【作者】 连慧莉；

【导师】 雎刚；

【作者基本信息】 东南大学 ， 热能工程， 2005， 硕士

【摘要】 对于燃煤机组,锅炉的运行性能与入炉煤质特性密切相关。由于目前电厂一般只能进行煤的工业成分分析,不能对煤的元素成分、燃烧特性参数进行实验室分析,不利于锅炉的安全经济运行。论文以神经网络技术为工具,研究建立基于煤的工业成分的燃煤特性预测模型,具有重要的理论意义和应用研究价值。神经网络具有较强的逼近非线性函数的能力,并具有自适应学习、并行分布处理和较强的鲁棒性及容错性等特点,为解决未知不确定非线性系统的建模问题提供了一种有效途径。本文在已有RBF神经网络的学习算法的基础上,提出了新型的RBF神经网络的构造方法,并将RBF神经网络应用于燃煤特性预测模型的建立。主要工作及取得的研究成果如下:(1)神经网络建模方法研究。对RBF神经网络进行了综述,并提出了一种改进的基于免疫原理的RBF神经网络混合学习算法,利用人工免疫系统的记忆、学习和自组织调节原理进行初始RBF中心的选择,然后再利用梯度下降法对RBF神经网络的参数进行有监督学习。对提出的算法进行了仿真,仿真结果表明了这种方法是有效的。(2)电站锅炉燃煤特性研究。包括煤的工业分析、元素分析、燃烧特性,煤灰的结渣特性诸多方面。特别对燃煤燃烧特性和煤灰结渣特性的影响因素和判别方法进行了研究总结。(3)用RBF神经网络建立燃煤特性预测模型。用RBF神经网络建立了煤的工业分析成分和煤的元素分析成分之间转换模型、煤的工业分析成分预测煤的燃烧特性的模型以及煤灰成分预测燃煤结渣指数的模型。仿真结果表明,所建立的模型具有较高的精度,并具有较好的泛化能力。(4)燃煤特性神经网络预测模型的应用研究。提出了采用煤的元素成分神经网络预测模型,将煤的工业分析成分转换成煤的元素成分来计算锅炉效率的方法,并与直接采用煤的工业分析成分计算锅炉效率的结果进行比较,结果表明采用元素成分神经网络模型计算锅炉效率,可有效提高计算的准确性。更多还原

【Abstract】 As to coal-burning plant unit, the running ability of boiler has intimate relation with the characteristics of the coal in the boiler. Recently in the power plant we could only analysis the industry component of the coal, but can’t analysis the element and combustion characteristic under lab circumstance. As a result, it is not good for the security and the economical operation of the boiler. The paper took neural network technology as a tool, researched the prediction model which based on the combustion characteristic of the industry component of the coal. There will be crucial theoretic significance and applied research value.Due to their powerful ability of approximating nonlinear functions, and with the characteristics of adaptive learning, parallel and distributed processing, strong robustness and fault tolerance, neural networks have been an effective approach to model and control the unknown and uncertain nonlinear system. Based on existing learning algorithms for RBF neural networks, novel learning algorithms are proposed in this paper, the application of RBF neural networks with proposed algorithms in nonlinear systems modeling is studied, and simulation study for characteristics of coal. The main works are:1.Research on neural network modeling method. The paper summarized the RBF neural network and advanced a mixing learning method based on hybrid RBF neural network, the memory, learning and self-organization abilities of artificial immune system are introduced into the selecting of the number and position of hidden layer radial basis function centers, the output layer weights are decided with the recursive least squares algorithm. The simulation results prove that it is with good generalization ability.2.Research on combustion characteristic of the utility boiler coal. Including the technical analysis, element analysis , combustion characteristic of the coal ,and so on. Especially, the paper summarized the research methods and influence factors of combustion characteristic and clogging characteristic.3.Set up a predication model of combustion characteristic by RBF neural network. The paper constructed transfer mold between the industry analysis element and the ultimate analysis element by RBF neural network. Just as the model predicating the combustion characteristic by the industry analysis element, and the model predicating the clogging index by the ultimate analysis element. The simulation results prove that the models has a good accuracy and the ability of generalization.4.The apply research on the predication model of combustion characteristic by neural network. The paper advanced the neural network predication model of element analysis model, and bring up the method which accounting the boiler efficiency by transforming the industry component analysis of the coal into element analysis, then compared with the results of boiler efficiency accounting by industry component analysis. The result proved that the accuracy will be enhanced effectively.更多还原