【作者】 王庆峰；

【导师】 林颐清；

【作者基本信息】 山东大学 ， 工程热物理， 2010， 硕士

【摘要】 目前城市集中供热已成为我国北方地区冬季供热的一种主要形式,并且发展较快。城市集中供热系统热能耗费数量很大,品味较低,通常是120℃以下的低位热能,却主要以高品位的一次能源来供应,故具有较大的节能潜力。尤其是在供热系统运行过程中,如何通过实施运行调节,使系统在最优状态下运行,提高供热系统的经济效益和节能水平,是供热领域亟须解决的问题,而与之相关的理论和应用研究已经得到关注。针对城市集中供热系统一次热网多采用质、量并调这种调节方式的现状,以运行能耗费用最低为优化目标,以供热管网的运行特性为约束条件,建立以供水流量和供回水温度为变量的集中供热系统一次热网运行能耗费用方程,通过求解该方程得到指导一次热网进行质、量调节的最优运行参数。使用非线性规划方法提供的分析思路和数值方法分析和求解方程,并通过在求解方程的不同阶段选择合适的方法建立了用于求解该方程的软件系统。通过对计算结果的分析,证明课题提出的运行能耗费用方程在指导集中供热系统一次热网的质、量调节时,比集中质调节具备更好的节能效益,同时证实用非线性规划方法开发的软件系统求解一次热网运行能耗费用方程比使用Matlab工具箱和步长法效率更高,结果更加准确。针对城市集中供热系统热负荷预测成为集中供热系统运行调节前提和基础的实际情况,考虑到供热系统在实际运行过程中供热热负荷与诸多影响因素之间具有非线性和动态性关系的特点,使用人工神经网络技术实现了某小区二级换热站热负荷的预测,提出了确立BP神经网络结构和参数的方法,并使用开发的软件系统对工程实例热负荷进行了预测,通过对预测结果的分析表明BP神经网络用于热负荷预测能够获得较高的精度,满足一般的应用需求。本文选择Visual C#程序设计语言和Microsoft Visual studio 2008程序设计平台实现上述算法和程序,它们所具备的显著特点,如面向对象技术、图形用户界面和软件部署等,为以后软件系统的后续研发提供了较好的基础框架。更多还原

【Abstract】 At present, district heating system is very popular in northern part of China and develops rapidly. The system consumes large quantity thermal energy which is of low level. The low level thermal energy which is usually 120℃is mainly supported by high level thermal energy, so it has much energy saving potential in district heating system operation, how to carry out the regulation to make the system run in best condition is a hot topic which is urgently to be solved, the relative theory and research have been focused.The primary heating networks are generally regulated by quality and quantity method. The optimization objective is the minize operation cost. The constraints conditions are dynamic characteristic of heating networks. The supply temperature and return temperature are variable value. The operation cost equation is set up and to be solved. The corresponding software program is also developed. Through the comparison, it is proved that the equation presented in this paper is of validity. Tt is also pointed out that the quality and quantity regulation is prior to the central heating quality regulation. And the non-linear optimize method is proved more efficient than toolbox of MATLAB and step method.For the heat load prediction is basis for the regulation of district heating system, considering the nonlinear characteristic and dynamic characteristic of the heat load which may be related to many factors, the neural works technology is adopted to implement the heat load prediction of secondary heat station. The method of confirm the Bp neural work structure and parameters is presented in this paper and the software is used to predict the real data. By the analysis of the prediction result, it is proved that he BP neural applied in the heat load prediction can get high accuracy and meet the common demand.Visual C# and Microsoft Visual studio 2008 Program platform are applied in this paper to implement the above algorithm and program. The excellent basic frame is provided for the following research and development of software.更多还原