摘要
为更加精确地建立火电厂循环流化床锅炉NO_x排放质量浓度的预测模型,提出了一种利用量子位Bloch坐标球面生成编码的自适应量子灰狼算法(AQGWO)来优化快速学习网(FLN)的模型。将AQGWO与差分、粒子群等算法的优化能力进行比较,进而验证该算法优化的精度和收敛速度。在不同工况下实时采集某火电厂300 MW循环流化床锅炉的实验数据,在相同条件下将AQGWO-FLN模型与利用其他算法优化的FLN模型、基本FLN模型的预测结果进行对比。结果表明:利用AQGWO-FLN的模型具有最好的预测精度和泛化能力,可有效准确地预测火电厂锅炉的NO_x排放质量浓度。
To accurately set up aprediction model for NOxemission from circulating fluidized bed boilers,an adaptive quantum grey wolf optimization（AQGWO）algorithm was proposed based on Bloch sphere coordinates to optimize the fast learning network（FLN）,of which the optimization accuracy and convergence rate were verified by comparing the AQGWO with differential evolution algorithm and particle swarm optimization.Using the experimental data of a 300 MW circulating fluidized bed boiler obtained under different working conditions,the prediction results of AQGWO-FLN model were contrasted with those of basic FLN model and the FLN models optimized by other algorithms.Results show that the AQGWO-FLN model has good generalization ability and high prediction accuracy,which may be used to predict the NOx emission from coal-fired boilers effectively and accurately.
作者
牛培峰
史春见
刘楠
马云鹏
吴志良
李进柏
NIU Peifeng;SHI Chunjian;LIU Nan;MA Yunpeng;WU Zhiliang;LI Jinbai(Key Lab of Industrial University,Computer Control Engineering of Hebei Province, Yanshan Qinhuangdao 066004, Hehei Province, Chin)
出处
《动力工程学报》
CAS
CSCD
北大核心
2018年第4期278-285,315,共9页
Journal of Chinese Society of Power Engineering
基金
国家自然科学基金资助项目(61573306
61403331)
关键词
循环流化床锅炉
NOx排放质量浓度
灰狼算法
快速学习网
Bloch坐标
circulating fluidized bed boiler
NOx emission concentration
grey wolf optimization algorithm
fast learning network
Bloch coordinates
