Research On Identification And Vibration Of Hydro-Turbine Generator Set Shaft System

The hydro-turbine generator set is the key part of a hydropower plant because it finishes most work of transferring energy. The vibration of the shaft system of the hydro-turbine generator set has direct influence on the safety of the set operation and decides the whole property of the plant. Therefore, the research of vibration and blind identification of the shaft system is of important value for engineering practice. This dissertation originates from the following three scientific research projects which are sponsored by the National Natural Science Foundation of China (key project No.5083 9003, project No.5127 9071), the Doctoral Foundation of Ministry of Education of China (Grant No.2013 531413 0002). The shaft system of the hydro-turbine generator set is taken as the research object and the vibration and blind identification problems are studied, so as to supply the theoretical basis and datum reference for the shaft vibration analysis. The main work can be summarized below:(1) The influences of different factors on the vibration characteristics of shaft system of hydro-turbine generator set are analyzed by the finite element method. Two kinds of higher order elements are proposed,The nonlinear model of the shaft system of the hydro-turbine generator set is established by the rotordynamics, in which the gyroscopic effect, the shear effect and its own characteristics are considered. Based on finite element method (FEM), Two kinds of higher order elements are proposed:a three-node elastic shaft segment element NODE3 suitable for calculating the vibration of the shaft system and a three-node shaft segment element NODE3shear considering the shear effect. And the dynamic equations of the elements and the whole shaft system are obtained. The two proposed elements are validated by numerical examples and then used in the calcu-lation of shaft system in a hydro-turbine generator set. The influences of gyroscopic effect, the different rotation speed and stiffness on the whirl frequencies, critical speed and calculation accuracies are obtained. The results show that, for the shaft system of the hydro-turbine generator set, the shear effect has a certain influence on both the natural frequencies and the critical rotation speed:when the stiffness of the shaft is large enough, the shear effect must be considered, otherwise there would be huge errors.(2) A hybrid genetic algorithm (HGA) and a blind identification method are proposed for identifying the time domain kernels parameters in a Volterra system. A blind identification based on the third order cumulants is realized for the shaft system of a hydro-turbine generator set.To solve the problem that the model of the shaft system of the hydro-turbine generator set is hard to be established, the quadratic Volterra system is introduced to describe the shaft system. The water force on the hydro-turbine runner is taken as the input signals of the system while the lateral swing at the water guide bearing as output signals, and then the Volterra system of the shaft system is established. To identify this model, an HGA is proposed by combining the simple genetic algorithm (SGA) with the traditional direct search method, and the corresponding program is developed. Three different numerical examples show that the HGA has the higher efficiency and stability than the SGA and the direct search method.Due to the difficulties that the input signals of the hydraulic force can hardly be measured, a kind of third order cumulants (TOC)-based blind identification method is introduced. There are plenty of coefficients that are needed to be determined when the TOC are calculated, a computer loop judgment method is thus proposed and the corresponding program is developed. Three different numerical experiments are carried out including the Hammerstein model and the Volterra model, and the effects of the different signals to noise ratio (SNR), memory length, sampling length are studied, which show high efficiency and strong universality of this blind identification method. Finally, taking different operating conditions into consideration, the time domain kernel parameters are blindly identified in the shaft system of the hydro-turbine generator set and the kernel parameters are obtained.Since the hydraulic force is always hard to describe, we propose a backwards recursive method to approximate the input signals. The feasibility of the recursive method is tested by numerical experiments. It is verified that the approximate input signals from this method can be treated as random excitation acting on the runner of the shaft system, and used as the hydraulic excitation for the shaft system analysis.(3) The dynamic response of the hydro-turbine generator set shaft system is analyzed under complex actions.The calculation model of the shaft system of the hydro-turbine generator set is established, and the random hydraulic force, the nonlinear oil film force, the unbalanced mechanical force and the nonlinear unbalanced magnetic pull are considered. The dynamic equations of the shaft system are derived. With the calculation results, the influence on the shaft system of some hydraulic, magnetic and mechanical parameters, including the rotation speed, the mass eccentricity, the excitation current and the random hydraulic excitation, are analyzed. The results reveal that, the random excitation (zero mean, independent identically distributed signals) on the runner has hardly influences on the up guide bearings and the generator rotor; has certain influence on the down guide bearing; has great influence on water guide bearing and runner. And the effects become stronger as the random excitation increases.(4) The prototype observations are carried out for the shaft system of 4# hydro-turbine generator set in a certain hydropower plant and the observed signal data are analyzed and processed.The shaft system of a hydro-turbine generator set is briefly introduced. The observed data are analyzed and filtered, and their graphs are plotted, so as to blindly identify the shaft system conveniently. Furthermore, the identified output data are compared with those in real hydro-turbine generator set, so as to validate the accuracy and the feasibility of the identification, and the reliability of the backwards recursive method to get the hydraulic excitation.