CAJ | 학술논문

为适应微网的建设和发展对其负荷预测效率及精度的要求,针对微网负荷基数小、间歇性、随机性大等特点,提出一种基于果蝇优化算法(fruit fly optimization algorithm,FOA)优化Elman神经网络的微网短期负荷预测模型.考虑到微网负荷受气象因素累计效应的影响,引入人体舒适度指数以降低输入向量维数.为克服常规学习算法收敛速度慢、易陷入局部最优解、编程复杂等缺陷,利用具有全局寻优性能的FOA对Elman神经网络的结构、权值和阈值进行优化,并以国内某微网示范工程项目为例,将FOA_Elman神经网络用于微网短期负荷预测.仿真结果表明,所提出的预测模型优于常规Elman神经网络模型,更具应用价值.
위괄응미망적건설화발전대기부하예측효솔급정도적요구,침대미망부하기수소、간헐성、수궤성대등특점,제출일충기우과승우화산법(fruit fly optimization algorithm,FOA)우화Elman신경망락적미망단기부하예측모형.고필도미망부하수기상인소루계효응적영향,인입인체서괄도지수이강저수입향량유수.위극복상규학습산법수렴속도만、역함입국부최우해、편정복잡등결함,이용구유전국심우성능적FOA대Elman신경망락적결구、권치화역치진행우화,병이국내모미망시범공정항목위례,장FOA_Elman신경망락용우미망단기부하예측.방진결과표명,소제출적예측모형우우상규Elman신경망락모형,경구응용개치.
To meet the requirement of the load forecasting efficiency and accuracy introduced by the construction and development of microgrid, according to the characteristics of microgrid load: small base load, high intermittent and big randomness, etc., a microgrid short-term load forecasting model based on Elman neural network optimized by fruit fly optimization algorithm (FOA) is proposed. Considering that the microgrid load is influenced by meteorological factors accumulative effect, the human body amenity index is introduced to reduce the input vector dimensions. To overcome the defects of conventional learning algorithm such as slow convergence speed, local optimal solution and complex programming, the fruit fly optimization algorithm possessing global optimization performance is utilized to the optimization for the structure, weights and threshold of Elman neural network. And taking a domestic microgrid trial project for example, the FOA_Elman neural network is used for microgrid short-term load forecasting. The simulation results show that the proposed forecasting model provides greater application value and is superior to the conventional Elman neural network model.