Research On Reactor Control Rod Movement Detection System

The control rod plays a vital role in the normal operation and safety of the pressurized water reactor nuclear power plant,so the normal operation of the control rod in both cases should be guaranteed.When in the protecting action,ensure that the control rod drive mechanism(CRDM)loses power and the latch release control rod falls into the reactor by gravity.The reactor is protected by a drop in the control rod that prevents the reactor from reacting.When in the normal operation,it is necessary to ensure that the claws transmit magnetic poles during normal timing,preventing the reactor from shutting down or reducing power due to malfunction.Therefore,research on the detection system of the control rod action is very necessary.This paper focuses on the research of the detection of reactor control rod's movement.The main work is summarized as follows:Firstly,the architecture of the control rod drive mechanism is introduced.Then the process of controlling the current signal of the rod is analyzed.The change of the current in the rising process and the completion of the action are analyzed.Then,according to the basic structure of the control rod action system and the current change timing in the action process,the overall structure analysis of the control rod movement detection system is carried out,which includes the software and hardware data acquisition board.Finally,the realization principle and index of hardware signal data acquisition board are analyzed.Secondly,the basic circuit model of the control rod is studied.According to the circuit model and the current formula,the rising edge of the current during the action of the control rod is analyzed.Then the movement detection based on wavelet decomposition is studied in MATLAB.The real-time detection of the action of the control rod is realized by using the wavelet decomposition singular value and the signal stratification energy respectively,and the optimal calculation method of the control rod movement detection is obtained and the state machine is designed according to the order of the control rods.Then,using SVM,the detection of the control rod movement in machine learning is analyzed,The energy value,the kurtosis factor and the peak factor of each layer of the signal after wavelet decomposition are used as the input data for classification and identification,and then the particle swarm optimization is used in the learning process of SVM.The algorithm performs parameter optimization of SVM machine learning,and finally analyzes the results of judgment under different SNR.It is found that the three eigenvalues have higher recognition rate when the noise is relatively low.When higher,the energy value has a relatively high accuracy.Finally,the host computer of the control stick movement detection system is designed under the Ubuntu system.Through the software requirements analysis software,the software interface,database design,UDP packet receiving module and data binary file storage are designed in the software design process.The wavelet multi-layer decomposition algorithm is implemented in C++.The implementation method flow of each function in the demand analysis is designed.It can display the action of the control rod in time in the interface,which is convenient for the operator to view in real time.