| Nuclear power charging pumps are vital equipment in nuclear power plants.In order to ensure the stable operation of the charging pumps,the vibration signals of their operation process need to be monitored,and the signals are analyzed by using signal processing technology to identify their fault characteristics and detect faults.Deal with the initial occurrence.Because the fault signals of most rotating machinery are non-linear and non-stationary,and Hilbert Huang Transform(HHT)has advantages in processing such signals,this method has been widely used in fault diagnosis,speech processing,exponential prediction,and engineering vibration.field.However,the current Hilbert Huang transform still has the disadvantages of endpoint effects and cannot be decomposed normally at both ends of the signal.In view of this problem,this article has been improved,and the effectiveness of the improvement is verified through Matlab simulation and actual failure experiments.The specific work of this paper as follows:(1)This article briefly introduces the research purpose and significance of this topic,analyzes the structural composition of nuclear power charging pumps,discusses the fault diagnosis of centrifugal pumps from domestic and foreign research status,and then introduces a common time-frequency analysis method: short-term Fourier Ye transform,Wegener distribution,wavelet transform and HHT,and compare the signal processing methods based on the characteristics of the signal,determine the method in this paper,analyze several common faults of nuclear power charging pumps,and clarify the gear faults through mechanical models,Frequency domain characteristics of rolling bearing failure and rotor rubbing failure;(2)In view of the endpoint effect during empirical mode decomposition in HHT,the legend is used to explain the causes and consequences caused by the endpoint effect,and then the endpoint extension method based on waveform matching and the minimum norm sum is used to improve the EMD.The continuation method in this paper is compared with several other types of continuation methods,and the validity is verified.(3)In order to analyze the gearbox failure in the nuclear power charge pump unit and the bearing failure in the XJTU-SY bearing data set,the signal was decomposed using the improved HHT algorithm to accurately determine its failure,and then the durability test of the nuclear power charge pump was performed In the test,the online condition monitoringsystem observed the increase of the multiple frequency and the decrease of the frequency in the vertical direction of the non-driving end.The diagnosis was made by the frequency characteristics of the rotor rubbing and the corresponding temperature change rate of the bearing.It indicates that the rotor has caused friction of the upper bearing due to imbalance;(4)A multi-stage pump impeller assembly optimization process based on improved particle swarm is proposed to solve the disorder phenomenon of the multi-stage rotor during the installation process.By optimizing the assembly phase,the rotor can be effectively reduced in low-speed and high-speed dynamic balance experiments.The vibration amplitude contributes to the reduction of the unbalanced failure rate of the nuclear charge pump;(5)Using Matlab GUI to develop the human-machine interface for the program involved in this article,and integrate time-domain analysis,frequency-domain analysis,and on-site dynamic balancing programs,combined with HHT analysis and rotor imbalance optimization in this paper,a nuclear power pump fault diagnosis system. |
PDF Full Text Request