摘要
针对贝叶斯网络判别学习方法在处理大数据集时,存在的模型训练时间长、算法迭代次数过多等问题,通过引入指数级参数,提出了混沌量子粒子群的权重类条件贝叶斯网络参数学习方法。该方法首先通过优化对数似然函数,解决生成学习的参数估计问题。然后,使用生成学习的结果,初始化判别学习的参数。最后,引入混沌映射序列,通过混沌量子粒子群优化(chaos quantum particle swarm optimization,CQPSO)算法,优化条件对数似然函数。使用权重类条件贝叶斯网络分类器对液体火箭发动机的故障进行分类,仿真结果表明,改进的方法分类精度高,误分类率低。同时,采用CQPSO与量子粒子群优化(quantum particle swarm optimization,QPSO)算法、标准粒子群优化(particle swarm optimization,PSO)算法相比,能够有效减少算法的迭代次数,提高算法的效率。
When dealing with big data sets, there are problems such as long model training time and too many iterations of the algorithm with the Bayesian network discriminative learning method. By introducing exponential parameters, a weighted class-conditional Bayesian network parameter learning method of chaos quantum particle swarm is proposed. The method addresses the estimation of the parameters of generative learning by optimizing the log-likelihood function. Then, it initializes the parameters of discriminative learning utlizing the result of the generative learning. Finally, the conditional log-likelihood function is optimized by constructing the chaos quantum particle swarm optimization algorithm with the chaos map sequence. The weighted class condition Bayesian network classifier is used to classify the faults of liquid rocket engines. The simulation results show that the improved method has high classification accuracy and low misclassification rate. Compared with the quantum particle swarm and the particle swarm optimization method, the chaos quantum particle swarm optimization algorithm can effectively reduce the number of iterations of the algorithm and improve the efficiency of the algorithm.
作者
刘久富
丁晓彬
郑锐
王彪
刘海阳
王志胜
LIU Jiufu;DING Xiaobin;ZHENG Rui;WANG Biao;LIU Haiyang;WANG Zhisheng(College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China;School of Electronic Science and Engineering, Southeast University, Nanjing 211189, China)
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2019年第10期2304-2309,共6页
Systems Engineering and Electronics
基金
国家自然科学基金(61473144)资助课题
关键词
贝叶斯网络
权重判别参数学习
量子行为粒子群
混沌映射序列
Bayesian network
weighted discriminative parameter learning
quantum behave particle swarm
chaos map sequence
