Converters Control Related Research Of Doubly-fed Induction Machine Wind Power Generation Systems

Along with the development of new and renewable energy industry, doubly-fed induction machine wind power generation systems are widely used in large wind power generation occasions. Research of the principle and existing problems of doubly-fed induction machine wind power generation system is really meaningful to renewable energy application.Both the stator and rotor of (DFIM) exchange active and reactive power with the sources connected to them, so the power relationship in the energy conversion is complicate. This dissertation analyses rotor's algebraic sign and reactive power, and give the necessity of minus frequency in super-synchronous speed area. The physical meaning of minus frequency is illustrated. When reactive power is computed by equivalent circuit,"reactive power character change" phenomenon occurs. The equivalent circuit of DFIM is divided according related physical element. The mechanical shaft related element is defined, so its reactive power. Then the reactive power flow and conservation character in energy conversion is illustrated. Instead of complicated equation, four-dimensional figures are used to express DFIM working characteristics in three-dimensional area.The energy conversion of DFIM is controlled by back-to-back converters between its rotor and the grid. This dissertation builds mathematical models of back-to-back converters. Grid side converter's grid voltage oriented vector control and machine rotor side converter's staton flux oriented control are employed and the decoupled control of back-to-back converter's active and reactive power is reached. The mathematical models of DFIM before and after grid connection are researched. The couple items and parameters of DFIM before and after grid connection are found different. The drawbacks of ignoring the difference of DFIM before and after grid connection in torque dynamic performance design and implement are illustrated. The grid connection scheme with voltage close loop is built and analyzed from the point of difference before and after grid connection. Then the corresponding experiment is done and the results are analyzed in comparison.In stator flux oriented vector control DFIM wind power generation system, the accuracy of complex power control is decided by stator flux detection performance. During low grid voltage period dc composition and negative sequence occurs in stator flux and magnetic circuit non-linear character occurs too. This puts more slashing requirements on stator flux detection algorithm. This dissertation analyzed the direct computation methods and close loop method of stator flux detection. The dynamic performance of different methods during low voltage ride through is emphasized. The system LVRT ability with different flux detection methods is researched in principle and experiments. The research provides groundwork in stator flux detection method choosing during LVRT.Rotor position is necessary in vector control of DFIM. This dissertation researches rotor position detection methods in both position sensor existing and position sensor non-existing conditions. In position sensor existing condition, a position sensor mounting detection method in synchronous reference frame is proposed. This method avoids phase judge of stator voltage before grid connection, which has a lot of harmonics. So the computation workload in position sensor mounting detection is reduced largely. In position sensor non-existing condition, the control algorithm has to compute rotor position firstly to implement vector control scheme. This dissertation researches open loop and close loop algorithm of rotor position detection. The model reference adaptive system based on stator flux and rotor current are analyzed in detail. Their equivalent close loop structure shows the sameness of the two algorithms. The input of adaptive mechanism in rotor current based MRAS is improved. As a result, the adaptive mechanism input has a linear relationship with rotor position detection error. The parameters design method is deduced for PI adaptive mechanism. The dynamic performance of system with different parameters is researched in experiment.