短期电网负荷预测方法研究

Study on the prediction method of short-term power load

短期负荷预测主要指对未来几日或者几周时间内的电力电网负荷进行预测,目的是为发电厂的符合分配,机组启停以及设备检修和燃料能源供应计划提供指导。考虑到常规的SVR预测模型采用人工经验的方法对RBF核函数参数、不敏感系数和惩罚系数等参数进行选取,这样常规SVR算法的缺陷就是其性能会因为随机选取的参数而变得随机和不确定,因此本文使用人工鱼群优化算法对SVR参数选取进行优化。为了提高人工鱼群算法全局搜索能力,将全局最优的信息融入到人工鱼的觅食、聚群、追尾移动选择过程中。改进后的全局人工鱼群算法,能够避免传统的人工鱼群算法在对人工鱼移动方向选择时没有考虑全局信息而引起的收敛效率和收敛精度低等缺点,因此能够更加快速精确地搜索到全局最优解。最后,通过实验方法,对本文研究的短期电网负荷预测方法进行验证。结果表明,本文研究的短期负荷预测,预测精度较高,具有较好的工程应用价值。

short term load forecasting mainly refers to the future for a few days or a few weeks in power load forecasting, the purpose is for power plant to meet the distribution, unit start stop and equipment maintenance and fuel energy supply plan to provide guidance. Con- sidering the conventional SVR prediction model using artificial experience method to select the parameters of RBF kernel function parameter, the insensitive factor and punishment coefficient, such conventional SVR algorithm flaw is that the performance will be because of the random selection of parameters and become random and uncertain. Therefore, this paper used artificial fish swarm optimization algorithm of SVR parameters selected to optimize. In order to improve the performance of artificial fish swarm algorithm is a global search capability,the global optimal information into the artificial fish foraging, Poly Group, rear end mobile selection process. The improved global artificial fish swarm algorithm can avoid the traditional artificial fish swarm algorithm in the choice of the direction of movement of artificial fish did not consider the global information and cause the convergence speed and convergence accuracy and other shortcomings, so it can more quickly and accurately search to the global optimal solution. Finally, through the experiment, this paper studies on the short -term power load forecasting method is verified. The results show that the short-term load prediction, the prediction accuracy is high, and has good engineering application value.