelligent control theory, artificial neural network (ANN; usually referred to as neural network, NN) control was considered in the field of nuclear power. Additionally, the application of ANN in NPPs is also being developed. In this study, we review and analyze the progress and characteristics of NN control (NNC) applied research in NPPs. This work summarizes and classifies the NNC methods (NNCMs) of NPPs, which have been studied and proposed, including NNPID control, adaptive control, self-tuning control, adaptive critic-based control, predictive control, compound control, supervisory control, and direct control method. The research progress of NNCMs for NPPs show that the research on NNC of NPPs advances with the development of NN theory, NNC theory, automatic control theory, and other related theories and methods. NNC is affected by the development of NN theory, and its application in NPPs is still in the stage of theoretical exploration and simulation research. It still needs further research and verification for the practical application of NNCMs in NPPs.