对煤粉炉中影响飞灰含碳量的主要因素进行分析,并根据正交实验原理,对某台670t/h的燃煤锅炉飞灰含碳特性进行多工况热态实验,采用基于LM(Levenberg-Marquardt)算法的BP神经网络建立了锅炉飞灰含碳量的神经网络预测模型,实验验证结果表明该算法不仅收敛速度快,而且模型能根据各种操作参数准确地预报锅炉在不同工况下飞灰含碳量。 The main factors affecting the carbon content of fly ash in pulverized coal furnace were analyzed. According to the orthogonal experiment principle, the carbonaceous characteristics of a 670t / h coal-fired boiler fly ash were tested under multi-working conditions. The BP neural network based on Levenberg-Marquardt algorithm is used to establish a neural network prediction model of carbon content in boiler fly ash. The experimental results show that the proposed algorithm not only has fast convergence rate, but also can accurately predict the boiler operating conditions in different workers Fly ash carbon content under the condition.