【作者】 刘明；

【导师】 高丙坤；

【作者基本信息】 东北石油大学 ， 油气信息与控制工程， 2011， 博士

【摘要】 本文就油田热水供暖系统热负荷智能预测技术进行了较深入的研究。首先,分析了研究油田热水供暖系统热负荷智能预测技术的现实意义和必要性,介绍了供热参数辨识领域的主要研究方法；接下来,系统阐述了供热运行调节的理论基础及主要调节方式；通过分析总结前人的经验与不足,在供热运行调节理论的指导下,提出了本文的热负荷预测方法。第一,对数据提取技术进行了研究。采用WebService技术编写数据访问接口程序,自动实时提取天气预报数据和各整点天气实况数据,为供热运行调节提供依据；采用C#.Net组件技术编写数据访问接口程序,供热运行实时数据采集系统中定时提取供热运行参数的实测数据,以便随时掌握供热系统的运行状况。第二,建立室外温度日变化模型。室外温度变化模型的建立是进行热负荷预测的前提和基础。该模型根据当天天气预报数据、历史气温变化规律、日出日落时间等因素确定若干个关键时刻的气温值,然后采用三次样条插值的方法预测全天的气温变化。第三,提出一种基于最小二乘法的热负荷参数预测方法。假定室外气温变化模型已建立,供热系统的设计热指标及供热网的实际参数已设定。通过整理分析历史供热运行数据(其中包括室外温度、出口温度、回水温度、出口流量、回水流量等),采用最小二乘法找出供水温度、回水温度、室外温度及流量之间的函数关系,建立供回水温度预测模型。第四,提出一种基于BP神经网络的热负荷预测方法。针对区域供热系统的特点,在对三层BP神经网络算法分析的基础上,结合该项目的实测数据,建立了完整的供热系统参数模型。第五,提出一种基于小波神经网络热负荷参数预测方法,针对实际供热情况,在对小波神经网络算法分析的基础上,结合项目实测数据,建立了完整高效的热负荷参数模型,充分发挥了小波神经网络的众多优点。最后,提出一种基于改进粒子群算法的热负荷组合预测方法,在充分发挥了改进粒子群算法的优势下,通过对前三种单项预测模型的在组合预测模型中权重的计算,,建立了基于改进粒子群算法的热负荷组合预测模型。更多还原

【Abstract】 This article makes an intensive research on the intelligent forecasting technology of heat load of oilfield hot water heating system. first, the practical significance and necessity of the intelligent forecasting technology of heat load of oilfield hot water heating system,and the main research methods in the field of heating parameter identification are introduced; then, the theoretical basis and ways of control and regulation of heat supply are formulated; finally, this article puts forward a heat load forecasting technique, guided by the theory of control and regulation of heating operation, through analyzing previous experiences and shortcomings.The first, the data fetching technique is studied. Data interface program is programmed by using WebService, automatically and real-timely acquiring weather forecasts data and hourly actual weather data, to provide a basis for heating conditioning; data interface program is programmed by using C#.Net components, acquiring data of heating operational parameter regularly by real-time data fetching system, in order to know the operation conditions of heating system anytime.The second, an outdoor air temperature daily variation model is built, which is the precondition and basis of heat load forecasting. The model determines several temperature values at some critical moments, according to such factors as weather forecasting data of the present day, the historical change law of air temperature, and the time of sunrise and sunset etc, and employs cubic spline interpolation to forecast the temperature variation of the whole day.The third, a heating parameter forecasting method is come up with on the basis of method of least square. Supposed that the outdoor temperature variation model has been established, and the designed thermal parameter and the actual parameter of heating system have been set, through analyzing historical heating operation data (including, out let flow, return water flow etc.), employing the method of least square to find out the functions between outlet temperature, temperature of return water, outdoor temperature and water flow, the model to forecast the temperature of water supply and return water is established.The fourth, this article puts forward a heating parameter forecasting method on the basis of BP neural network. In view of characteristics of district heating system, on the basis of analysis of the three-layer BP neural network arithmetic, combined with the measurement data from the project, a complete heating system parameter model is established.The fifth, this article puts forward a heating parameter forecasting method on the basis of wavelet neural network. For the case of the actual heating, on the basis of analysis of the wavelet neural network arithmetic, combined with the measurement data from the project, a complete heating system parameter model is established, taking advantage of the strengths of wavelet neural network.Finally, a combination forecasting method of heat load is come up with on the basis of improved PSO, by advantaging into full play improved PSO, calculating the weight of the first three models in combination forecasting model, a combination forecasting method of heat load on the basis of improved PSO is established更多还原