集中供热系统热负荷预测方法研究

发布时间：2018-05-14 07:44

本文选题：集中供热系统 + 热负荷预测　；参考：《长春工业大学》2017年硕士论文

【摘要】：我国北方城镇冬季供暖所需的供热能耗在社会能源消耗中占的比重很大,随着国家对节约能源的日益重视,大部分地区都采用了集中供热的供暖方式。但是由于集中供热系统覆盖区域广阔,控制和调节十分困难,因此在供热系统运行调节过程中对热用户的实际供热量进行预测显得十分重要。为了能够更加准确的对集中供热系统进行热负荷预测,本文对热负荷的影响因素进行了相关性分析,在确定预测模型的输入变量和评价标准的基础上,将神经网络和支持向量机用于热负荷预测,建立了各自不同的优化改进模型,并在热负荷预测方面的表现做具体的研究。对于神经网络算法,本文建立了BP神经网络预测模型,然后用小波分析理论对模型进行了改进并建立了小波神经网络预测模型;对于支持向量机算法,本文建立了支持向量机回归预测模型,采用粒子群算法对支持向量机预测模型进行参数寻优,并由此建立基于粒子群支持向量机预测模型。为了对支持向量机学习能力进一步提高,故本文采用了动态多种群粒子群优化支持向量机算法,并由此建立了基于动态多种群粒子群支持向量机热负荷预测模型。通过对各种算法的集中供热热负荷预测模型的分析和计算结果表明:支持向量机算法比神经网络在处理与供热负荷有关的较多影响因素的高维数学问题方面更为先进;动态多种群粒子群算法在参数寻优中的搜索能力明显要优与粒子群算法;优化后的预测模型的预测精度要高于原始预测模型;使用支持向量机及其优化算法建立的集中供热系统热负荷预测模型的预测效果整体优于使用神经网络及其优化算法建立的预测模型。在基于供热的实测数据的基础上,通过分析、对比各个模型,并综合评价因素,本文采用的基于动态多种群粒子群支持向量机热负荷预测模型稳定性好,预测精度高,能够精确有效的为供暖企业科学生产提供有效的参考,为热源分配、调度提供必要依据。
[Abstract]:The energy consumption of heating in northern cities and towns in winter accounts for a large proportion of the social energy consumption. With the increasing attention paid by the state to energy conservation, central heating is used in most areas. However, because the central heating system covers a wide area, it is very difficult to control and regulate, so it is very important to predict the actual heat supply of heat users in the operation and regulation process of the heating system. In order to forecast the heat load of the central heating system more accurately, this paper analyzes the influence factors of the heat load, on the basis of determining the input variables and evaluation criteria of the prediction model. Neural network and support vector machine are applied to heat load forecasting. Different optimization and improvement models are established, and the performance of heat load forecasting is studied in detail. For the neural network algorithm, the BP neural network prediction model is established, and then the wavelet analysis theory is used to improve the model and establish the wavelet neural network prediction model. In this paper, the regression prediction model of support vector machine is established, and the parameter optimization of support vector machine prediction model is carried out by using particle swarm optimization algorithm, and the prediction model based on particle swarm optimization support vector machine is established. In order to further improve the learning ability of support vector machine (SVM), the support vector machine (SVM) algorithm based on dynamic multi-swarm particle swarm optimization (DMLPSO) is adopted in this paper, and a heat load forecasting model based on DVM (support Vector Machine) is established. The analysis and calculation results of various algorithms show that the SVM algorithm is more advanced than the neural network in dealing with the high-dimensional mathematical problems related to the heat supply load. The search ability of dynamic multi-swarm optimization algorithm in parameter optimization is obviously superior to that of particle swarm optimization, and the prediction accuracy of the optimized prediction model is higher than that of the original prediction model. The forecasting effect of the heat load forecasting model of central heating system based on support vector machine and its optimization algorithm is better than that established by neural network and its optimization algorithm. On the basis of the measured data based on heating, through analyzing, comparing each model, and synthetically evaluating factors, the heat load forecasting model based on dynamic multi-swarm optimization support vector machine has good stability and high prediction precision. It can provide an effective reference for the scientific production of heating enterprises and provide the necessary basis for the distribution and dispatch of heat sources.
【学位授予单位】：长春工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU995;TP18

【参考文献】