The Automatic Generation Control Of Isolated Two-area Microgrid With Multiple Resources Based On Learning Optimization

The isolated microgrid has the advantages of less pollution,high reliability,and high energy utilization efficiency.However,new energies such as photovoltaic and wind power are intermittent,random,and volatile,with the random changes of load demand,which could make the active power of the island microgrid out of balance,as well as the frequency greatly deviate from the rated value;eventually,the standalone microgrid will be beyond the range of safe operation.Therefore,to stabilize the frequency of microgrid while maintaining the voltage of it stable,it is necessary to control frequency of isolated microgrid.An isolated two-area microgrid system consisting of micro-turbine units,small-scale hydropower units,flywheel energy storage systems,wind turbines,and photovoltaic power is considered in this dissertation.The automatic generation control(AGC)for isolated two-area microgrid with heterogeneous energies is addressed for improving the performance of frequency control.A Load Frequency Control(LFC)model of isolated two-area microgrid with multiple distributed energy resources and energy storage device is established.The wind power,the photovoltaic output and load demands are formulated as discrete time Markov processes based on their stochastic dynamic characteristics.For the automatic generation control of an isolated two-area microgrid,the frequency controller is designed by the learning optimization algorithm combining the AGC control method of the interconnected power grid.The simulation experiment was performed on the LFC model of the microgrid system.Compared with the traditional Q learning and PI controller,the results verify the rationality of the Q learning optimization algorithm based on simulated annealing applied to the two-area microgrid model and the efficient performance of the frequency controller.In addition,considering the situation of insufficient capacity in the microgrid system,the power supply and the load demand cooperate with each other.Aggregate and classify the load demand,divide it into rigid load and flexible load,equalizing the flexible load to virtual negative power supplies to coordinate with distributed power.A discrete-time Markov decision process model for a microgrid AGC system is established,regarding the flexible load as the system state quantity.The frequency controller based on a learning optimization method is used to complete the power balance and frequency of the isolating two-region microgrid system.Finally,simulation results and analysis are presented.