基于ADQPSO-SVR的锅炉飞灰含碳量预测研究  

Research on Prediction of Carbon Content in Boiler Fly Ash Based on ADQPSO-SVR

作　　者：彭道刚 李丹阳 顾立群[2] 赵慧荣 PENG Dao-gang;LI Dan-yang;GU Li-qun;ZHAO Hui-rong(College of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China;Power Plant of Baoshan Iron&Steel Co.,Ltd.,Shanghai 201900,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090 [2]宝山钢铁股份有限公司电厂,上海201900

出　　处：《计算机仿真》2020年第3期72-77,共6页Computer Simulation

摘　　要：针对锅炉飞灰含碳量的预测问题,提出了自适应扰动量子粒子群优化的支持向量回归机方法(ADQPSO-SVR),即在量子粒子群优化算法(QPSO)的基础上加入自适应扰动,克服了支持向量回归机(SVR)经验选择学习参数的弊端。用此改进算法对SVR的学习参数进行寻优,经过实例研究表明,ADQPSO算法的寻优能力较强,利用ADQPSO算法得到的SVR模型有较高的预测精度,同时与GA-BP算法和GA-RBF算法相比,ADQPSO-SVR能够提高锅炉飞灰含碳量预测的准确性及稳定性。In order to predict the carbon content of boiler fly ash,an adaptive perturbation quantum particle swarm optimization Support Vector Regression(ADQPSO-SVR)method is proposed,that is,adaptive perturbation is added to the QPSO(Quality of Particle Swarm Optimization)algorithm,which can overcome the drawbacks of the experience vector selection machine(SVR)to select learning parameters.This improved algorithm was used to optimize the learning parameters of SVR.A case study shows that the ADQPSO algorithm has a strong search capability,the SVR model obtained by using the ADQPSO algorithm has a higher prediction accuracy,and compared with GA-BP algorithm and GA-RBF algorithm,ADQPSO-SVR can improve the prediction accuracy and stability of boiler fly ash carbon content.

关 键 词：锅炉飞灰含碳量 学习参数选择 自适应扰动量子粒子群 支持向量回归 

分 类 号：TK[动力工程及工程热物理]