【摘要】 循环流化床锅炉具有污染物排放少、燃料适应性广、负荷调节能力强等优点,近年来在电力、供热等行业中得到广泛应用。然而目前大部分循环流化床锅炉均存在自动投入率低,操作依赖人工经验的特点,造成这一状况的一个重要原因是缺乏合理的数学模型。首先对工艺流程进行分析,选取对锅炉效率影响最大的7个参数作为建模对象:过量空气系数、床温、排烟温差、飞灰含碳量、一次风机电流、二次风机电流、引风机电流。每个参数有各自不同的特点,对不同的统计模型的适用性也不尽相同。为了达到最佳建模效果,分别应用多元线性回归、多元逐步回归、偏最小二乘回归及BP神经网络对这些参数进行建模。实例研究表明,过量空气系数和二次风机电流适合采用偏最小二乘回归法建模;床温、排烟温差、一次风机电流和引风机电流适合采用多元线性回归法建模;飞灰含碳量采用BP网络模型对其预测效果相对较好。本文所建的模型对循环流化床锅炉的节能分析和进一步的操作优化研究具有一定的实际意义。更多还原

【Abstract】 Circulating fluidized bed boiler with many advantages such as less pollutant emission,fuel adaptability,strong load regulation capability has been widely applied in power,heating and other industries in recent years.Because of the lack of a reasonable mathematical model,the operation of the circulating fluidized bed boilers is depended on the artificial experience.Firstly,seven most important parameters were selected as the modeling object:excessive air coefficient,bed temperature,exhaust gas temperature,carbon content in fly ash,current of primary air fan,current of second air fan,current of induced draft fan.Each parameters has their own characteristics.In order to achieve the best effect of models,four different statistical models as multiple linear regression,multiple stepwise regression,partial least squares regression and BP neural network were introduced.The case study shown that,the excess air coefficient and the two fan current is suitable for partial least squares regression model;bed temperature,exhaust gas temperature,current of primary air fan and current of induced draft fan is suitable for multivariate linear regression model;forecast effect of carbon content of fly ash with BP network model is better.The models for circulation fluidized bed boiler proposed in this paper have certain practical significance for operation optimization analysis and further study.更多还原