Research On Power Dynamic Prediction Method Of Park-level Microgrid

Microgrid is a small power system close to the user side,which is an integration of power technology,information technology and control technology.Microgrid usually contains distributed power storage device,energy conversion device and load,etc,because its internal components can be self-controlled and self-regulated and automatically switch operation mode between grid and off-grid.Microgrid can realize clean and efficient utilization of new energy and multi-energy complementation.As the important practice of the above technology,microgrid can use energy management technology for real-time collection of load and grid information,free switching of each power unit,optimal power distribution between power storage system and load,finally the park to achieve maximum economic efficiency.To realize the energy management of the microgrid in the park,it is necessary to accurately predict the internal load and photovoltaic power at a long time scale.The prediction results can not only be used as the basis for formulating the subsequent distributed power generation plan in the park,but also as the basis for setting the parameters of the energy storage system.Although there have been abundant research results on load prediction and photovoltaic prediction at home and abroad,due to the particularity of the application scenario of park-level microgrid,the actual effect of its prediction results is different from that of other application scenarios such as large power grid.Therefore,based on the background of park-level microgrid and combined with practical problems encountered in engineering practice,this paper proposes a power dynamic prediction method suitable for park-level microgrid.(1)Considering the fluctuation and randomness of park-level microgrid load,an online dynamic prediction strategy for park-level microgrid load is proposed.Since the load scale of the park-level microgrid is small and mainly affected by the power consumption behavior of the people inside the park,its load has greater volatility and randomness than that of the large-scale power system.The strategy designed in this paper is based on the day-ahead prediction results,and real-time tracking and dynamic correction are carried out within the day.(2)For the park-level PV power generation scenario,a short-term prediction method of PV power based on dynamic update is proposed.Since PV power generation output is mainly affected by meteorological conditions,the sudden change of intraday meteorological conditions may lead to inaccurate prediction of PV power generation and further cause scheduling problems of power supply and energy storage system.So the proposed method scroll updates predicted results using the real-time updated meteorological data and photovoltaic power prediction deviation,and finally improves the accuracy of short-term photovoltaic power generation forecasting.(3)Aiming at the application of power prediction algorithm in practical engineering,the system design scheme is proposed,and the common problems encountered are summarized.Firstly,the data processing process and model update iterative method are introduced.Finally,the model update algorithm including load prediction algorithm photovoltaic prediction algorithm is introduced.The design logic of the whole system of dynamic detection algorithm is explained,and the working process of the system is described.In addition,the data platform interface is displayed,which directly shows the good operation effect of the whole system.