考虑概率区间的微电网短期负荷多目标预测方法
发布时间：2018-12-20 15:15
【摘要】：微电网负荷随机性强、波动大,负荷单点预测已经难以满足微电网稳定运行需要.提出一种考虑概率区间的微电网短期负荷多目标预测方法,以循环神经网络为预测模型,以逼近理想解排序策略、网格筛选策略对基本多目标人工蜂群算法进行改进,优化循环神经网络的权值和阈值,避免单目标区间预测中惩罚系数难以选择的问题,对历史负荷数据进行记忆并修正预测结果,有效提高微电网短期负荷区间预测准确性与可靠性.仿真结果表明,本文所构建的考虑概率区间的微电网短期负荷多目标预测方法,预测性能优越、结果准确,可为微电网安全经济调度提供决策依据.
[Abstract]:The load of microgrid has strong randomness and large fluctuation, so it is difficult to meet the need of stable operation of micro-grid by single-point load forecasting. This paper presents a multi-objective forecasting method for short-term load of microgrid considering probabilistic interval. Cyclic neural network is used as the prediction model, and the basic multi-objective artificial bee colony algorithm is improved by approximate ideal solution ranking strategy and grid screening strategy. The weights and thresholds of the cyclic neural network are optimized to avoid the problem that the penalty coefficient is difficult to select in the single-objective interval prediction. The historical load data are memorized and the prediction results are corrected. The accuracy and reliability of short-term load interval forecasting for microgrid are improved effectively. The simulation results show that the multi-objective forecasting method for microgrid short-term load based on probabilistic interval is superior in performance and accurate and can provide decision basis for safe and economic dispatch of microgrid.
【作者单位】： 江南大学物联网技术应用教育部工程研究中心;
【基金】：国家自然科学基金(No.61579167,No.61572237) 高等学校博士学科点专项科研基金(No.20130093110011)
【分类号】：TP18;TM715
[Abstract]:The load of microgrid has strong randomness and large fluctuation, so it is difficult to meet the need of stable operation of micro-grid by single-point load forecasting. This paper presents a multi-objective forecasting method for short-term load of microgrid considering probabilistic interval. Cyclic neural network is used as the prediction model, and the basic multi-objective artificial bee colony algorithm is improved by approximate ideal solution ranking strategy and grid screening strategy. The weights and thresholds of the cyclic neural network are optimized to avoid the problem that the penalty coefficient is difficult to select in the single-objective interval prediction. The historical load data are memorized and the prediction results are corrected. The accuracy and reliability of short-term load interval forecasting for microgrid are improved effectively. The simulation results show that the multi-objective forecasting method for microgrid short-term load based on probabilistic interval is superior in performance and accurate and can provide decision basis for safe and economic dispatch of microgrid.
【作者单位】： 江南大学物联网技术应用教育部工程研究中心;
【基金】：国家自然科学基金(No.61579167,No.61572237) 高等学校博士学科点专项科研基金(No.20130093110011)
【分类号】：TP18;TM715
【相似文献】
相关期刊论文 前10条
1 郑漳华;艾芊;;微电网的研究现状及在我国的应用前景[J];电网技术;2008年16期
2 麻敏;韩继明;麻秀慧;陈衍妍;;浅谈微电网[J];科技资讯;2009年15期
3 楼书氢;李青锋;许化强;刘鲁丹;;国外微电网的研究概况及其在我国的应用前景[J];华中电力;2009年03期
4 杨为;丁明;毕锐;高研;丁银;;微电网实验平台的设计[J];合肥工业大学学报(自然科学版);2010年01期
5 袁清芳;周作春;陈艳霞;李香龙;陈国锋;;微电网发展应对策略[J];农村电气化;2010年10期
6 安智敏;罗时光;;微电网的概念及发展[J];内蒙古石油化工;2011年21期
7 王t，
本文编号：2388175
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2388175.html
