热水锅炉燃烧系统建模与优化控制研究

发布时间：2018-10-20 07:45

【摘要】：在我国北方地区,冬季非常寒冷,采暖供热是居民生活的重要需求。因此,供暖锅炉应用非常广泛。但是由于供暖锅炉的燃烧控制系统多由人工进行操作,自动化程度很低,造成了能源利用率低,大气污染严重等情况。供暖锅炉的燃烧系统是一个较难控制的对象,具有非线性、强耦合、大延时、时变性等特点。如何根据负荷的变动,合理有效地控制供暖锅炉的燃烧系统,对确保锅炉高效、环保与安全运行具有重要的意义。本文针对供暖锅炉中应用最多的链条式燃煤热水锅炉进行研究,分析了其主要工作过程与原理以及锅炉燃烧系统的主要任务。在此基础之上,通过研究热水锅炉的燃烧特质和控制方法,明确了燃烧控制系统中各变量之间的相互关系,提出了由负荷控制、送风控制和引风控制三个相对独立的子系统组成的燃烧控制系统整体控制方案。本文以大连鑫昱供热公司20t/h链条式燃煤热水锅炉为实验对象,通过分析燃烧系统的主要参数,给出了锅炉燃烧系统的三输入三输出模型。通过分析多输入多输出系统参数辨识的原理,将热水锅炉三输入三输出系统转化为由三个三输入单输出传递函数组成的系统,根据锅炉实际运行的数据,采用递推最小二乘参数估计方法对锅炉燃烧系统进行数学建模,取得了良好的建模效果。针对燃烧控制系统多输入多输出且各参数之间相互耦合的特点,给出了PID神经网络控制算法,但由于其初始权值随机获取,控制效果易陷入局部最优,本文提出了一种新的自适应变异粒子群优化算法。将系统辨识模型作为控制对象进行Matlab仿真实验,自适应粒了群优化的PID神经网络器与PID神经网络器、基本粒子群优化的PID神经网络器相比,不仅响应快速,逼近目标时间更短,而且稳态相对误差更小,解决了PID神经网络算法初始权值易陷入局部最优值以及粒子群算法容易早熟收敛的问题,得到了更好的控制效果。
[Abstract]:In northern China, the winter is very cold, heating and heating is an important demand of residents' life. Therefore, heating boilers are widely used. However, the combustion control system of heating boiler is operated by manual, and the degree of automation is very low, which results in low energy efficiency and serious air pollution. The combustion system of heating boiler is a difficult object to control, which has the characteristics of nonlinear, strong coupling, long time delay and time varying. How to control the combustion system of heating boiler reasonably and effectively according to the change of load is of great significance to ensure the high efficiency, environmental protection and safe operation of the boiler. In this paper, the chain type coal-fired hot water boiler, which is widely used in heating boiler, is studied, its main working process and principle and the main task of boiler combustion system are analyzed. On this basis, by studying the combustion characteristics and control methods of the hot water boiler, the relationship between the variables in the combustion control system is clarified, and the load control is proposed. The whole control scheme of combustion control system is composed of three relatively independent subsystems: air supply control and air supply control. This paper takes the 20t/h chain-fired hot water boiler of Dalian Xinyu heating Company as the experimental object. By analyzing the main parameters of the combustion system, the three-input and three-output model of the boiler combustion system is given. By analyzing the principle of parameter identification of multi-input multi-output system, the three-input three-output system of hot water boiler is transformed into a system composed of three three-input single-output transfer functions. The recursive least square parameter estimation method is used to model the boiler combustion system, and a good modeling effect is obtained. In view of the characteristics of multiple inputs and multiple outputs of combustion control system and the coupling of various parameters, a PID neural network control algorithm is presented. However, due to the random acquisition of the initial weights, the control effect is easy to fall into local optimum. In this paper, a new adaptive mutation particle swarm optimization algorithm is proposed. The system identification model is used as the control object in the Matlab simulation experiment. Compared with the PID neural network and the PID neural network based on the basic particle swarm optimization, the PID neural network based on self-adaptive particle swarm optimization not only has a fast response, but also has a shorter time to approach the target. Moreover, the relative error of steady state is smaller, which solves the problem that the initial weight of PID neural network is easy to fall into local optimal value and particle swarm optimization is easy to converge prematurely, and the better control effect is obtained.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TU832.21;TP273

【参考文献】