The Heating Load Forecasting Investigation Based On Fuzzy Neural Network

Along with economy in our country is developing, the heating area isexpanding, heating energy consumption is also constantly increased. In north areas ofour country, the heating energy consumption has achieved27.2percent of totalnational energy consumption. In addition, our country’s energy depletion and thecontinuous development of the automatic control technology, the investment on theheating pipe network is gradually increasing, the research on network is more andmore important. Heating pipe network is a complex distribution network, it is greatsignificant to forecast accurately heating load for the movement management of theheating system、raising heating quality、environment and energy saving.The paper summarized the characteristics and present situation of heating loadforecast, thorough analysis and studies the various factors that affect the heating load,proposed to the processing approach for influencing factors using quantitative offuzzy data. On this basis, this paper used a new type of fuzzy neural networkforecasting system. In the system, the sophisticated BP network is used as the designcore, and the quantitative parameters of fuzzy data of influencing factors are used asinput values, then get the forecast network model.After the model operation parameters determined, using MATLAB7.0simulationto predict, the results show that under the condition of equal or approximate itaccuracy can achieve ideal result, and the relative error in reasonable range. Besides,the hybrid algorithm than simple fuzzy algorithm and neural network algorithm hasbetter prediction precision, stronger generalization ability, which can improve thebetter heating quality and save energy.Finally, on the basis of the realization of heat load forecast, study the control ofcentral heating, due to the result of the heating system nonlinear characteristics, itmake the control complexity, this paper discusses the necessity of control and thecontrol strategy.