基于模糊神经网络的供热负荷预测

首先在对供热负荷预测算法的发展现状主要成果阐述的基础上,对影响供热预测因素采用模糊量化的方式进行研究处理,并由此推断将模糊神经网络算法应用于供热负荷预测可以得到良好的效果.研究模型的设计核心是BP神经网络,即将模糊量化后的影响因素作为系统的输入值,去调整神经网络的权值,从而得到预测的网络模型.建立预测模型和预测策略后,可以采用Matlab科学计算软件开发程序对预测模型效果进行模拟仿真,结果表明,预测的结果能够满足要求,相对误差在合理的范围内,并且模糊神经网络算法比单纯神经网络算法具有更好的预测精度和鲁棒特性,从而达到节能的目的.且适应性强,可以应用到类似的供热工程上.

Heat load forecast based on the fuzzy neural network controller

The characteristics and present situation of heating load forecast were summarized, through analyzing and studying the various factors that affect the heating load. The processing approach for influencing factors was proposed by using quantitative of fuzzy data. On this basis, a new type of fuzzy neural network forecasting system was used. In the system was used, the sophisticated BP network method was used as the design core, and the quantitative parameters of fuzzy data for influencing factors were used as input values, and then the forecast network model was gotten. After the model operation parameters were determined, Matlab7.0 was used to simulate and predict. The results show that its accuracy can achieve ideal result under the condition of equal or approximate, and the relative error is under a reasonable range. Moreover, the hybrid algorithm rather than the simple fuzzy algorithm or neural network algorithm has a better prediction precision and stronger generalization ability, which improve the heating quality and save energy.