Research Of State Estimation Based On Kalman Filter Algorithm And Reliability Modeling And Optimization For Wind Turbine

The thesis consists of two parts. On the one hand, due to the importance of the state estimation several state estimation methods based on Kalman filter (KF) algorithm are studied. According to the principle of the basic Kalman filter algorithm, extended Kalman filter (EKF) algorithm, Kalman filter algorithm based on Unscented Transform (UKF), and UKF algorithm with state constraints are studied, and an improved extended Kalman filter-Anti-fading extended Kalman filter algorithm is proposed. Furthermore, the applications of the algorithms above are studied.On the other hand, taking wind power generation as the background, reliability modeling and optimization method of complex system is studied. Wind power generation is one of the most promising new energy, firstly, the calculation method of wind power plant operation reliability is studied, then a reliability model based on Markov processes of wind power generation system is established, and at last a new maintenance strategy optimization method of wind power plant is proposed. The main contribution and the results of the thesis are as follows:1. Several state estimation algorithms based on the KF are studied, including KF algorithm, EKF algorithm, UKF algorithm and constrained UKF algorithm. In order to compare the performance of EKF and UKF algorithm, state estimation of important variables in three-phase induction motor control system is implemented.2. Kalman filter is applied in the motor control system. Based on semi-physical simulation system of wind turbine in the laboratory, modeling and controlling of the generator are performed according to the field collected data. The results show that IMC-PID control method with Kalman filter obtains better control effect.3. According to the problem that the wrong system noise attributes will lead to filter divergence, the thesis proposes a new Anti-fading extended Kalman filter algorithm and shows the effectiveness of the new algorithm through the simulation results.4. In view of excellent performance of UKF algorithm, and features on complex reaction process, highly nonlinear and difficulty to measure the important reactants variables online of the fermentation process, the thesis presents an application innovation--the UKF algorithm is applied in state estimation of saccharomyces cerevisiae fermentation reaction process for the important state variables, simulation results show the effectiveness of this application.5. In the thesis, a horizontal axis double-fed wind generator which is used in many countries around the world nowadays is studied. A reliability model is established and a maintenance strategy optimization method is proposed for the whole wind turbine system. Firstly, a reliability model is established based on Markov processes theory. Then, a Markov processes model of a wind turbine's deterioration, failures and maintenance is built. Taking maintenance interval as the optimization variable, developed model is applied for reliability optimization of the maintenance strategy. And further, Markov decision method is applied to optimize the maintenance strategy based on a determined maintenance interval time of the wind turbine.