基于改进极限学习机的电力需求预测研究

Research on Electric Power Demand Prediction Based on Improved Extreme Learning Machine

针对电力需求预测的准确与否对电力企业的健康发展以及国民经济具有重要影响的问题,论文提出了基于改进极限学习机的电力需求预测模型。首先,对电力需求预测进行了分析,根据电力需求的特点将电力需求分为工业、农业、交通、居民和动力用电,并分析了电力需求预测的原理和过程;其次,提出了基于GA改进的ELM模型,通过GA的优化寻优能力提升ELM网络模型的拟合能力;然后通过实际的案例仿真,分别对某地区的大工业、非普工业、商业、居民生活用电和非生活用电等方面进行了模型预测研究,算例仿真误差较小,验证了论文所提方法的有效性和适用性。

In view of the electric power demand forecasting accuracy or not for the healthy development of the electric power enterprise,and has the important influence on national economy,this paper puts forward the electric power demand forecasting model based on improved extreme learning machine.First of all,the electric power demand forecasting is analyzed,according to the characteristics of the demand for electricity power demand can be divided into industry,agriculture,transportation,residents,and power consumption of power,and the principle and process of electric power demand forecasting is analyzed.Secondly,an ELM model based on GA is proposed to improve the fitting capacity of ELM network model through GA optimization.Then,through the actual case simulation,the model prediction research is carried out on the large industry,the non-pup industry,the commercial,the residential electricity and non-living electricity consumption of a region.The simulation error is small,and the validity and applicability of the proposed method are verified.