Research On Decentralized-Coordinate Control Of Multiple Power Generation Process

At present, most of electric power is generated by burning of fossil fuels. With the continuing increase of energy demand, fossil energy will be exhausted eventually. Therefore, alternative energy sources are deeply researched by all the countries governments. Wind power, the most mature technology of renewable power generation, has been deeply researched and widely used. There are a lot of successful commercial cases of wind power. It is obvious that voltage stability and system damping of local power system will be impacted by grid-connected wind farms. Decentralized-coordinated controls of conventional power systems are design based on the synchronous generators. Howerver, indcution generator, which belongs to asynchronous generator, is the dominant type used in the wind farms. There are great differences between the asynchronous and synchronous generators, such as electromechanical characteristic and control schematic. It can be seen that conventional decentralized-coordinated control of power system can not satisfy the demand of renewable power system, where the large-scale wind power has been integrated.Based on the above considerations, renewable power system interaction measurement model is developed firstly, where the interactions between the different generators are represented by using local measurable signals. This model provides the basis of decentralized-coordinated control design of renewable power system. State regulator based decentralized-coordinated control of renewable power system is designed and compared with the configurable decentralized-coordinated control. The results of comparative theoretical analysis and simulation study show that decentralized-coordinated control of renewable power system based on interaction measurement method has a better control performance, which is flexible and optimal for the unrestricted controller structure.In order to improve the tracking control and system damping performances of renewable power systems, neural adaptive decentralized-coordinated model predictive control (NADMPC) is proposed, where the interaction measurement models of renewable power system at the chosen typical operating points are used to establish the predictive model bank. An online trained Elman artificial neural networks (ANN) is employed as the weighting controller to calculate the weightings of models in the model bank. The proposed weighting method can be regarded as a closed loop, nonlinear and adaptive weighting method. Then, the weighted interaction measurement model of renewable power system is used by the model predictive contorller for computing the output of the NADMPC. The neural adaptive power system stabilizer (NAPSS) is also proposed to enhance the fault ride-through capability of double fed induction generator (DFIG) based wind farm. Dominant eigenvalue analysis and dynamic simulation show the contributions of the proposed NADMPC and NAPSS.For multi-objective optimization and precise control, the decentralized coordinated mixed H2/H∞ fuzzy proportional integral control (DMFPIC) with regional pole placement is proposed in this thesis. The interaction measurement model of renewable power system is used to obtain the fuzzy state observer, where the approximated error has been considered. With the constrains of regional pole placement and PI controller structure, the solution of the proposed mixed H2/H∞ control problem is transferred to three eigenvalue problems (EVPs), and each EVP can be described by the terms of linear matrix inequalities (LMIs). However, the LMI is a convex optimization, which can be solved easily and conveniently by using MATLAB LMI Toolbox. A robust technique is also proposed to override the effect of approximated error. Eigenvalue analysis and dynamic simulation show the effectiveness of the proposed DMFPIC.