热力站热负荷神经网络预测模型对比研究  

Comparative Study on Neural Network Prediction Models for Heat Load of Heating Station

作　　者：张昌豪 路绪坤 ZHANG Changhao;LU Xukun

机构地区：[1]郑州热力集团有限公司,河南郑州450000

出　　处：《煤气与热力》2024年第6期V0001-V0005,共5页Gas & Heat

摘　　要：以郑州某热力站为研究对象,通过相关性分析确定热负荷的影响因素。分别建立BP神经网络预测模型、RBF神经网络预测模型以及采用遗传算法优化的BP神经网络预测模型,对3种预测模型的预测效果进行评价。3种预测模型的预测热负荷与实测热负荷的变化趋势基本一致,均可较为客观地反映热负荷的时序特征。与BP预测模型、RBF预测模型相比,BP-GA预测模型的预测值更接近实测值,预测值的误差、相对误差更小。在3种预测模型中,BP-GA预测模型预测效果最佳,且训练时间最短。Taking a heating station in Zhengzhou as the research object,the influencing factors of heat load were determined by correlation analysis.The BP neural network prediction model,the RBF neural network prediction model and the BP neural network prediction model optimized by genetic algorithm were established respectively,and the prediction effects of the three prediction models were evaluated.The variation trends of the predicted heat load of the three prediction models are basically consistent with those of the measured heat load,and the time series characteristics of the heat load can be objectively reflected.Compared with the BP prediction model and RBF prediction model,the prediction value of the BP-GA prediction model is closer to the measured value,and the error and relative error of the prediction value are smaller.Among the three prediction models,the BP-GA prediction model has the best prediction effect and the shortest training time.

关 键 词：热负荷预测 BP神经网络 RBF神经网络 遗传算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]