Research On Optimal Scheduling Algorithms Of Microgrid Energy Management System

Renewable energy replaces traditional fuels such as coal,natural gas and oil to obtain electricity has become the current trend of social development.With the continuous development of renewable energy around the world,microgrid technology has been booming because of its ability to realize the use of renewable energy.Microgrid energy management is the key to the efficient use of renewable energy and the economic operation of the system.Therefore,this dissertation focuses on the optimization of island type and interconnected microgrids energy management.The main research contents are as follows.1.Aiming at the problem of inaccurate power prediction in the microgrid,this dissertation proposes a prediction method that can identify and adaptively compensate for photovoltaic power deviation.The algorithm uses the improved k-means method to classify the data.The least angle regression algorithm is adopted to select key feature vectors.The neural network is choosed to capture the characteristics of prediction errors.According to the characteristics of photovoltaic power forecasting deviation,the system uses the combination of historical data and training network to compensate for predictive errors.This method improves the accuracy of renewable power prediction,which in turn improves the accuracy of day ahead energy scheduling.2.The current microgrid real-time energy scheduling optimization algorithm has the problem that it is difficult to achieve a global optimization,this dissertation proposes a real time energy optimization algorithm that can take into account both the accuracy and global optimality of microgrid energy scheduling.The algorithm establishes a power prediction model with time series characteristics and rolling update performance.For the non-smoothness and nonconvexity of the optimization problem,the equivalent transformation and Semidefinite relaxation are used to transform the problem into a smooth convex optimization.Finally,the Averaging Fixed Horizon Control and Semidefinite Programming are employed to solve the problem.The convergence of the algorithm is not affected by the prediction accuracy,and the online global optimal energyscheduling can be achieved.3.In view of the energy coupling existing in the distribution between the main grid and the microgrid,and between the microgrid and microgrid in the interconnected microgrids,the dissertation proposes an online decoupling optimization algorithm.The algorithm uses the alternating direction method of multipliers to decouple the coupling constraints in energy scheduling,and takes the soft threshold operator method to solve the problem of time coupling in the objective function.This algorithm also uses the relaxation factor to achieve the convex transformation of the problem,and finally uses the Averaging Fixed Horizon Control to achieve online energy optimal scheduling.The algorithm is universally applicable to different microgrid structures,and can realize autonomous energy management of interconnected microgrids.