广州电网负荷特性分析及短期预测模型设计

结合广州电网负荷短期预测系统的开发工作，设计了合理的预测模型。分析表明，该地区负荷表现出较强的以周为间隔和以日为间隔的周期性，民用负荷占据较大份额，日负荷分时段特性明显。的预测系统基于人工神经网络技术，针对性地分别建立了人工神经网络训练周模型和日模型，在对历史电网负荷和气象数据进行预筛选的基础上，结合对日负荷的分时段预测处理，开发短期预测系统。系统具有较高的预测效率和满意的预测准确度。针对该地区夏季高温季节出现的负荷饱和特性，设计了基于专家知识的预测检验环节，运行结果理想。

Guangzhou regional load analysis and short-term forecasting model design

Based on the development of Guangzhou power short term load forecast system,a feasible forecast model is designed.By analysis,the load characteristics of Guangzhou region show a strong periodicity of week and day,and the civil load takes a great part with obvious time sections of daily load.The developed forecast system is based on ANN (artificial neural networks)technology,and the daily training model and weekly training model are designed.According to the filtered historic net load data and weather data,and combined with the time sectional pre processing of daily load,a short term forecast system is developed.It has higher forecast efficiency and satisfying forecast precision.Regarding to the load saturation appeared in summer with ultra higher temperatures,a forecasting result verification module based on expert knowledge is designed.The operating results are excellent.