论文部分内容阅读
针对大型电站锅炉空气预热器受热面积灰状况进行了分析研究。应用3层神经网络构建了300MW电站锅炉空气预热器受热面积灰监测数学模型,选择锅炉负荷、烟气差压、排烟温度等参数作为输入向量,以反映空气预热器积灰状况的污染系数作为输出向量,利用电厂DCS系统采集的机组实时数据,经规格化处理后作为样本集对网络进行训练。训练过程中,通过添加动量项和变步长改进了BP算法。将建立的模型应用于华电国际青岛发电公司#2炉的空气预热器在线积灰监测,取得了较好的结果。
Aiming at the gray area condition of heating area of air preheater in large power plant boiler, Based on the three-layer neural network, the mathematic model of the monitored area of the air preheater in the 300 MW power plant boiler is built. The boiler load, flue gas pressure and exhaust gas temperature are selected as input vectors to reflect the pollution of the air preheater Coefficient as an output vector, the use of power plant DCS system acquisition unit real-time data, after the normalized processing as a sample set of network training. During training, the BP algorithm was improved by adding momentum and variable steps. The established model was applied to online fouling monitoring of air preheater in # 2 furnace of Huadian Power International Qingdao Power Generation Company, and achieved good results.