Integrated Dispatch And Control Of Micro Grid Based On Extreme Dynamic Programming Algorithm

Under the background of increasingly mature control technology and energy management technology,MG has gradually become the inevitable trend of distributed generation network structure.As a solution to integrate distributed power supply,MG has effectively integrated a large number of different distributed generation into the distribution network in the form of integration.The distributed generation has quite smaller inertial than the conventional DC generators,and the renewable energy source has great intermittency and uncertainty,which will lead to the unstable voltage frequency of the micro grid.In the meantime,due to the diversity of distributed energy sources,the power quality control of MG becomes more and more serious.In the traditional mode,the control strategy of MG generally adopts the hierarchical control strategy,forming a three-layer control combination strategy of "droop control + automatic generation control + economic dispatch".However,the hierarchical control strategy has some problems.The control algorithm aims at reducing frequency deviation,while the optimization algorithm aims at reducing economic cost.The result of the combination of this two is that the effect of the control algorithm is emphasized while the economic benefit is ignored.Moreover,the deviation of the prediction of generation capacity and load situation may lead to the imbalance of power generation plan and load,thus resulting in the diseconomy of distributed generation after the droop control and automatic generation control.In view of the above problems,this paper designs a unified time scale intelligent algorithm,which can output power generation instructions of multiple micro power generation at one time to take into account frequency deviation and power generation cost.The algorithm,shown in Fig.2,carries the extreme dynamic programming algorithm that is based on adaptive dynamic programming algorithm.The proposed algorithm takes the adaptive dynamic programming algorithm as the framework,the extreme learning machine(ELM)as the kernels of the evaluation module,the model module,the implementation module and a new prediction module,which is named: extreme dynamic programming(EDP).The proposed algorithm,due to the addition of extreme learning machine,has the following advantages:1)the required pre-learning time is greatly reduced compared with the original ADP algorithm;2)compared with existing GCD algorithms,the actual calculation time required in usage is shortened,avoiding the problem that the actual calculation time exceeds the allowedinstruction cycle;3)the problem of poor performance caused by the combination of control algorithm and optimization algorithm can be avoided,and the generation power instruction of each unit can be output at one shot.