Study On Condition Monitoring And Fault Diagnosis For Gearbox Based On OMA

Gearbox is an indispensable mechanical transmission device, which widely used in metallurgy, mine, petroleum, power generation, engineering machine and other industries. Health monitoring and fault diagnosis of the working gearbox, can predict the health status of gearbox to prevent equipment damage occurs suddenly, causing casualties, and enormous losses. In this paper, the OMA(operational modal analysis)technique will be applied to running gearbox in the actual conditions for data collection and real-time analysis. After identifying the operational modal parameters of working gearbox, then compare with the normal data, and further achieve condition monitoring and fault diagnosis.In this paper, first of all, taking into account different working state of the gearbox, gearbox test-bed is built and the corresponding experimental plan is established for the study of OMA, then conduct test point arrangement and sensor installation. Secondly, to OMA to gearbox with Op.PolyMAX method and the improved multi-reference least square complex exponential method respectively, then identify the modal frequencies and modal shape based on stabilization program within the considered frequency range. Op.PolyMAX method has an excellent effect in identifying the poles of the system, which essentially uses the maximum likelihood estimation principle which is a kind of modern parameter estimation method, and uses multiple reference cross-power spectrum matrixes instead of frequency response function matrix. In the process of improving multi-reference least squares complex exponential algorithm frequencies of periodic excitation signal will be considered as the known characteristics root (zero damping), which are explicitly included in the identification process, and this method plays a good effect. Thirdly, since the gear transmission can cause the vibration including harmonic components, this is contrary to white noise input assumption of OMA, the identification method may lose robustness, leading to the modal parameters identified are not accurate, traditional harmonic frequency analysis methods combining time domain and frequency domain is utilized for the harmonic components of the response signal to achieve the extraction of the fault. Fourthly, in order to solve the fault location problem, the concept of GODS(generalized operating deflection shape) is posed and corresponding matlab program is self-compiled based on transmissibility function to identify GODS of gearbox. After representing GODS curves of all measuring point with two-dimensional graph, we can find the fault type and location through differences of the peaks of vibration response.The research projects, breaking the traditional testing and diagnostic methods, provides a new way for condition monitoring and fault diagnosis of equipment on multi-operating conditions.