| To extract the weak fault signal features, a method research of extracting the fault feature based on the Local Mean Decomposition and 1.5-Dimension Spectrum was carried out. A Local Mean Decomposition method based on all phase pretreatment was proposed, and a pattern recognition method based on manifold learning was studied, to achieve a bearing fault intelligent diagnosis realized the intelligent diagnosis for bearing fault based on the Manifold Learning. To achieve the dynamic balance of the aero-engine rotor system, a research of dynamic balance technology based on the acceleration signal was carried out. A balance method of the Influence Coefficient Method based on the acceleration signal was proposed.The main contents were as follows:(1) A feature extraction method based on the Local Mean Decomposition was studied. An endpoint processing method based on the inside shrink continuation for the Local Mean Decomposition was proposed, and the reliability of the continuation method was analyzed. Combined with the all phase pretreatment, a Local Mean Decomposition method based on all phase pretreatment was proposed, namely carrying out the all phase pretreatment method before the Local Mean Decomposition. Taking rolling bearings as an object and using the vibration signal and the acoustic emission signal, the different types of single fault and compound faults were analyzed, which verified the effectiveness of the fault feature extraction method of the Local Mean Decomposition based on all phase pretreatment.(2) A feature extraction method of 1.5 Dimensional Spectrum based on the pretreatment of the Local Mean Decomposition was proposed, which combined the Local Mean Decomposition based on all phase pretreatment with 1.5 Dimensional Spectrum. The method was applied to fault diagnosis of low-speed bearing, and its application on different types of single fault and compound faults signal was analyzed, and the relationship between the enhanced frequency of 1.5-Dimension Spectrum for low-speed bearing signal and the corresponding harmonic order was discussed. Moreover, an intelligent recognition method based on the Manifold Learning was studied, which combined the phase space reconstruction with the Supervised Locally Linear Embedding algorithm and realized the intelligent diagnosis for bearing fault.(3) The dynamic balance technology based on the Influence Coefficient was studied, and the experimental verifications of both the single plane and the double plane were carried out. The conversion relationship between the influence coefficient based on the quality and the one based on the mechanics was deduced, and was used to determine the mechanical form of the unbalanced load. Finally combined with the phase detection of the acceleration signal using the adaptive notch and Cross Power Spectrum method, the Influence Coefficient based on the acceleration signal was defined, which was applied to the dynamic balance based on the acceleration signal. And the experimental study of the Influence Coefficient method based on the acceleration signal was carried out, the effectiveness of which was verified. |
