Research On Fault Diagnosis For Hydroelectric Generating Units Based On Shaft Orbit

With the continuous development of science and technology, the structure of theHGS(Hydroelectric Generating Sets, HGS) becomes more and more complicated, the closeties between the different parts lead to the increasingly high manufacturing cost, etc., therisk of failure is also gradually grew. It is necessary to sample the state information of theHGS timely and accurately, then identify the working state of it, thus to improve thereliability of the HGS itself.Shaft orbit is an important means for rotating machinery fault diagnosis. It issynthesized by vibration signal, and reflects various fault information of rotatingmachinery, the shape characteristics of shaft orbit is very necessary for diagnosis fault ofrotating machinery.For the difficult problems in the fault diagnosis of HGS, that including shaft orbitpurification, signal detection and extraction, such as Harmonic Wavelet, Walshtransformation, Support Vector Machines, etc, are introduced in this paper, and thesystematic research has been performed, and the main contents are listed as follows:(1) Shaft orbit, as a mapping way of status information, reflects the operation thehydroelectric generating unit from the viewpoint of its graphical features. In this paper, atfirst, the Harmonic Wavelet for the shaft orbit of purification has been used. Second, thegraphic features of the shaft orbit by the Walsh transformation have been extracted. Finally,in order to classify the shaft orbit quickly and efficiently. The SVM (support vectormachine, SVM) has been utilized.(2) Taking into account the requirements of the real-time identification of the orbitand lack of prior knowledge in fault diagnosis, the Walsh spectrum features of the shaftorbit are extracted by calculating and transforming the distance vector, then the mappingrelations between the orbit features and the orbit types are learned by utilizing SVM theory,after that, the SVM is used for the identification of shaft orbits. The results of experimentsshow that the proposed method is feasible.(3) According to the above investigation, the four kinds of typical mechanical faultconditions of the rotating machinery are simulated through the experiment on the DHRMTmufti-functional rotor test-rig. The diagnostic results can verify the correctness of this newfault diagnosis method. Finally, the shaft orbits of HGS have been used to testify thismodel, and the result shows that this model is effective.