Research On The Method Of Weak Feature Extraction And Incipient Fault Diagnosis For Complex Electromechanical Equipment

The measured signals not only contain the fault information but also contain a lot of noise influenced by background noise, transmission path of vibration and so on in the fault diagnosis of electromechanical. Especially, because the feature of the incipient fault is weak and submerged in heavy noise, it is difficult be extracted. Aiming at electrical and mechanical equipment, this dissertation focuses on the feature of incipient fault detection and practical diagnosis techniques in incipient fault diagnosis.(1) The mathematical model and weak signal detected principles of differential resonator is described. It is discussed that the detection performance influenced by parameter setting. The principles of parameter selection are given. The detection sensitivity and fault-tolerant of differential resonator are analyzed for the first time. The convergence rate and detection sensitivity is enhanced if the system parameters are increased. On the contrary, fault-tolerant of detection is enhanced. Applying this method to the rolling bearing and gear of mill incipient fault, the weak feature of incipient fault was successfully extracted. It is proved to be an effective method of incipient fault features extracted using the differential oscillator.(2) The mechanism and method is first proposed that using the phase of differential resonator reflected amplitude of weak signal. According to characteristics of differential resonator phase diagram, a new method is proposed to identify the state of differential resonator phase. The linear relationship between the phase diagram size and signal amplitude is proved from both theoretical analysis and simulation test with the same parameters of differential resonator. Therefore we can use differential resonator phase diagram size to represent the amplitude of signal. Incipient bearing fault data of mill are analyzed using the differential resonator. The deteriorated process of bearing is founded. And it shows the methods'efficiency.(3) It is analyzed for the first time that band width of detected with differential resonator is influenced by system parameters. A new method is prospered to detect the complex frequency signal using differential resonator array. The signal frequency spectrum is reconstructed without noise. The small-scale system parameter correspond to the larger detection bandwidth, the large-scale system parameter correspond to narrow bandwidth. The complex signal frequency spectrum is reappeared according signal frequency and amplitude information that it reflected by phase diagram of differential resonator.(4) A new method is proposed to accurate detect weak signal that differential resonator and stochastic resonance are fused. There is odd multiple frequency in the stochastic resonance detection. The spectral peaks of stochastic resonance are detected by difference resonator. False frequencies are excluded using the frequency selection of differential resonator. The incipient fault data is analyzed. And the feature of fault is extracted. The incipient fault is expected to detect using this method.(5) There is subjective that the state of chaotic oscillator is detected using eye-measurement. The phase diagram of chaotic oscillator is described by Hu moment from convergence and symmetry when the chaotic oscillator transformed. The threshold value of chaotic oscillator detection is obtained. The state of chaotic oscillator can be automatic identified. It is proved to be an effective method through analysis the data of incipient bearing pitting fault of high-speed rolling mill.Satisfactory results have been achieved when using above method to extract the feature from the weak signal of complex electromechanical. It have widespread application prospect in the domain of incipient fault diagnosis.