基于BP神经网络的CFB锅炉飞灰含碳量建模

Modeling of carbon content in fly ash of CFB boiler based on BP neural network

飞灰含碳量是影响锅炉热效率的重要指标,影响着机组的经济运行。建立一种基于Levenberg-Marquardt(L-M)算法改进的BP神经网络模型,对某电厂150 MW CFB锅炉的飞灰含碳量进行建模预测,包括1个母模型和3个子模型。母模型选取煤的工业分析、低位发热量等7个参数作为输入参数,子模型研究煤质参数偏差对母模型其他输入参数的影响。利用改进的BP神经网络分别对样本进行训练,预测飞灰含碳量。将训练结果与传统多项式回归法或经验方法得出的结果进行对比。结果表明,BP神经网络、多项式线性回归(PLR)、多项式非线性回归(PNR)的相关系数R2分别为0.9571、0.6051、0.7667,相对平均误差RME分别为4.84%、17.02%、12.46%。改进的BP神经网络模型对飞灰含碳量具有更高的预测精度和更好的泛化能力。

The carbon content of fly ash is an important indicator that affects the thermal efficiency of the boiler and the economy operation of the units.An improved BP neural network model based on the Levenberg-Marquardt algorithm was established to predict the carbon content of fly ash in a 150 MW CFB boiler,including a parent model and three sub-models.The parent model selected seven parameters such as the technical analysis and low calorific value of coal as input parameters.The sub-models investigated the coal quality deviation parameters on other input parameters of the parent model.The improved BP neural network was used to train the samples to predict the carbon content of fly ash.The training results were compared with the results obtained by traditional polynomial regression methods or empirical methods.The results show that the correlation coefficient R^(2)of BP neural network,Polynomial linear regression,and Polynomial nonlinear regression are 0.9571,0.6051,0.7667,respectively,and the relative mean error RME are 4.84%,17.02%,12.46%,respectively.The improved BP neural network model has higher prediction accuracy and better generalization ability for fly ash carbon content.