Study On The Control Method Of Urban Central Heating Network

The energy issue is an important contradiction in the world today,and the problem may bring the wars.With the progress of China's reform and opening up,the pace of urban construction is getting faster and faster.In this process,the city's central heating system becomes more and more complex.Followed by the increasingly is the bad control of the load distribution.Therefore,the network load forecasting of the heating network system have a great significance.This paper introduces the background and significance of the topic.With the country vigorously develop urban construction,the city's heating system has become increasingly large,more and more complex,and the heat distribution is uneven highlights all the phenomena.And then introduced the development of domestic and international heating network status and its control methods.Secondly,the operation mechanism and control scheme of the heating network are introduced from the physical structure of the urban heating network.The concrete factors influencing the load of the heating network are analyzed in detail,and the time series characteristics of the heat load are analyzed in detail,and it is pointed out that it is predictable.Finally,the error of the forecast data is analyzed in detail,and the evaluation index of the prediction error is set.Then,based on the background of particle swarm optimization,the theoretical basis and algorithm parameters of particle swarm algorithm are introduced,and the parameters of the algorithm are analyzed.Then the concept of BP neural network is introduced,and its structure is analyzed deeply.Meanwhile,the method of determining the number of hidden layer nodes is briefly introduced.Finally,the morbidity of the collected network data is analyzed in detail,and the corresponding solutions are put forward,such as abnormal data identification,data completion,data normalization and data denoising.This paper focuses on the data noise reduction,and the method is the soft threshold wavelet denoising method.According to the wavelet coefficients of the noise signal and the wavelet coefficients of the signal,the wavelet coefficients of the noise signal are zeroed,and the wavelet coefficients are processed and the wavelet Inverse operation,and finally get the noise load after the heat load operation data.The standard BP neural network prediction model and the improved PSO-BP neural network prediction model are put forward.The same data are used to train the two kinds of forecasting models respectively.The same data are used to test and the comparison and prediction results are analyzed.PSO-BP neural network prediction model can predict the short-term heat load more accurately.