用神经网络预测煤及配煤的着火特性

【作者】 周坤；

【导师】 赵虹； 杨建国；

【作者基本信息】 浙江大学 ， 工程热物理， 2005， 硕士

【摘要】 随着我国工业的快速增长,煤的消耗量增加,加上运输困难、煤矿分布不均匀以及国家有关对燃煤锅炉尽量燃用劣质煤的政策等因素,造成电厂不可能长期燃用单一煤种,而不得不燃用配煤。而煤的着火特性对锅炉的安全经济运行非常重要,所以本文重点研究配煤的着火特性。 煤粉自身特性是影响着火特性的一个非常重要因素,而着火特性和煤质之间的内在关系是复杂的非线性关系,从某个单一煤质指标、人为经验或经验计算公式往往不能准确地描述,为了解决这一问题,本文采用BP神经网络,以大量的煤质分析数据和热天平试验数据为基础,建立了三个神经网络模型,如下: 1.以单煤煤质和掺混比例为输入数据,配煤的相应的煤质为输出数据,建立了通过单煤煤质预测配煤煤质的神经网络模型。 2.分三种情况建立了工业分析、发热量和着火温度的BP神经网络模型。 3.分三种情况建立了工业分析、发热量和着火稳定性指标的BP神经网络模型。 结果表明,用BP神经网络可以达到很高的预测精度,同时还可以比较输入和输出的相关性。 本文还针对五组不同煤质相混的配煤的热分析曲线进行了分析,研究了配煤的着火特性。并选取了10个配煤对已建立的神经网络模型预测的正确性进行了验证。更多还原

【Abstract】 Along with the rapid development of our country’s Industy, the factors induce that the power plant combusts the single coal impossibly for a long time, which are the increase wastage of coal, the difficulty transportation, the asymmetric distributing of coal mine and the policy of nation about that power boiler combusts inferior coal, therefore, the power plant must combusts blended coals. However, the characteristics of ignition is very important to the safe and economical boiler operation, so that the article researches on the characteristics of blended coals’ ingnition.The coal characteristics is one of very important factors affecting the characteristics of ignition, but it is complex nonlinear correlation between inner-relation of coal quality and igniting characteristics, which cannot be described exactly by certain single coal quality index, the people’s experience and the empiric formulas. For the sake of solving this problem, BP neural network is adopted in this article. Based on mass data of coal analysis and thermogravimetric analysis, the article establishes three models of BP neural network, as follows:1 , The single coal grade and mixed proportion are import data, and the corresponding blended coals grade is export data. Therefore, the article establishes BP neural network model which forecast the blended coals grade through single coal grade.2 , This article is divided into three kinds of condition and establishes BP neural network model from proximate analysis and calorific value to ignition temperature.3 , This article is divided into three kinds of condition and establishes BP neural network model from proximate analysis and calorific value to ignition stability index.The prediction result shows that the high precision of prediction can be achieved by using BP neural network, and that the relativity of input data and output data can be compared by using the BP neural network too.The article also analyses the five thermal analysis curves of blended coals made up of different coal grade, researches the characteristic of blended coals ignition, and chooses ten blended coals validate the result of establishing BP neural network models.更多还原