摘要
在城市集中供暖方面,热电厂的短期热负荷预测对提高热电厂的经济效益和热能利用率十分重要。该文以山西某热电厂的供热系统的换热站作为研究对象,使用遗传算法和粒子群算法改进BP神经网络,基于热负荷相关的历史数据构建改进型神经网络的热负荷预测系统。仿真结果显示,BP神经网络预测系统的波动程度比较大,预测精度低,而改进型的神经网络算法克服了这些缺点,在历史样本数据较少的情况下,仍然保持很高的预测精度,改进后的预测系统精度较高、稳定性较强,满足工业生产需求。
In terms of urban central heating,the short-term heat load prediction of thermal power plants is very important to improve the economic efficiency and thermal energy utilization rate of thermal power plants.This paper takes the heat exchange station of the heating system of a thermal power plant in Shanxi as the research object,uses genetic algorithm and particle swarm optimization to improve the BP neural network,and builds an improved neural network thermal load prediction system based on the historical data of thermal load.The simulation results show that the BP neural network prediction system has a relatively large degree of fluctuation and low prediction accuracy,and the improved neural network algorithm overcomes these shortcomings.In the case of less historical sample data,it still maintains high prediction accuracy and improvement.The prediction system afterwards has high precision and strong stability,and meets the needs of industrial production.
作者
王琦
胡磊
杨超杰
WANG Qi;HU Lei;YANG Chaojie(Department of Automation, Shanxi University, Shanxi Taiyuan 030013,China)
出处
《工业仪表与自动化装置》
2020年第6期11-16,共6页
Industrial Instrumentation & Automation
关键词
热负荷预测
BP神经网络
改进型神经网络
预测精度
thermal load forecasting
BP neural network
improved neural network
prediction accuracy
