Calculation,Purification And Automatic Recognition Of The Rotating Machinery's Orbit Of Shaft Center-line

As a kind of important graphic signs, the orbit of shaft center-line include a large number of fault information, so the orbit of shaft center-line is an indispensable fault information in the diagnosis process. The effect of purification and automatic identification of the orbit determines the intelligent level of the fault diagnosis expert system. Therefore, a comprehensive study on the orbit of shaft center-line is necessary. In the thesis, based on the rotor-bearing system and the measured orbits, we have a comprehensive study on the orbit of shaft center-line in the following three aspects: theoretical calculation of the orbit, purification and automatic identification.Firstly, according to the bearing's structure, in order to obtain the analytical expression of the oil film pressure, we use the short bearing theory to replace the actual bearing system and obtain the rotor-bearing system's theoretic orbit of shaft center-line by solving the equations of motion under the lubrication theory.Secondly, apply the wavelet theory to the orbit's purification and illustrate that it need to choose an appropriate wavelet and an appropriate layers to get better results. On this basis, introduce the harmonic wavelet to purify the orbit of the shaft center-line because it has a determined formula and the algorithm is relatively simple. The result shows that the harmonic wavelet makes a better effect in the purification.Thirdly, extract the feature of the orbit of the shaft center-line. Based on the image processing theory, converter the orbit into image and take the invariant moments of the image as the shape features of the orbit of the shaft center-line. The simulation experiment prove the invariant moments of the image don't change when the image is rotated or zoomed, so it is feasible to take the invariant moments as the shape features of the orbit of the shaft center-line.Fourthly, identify the orbit of the shaft center-line automatically. Because the neural network has advantages for a highly nonlinear mapping capability, fault tolerant capability and a strong ability to learn, so it is very suitable for the problem of identification and classification. In the thesis, design the BP neural network for automatic identification of orbit of the shaft center-line. The simulation results show the use of neural networks to identify orbit automatically has high accuracy.Finally, verify the above methods with actual orbits acquired from the orbit of shaft center-line rotor-bearing system. The results show the method discussed in the thesis can also achieve better effect when used in the practice.