基于相似日的遗传程序设计短期负荷预测模型

Short-term Load Forecasting Model Established by Genetic Programming Based on Similar Days

针对短期负荷预测中对自动化、智能化和预测精度的进一步需求,提出了基于相似日的遗传程序设计短期负荷预测模型.首先从日特征量、日前趋势相似度及其综合来选择预测日的趋势相似日样本数据,然后重点介绍了利用相似日样本数据进行短期负荷预测遗传程序设计演化建模的算法设计流程.贵州电网日96点负荷实例预测结果表明,模型所对应的预测曲线与原始曲线拟合较好,模型具有较高的预测精度,能够反映短期负荷变化的规律.

For the further requirements and insufficient of automation, intelligence and forecasting precision in forecasting the short term electric power loads, a shortterm load forecasting model established by genetic programming （GP） based on similar days is pro posed in this paper. At first, the sample data of trendsimilar days of the forecasting day are chosen from charactersimilar days, trend similar days and their integration. Then using the sample data of similar days, the design process of evolutional algorithm for modeling the shortterm load forecasting by GP is introduced in detail. It is applied to forecasting Guizhou Power Grid 96 points load. The result shows that the prediction curves of the model can fit the original curve very well, and the model is of high precision so as to reflect shortterm load changing laws.