模糊耦合神经网络的系统供热预测研究

Research on Heat Forecasting Based on Fuzzy Neural Network

热流量的准确预测,对于日益庞大的城镇供热系统能效提升起着至关重要的作用。首先对影响热流量的各因素应用数理统计的方法进行分析,然后采用模糊耦合神经网络算法对供热系统建立热流量预测模型,并通过某热源的历史运行数据进行分析,与小波神经网络算法比较,证实了该耦合算法的高准确性。这种将模糊理论与神经网络相结合的预测方法兼顾了二者的优点,使其既具有模糊逻辑的不确定信息处理能力,又可以有神经网络的自学习能力。该研究符合城镇供热系统非线性的特性,可以应用于智能热网的热流量预测。

Accurate forecasting of heat supply is very important to improve energy efficiency of urban heating system.The paper analyzed firstly each factor influencing heat supply by mathematical statistics method,and then presented a hybrid algorithm of fuzzy combined with neural network for forecasting heat supply in heating system.It turned out that the hybrid algorithm has high accuracy and application promotion value through analyzing the historical operating data of some heat source and comparing with wavelet neural network approach.The hybrid algorithm possesses uncertain information processing capability,and self-learning ability of neural network,and can be applied to the prediction of heat supply for intelligent heat network.