Research On Fault Diagnosis Method Of Electric Valve In Nuclear Power Plant

Valves are one of the common equipment in nuclear power plants.Due to the nature of the valve that needs to be switched frequently,it is easy to wear and age,which has a certain impact on the occurrence of system operation accidents.Ensuring the normal operation of electric valves is the key to ensuring the safe operation of nuclear power plants.Electric gate valve is studied in this article,and data is collected in the four operating states of the valve.Key technologies such as signal processing,fault diagnosis,and fault diagnosis system design are being studied.A fault diagnosis method suitable for electric valves is proposed.The specific work content is as follows:First,the acceleration sensor is used to collect the vibration signals of the electric gate valve in different states.The wavelet packet decomposition method is used for signal processing,and the ratio of the wavelet packet node energy to the total energy is calculated.This data is used as the sample data of the fault diagnosis algorithm.Second,combine the Particle Swarm Optimization(PSO)with the Kernel Extreme Learning Machine(KELM),and the PSO is used to optimize the parameters of the KELM.The problem that the KELM introduced the kernel parameter which caused the sensitive problem of parameter setting is solved.Improved PSO algorithm is used to solve the problem that PSO algorithm is easy to fall into local extreme value.Select the vibration signal of the three operating states of the electric valve as the sample data of the model,and train the ELM,KELM,PSO-KELM,and IPSO-KELM model to realize the recognition of different modes.Third,acoustic emission sensors are used to collect internal leakage fault data under different flow rates of electric valves.Convolutional Neural Network(CNN)algorithm is used to classify electric gate valves with different degrees of internal leakage.Fourth,python 3.6 is used to complete the development of the electric gate valve fault diagnosis system software under the Spyder compilation environment.Relevant functions such as fault diagnosis of electric gate valve,model training of intelligent algorithms,and loading of operating data are realized.The effectiveness of the system is verified by the collected experimental data.Comprehensive experimental results show that PSO-KELM、IPSO-KELM can accurately distinguish the different operating states of electric valves,and CNN can effectively identify different degrees of internal leakage of electric valves.The functions of each part of the electric gate valve fault diagnosis system are effective and can achieve the expected goals.