Model Parameter Identification And Fault Diagnosis Of The Hydro Generator Unit

With the rapid development of the hydroelectric of China, the hydro generator unit (HGU) move towards the trend of large capacity and high head. The safe and stable operation of HGU makes a significant effect to improve the economy and guarantee the safety of the hydropower station. The HGU is a complex nonlinear system which is influenced by hydraulic, electrical and mechanical, and the operational mechanism of HGU is not all clearly to human people now. So, the model building and fault diagnosis are always the hot research topics. Identification molding of HGU which can improve the control quality and makes a contribution to fault diagnosis has attracted many researchers.In this paper, the detail model of HGU have been studied, and a detail hydro turbine model is proposed. In order to reduce the computing effort of the electrical power system, a dynamic equivalence model of HGU group is proposed. As the modeling method has a significant effect in fault diagnosis of HGU, the Volterra series, GFRF and NOFRFs have been employed in the fault diagnosis of HGU. The research content and results of this paper are as follows:(1) As the hydraulic loss of distributor has a great influence on efficiency of the hydro-turbine, and the hydraulic loss model is complex. A method base on curve fitting is proposed to expression the hydraulic loss of distributor, and the detailed hydro-turbine model is built. An IAFSA method is proposed to identify the model. An experimental study is carried out to test the proposed method, the simulation results indicate that the improved hydro-turbine model has high accuracy and can describe the real running state. The identification result of the distributor’s hydraulic loss is according to the theoretical formula analysis, and the model was validated by the step disturbance method. The results indicate that the proposed model is reliable and accurate.(2) As the traditional dynamic equivalents based on online identification have the problem of multi-solution. An improved identification method has been proposed. The proposed objective function not only used the dynamic data dynamic data of the tie line, but also the dynamic data of some point in the study system. This method ensures that the dynamic process of the equivalent system is similar to the original system. As the condition conversion of the hydropower system, there is an urgent need to improve the identification performance. An improve PSO method is also proposed. The result indicates that the proposed method overcomes the problem of multi-solution, and has higher efficiency in identification.(3) As the identification of the Volterra series involves the infinite-solution problems which are caused by the periodic characteristic of the excitation signal of rotor-bearing system. A key kernels-PSO (K.K-PSO) method is proposed for Volterra series identification. Instead of identifying the Volterra series directly, the key kernels of Volterra are found out to simply the Volterra model firstly. Then, the Volterra series with the simplest formation is identified by the PSO method. An experimental test has been done to get the Volterra series of a rotor-bearing test rig in different states, and a fault diagnosis system is built with a neural network to classify different fault conditions by the kernels of the Volterra series. The analysis results indicate that the KK-PSO method performs good capability on the identification of Volterra series of rotor-bearing system, and the proposed method can further improve the accuracy of fault diagnosis.(4) In order to recognize runner state of hydraulic turbine timely and effective. Modeling method was used to build the runner model, and effective feature was extracted for state recognition. As the input signal of hydraulic turbine runner is complex and difficult to be measured, the input unbalance force which acts on the runner is assumed as a random processing, and a state recognition method based on blind identification of Volterra series is proposed. The Volterra model of hydraulic turbine runner is built by blind identification method with the monitoring data of guide bearing. Generalized frequency response function (GFRF) is built by multi-Fourier transform. Runner state was recognized by analyzing the diversification in the frequency domain. The experimental research indicates that the proposed method describes the frequency domain characteristics very well, and the extracted frequency feature is effective.(5) As the fault diagnosis for hydroelectric generator unit is lacking of fault data, and most of the diagnosis methods are proposed without considering the characteristic of hydroelectric generator unit. A vibration dynamic modeling method for hydroelectric generator unit is proposed by using the finite element method, and further get the vibration data in different states. Next, the fault feature is extracted based on the nonlinear output frequency response functions (NOFRFs). At last, a diagnosis system with Support Vector Machine (SVM) is proposed and employed for diagnosis of the hydroelectric generator unit. The results indicate that the feature which extracted from NOFRFs has a strong effect, and demonstrates that the proposed method is feasible and helpful for faults diagnosis in hydroelectric generator unit.