Research On Load Forecasting And Energy Optimization Management Of Micro-grid

The increasing penetration rate of renewable energy,the rapid development of electric power replacement industry,as well as a variety of demand response strategies application,have led to more outstanding power load problems in volatility and randomness,which increases the difficulty of short-term load forecasting in micro-grid.Due to the effect of natural environmental factors such as wind speed and light intensity,the stability of the renewable energy is relatively poorer,which makes the traditional energy management strategies difficult to meet the requirement of safe,reliable and economical operation of micro-grid.Through an in-deep study of power load characteristic,mining the correlation between influence factors and power load and establishing targeted prediction model is an important step to improve the accuracy on short-term load forecasting.Based on electric power data of a micro-grid,this thesis presents a qualitative analysis of load characteristic firstly.This thesis presents a quantitative analysis of various factors through the grey correlation theory,and then extracts main factors.In response to the problem on short-term load forecasting,this thesis puts forward load forecasting method based on information interaction.It includes:(1)the preliminary prediction model,which uses the clustering optimized K-means learning algorithm to solve the initial network parameters firstly,and then adopts Quantum-behaved Particle Swarm Optimization(QPSO)algorithm to search the optimal adaptive network parameters,and utilizes the optimized Radial Basis Function(RBF)module to forecast the electric power preliminarily;(2)the information interaction model,which establishes an interactive mechanism between user terminal and micro-grid to obtain the statistics of users' future electricity information and to employ demand response strategies and evaluation function to adjust and regulate their consumption behavior;(3)the coordinate prediction model,which uses variance optimum combined method based on time-varying weight to adaptively adjust weight coefficient,and gives more weight to higher forecasting accuracy module,so as to improve the precision of short-term load forecasting in micro-grid.In response to the problem on energy optimization management,this thesis puts forward double layered energy management strategy based on day-ahead optimization and real-time scheduling.In the stage of day-ahead optimization,this thesis uses switch optimization rule to determine on-off state of controllable distributed generations,and then coordinate the work periods of shiftable load by means of optimal model based on load control to promote energy configuration between system load and distributed generation;this thesis solves charge-discharge power and state of charge of energy storage system(ESS)by means of optimal model based on energy storage control,meanwhile update working state of shiftable load.In the stage of real-time scheduling,combined with the actual running state and the optimized result,this thesis adjust the current charge-discharge power of ESS by means of scheduling model based on energy storage optimization,which ensures the actual running state of the system is consistent with the optimization result and balances the instantaneous power fluctuations.Besides,this thesis uses the optimal load distribution model to solve the active power output of controllable distributed generations,so as to make independent micro-grid safe,reliable and economical running.