论文部分内容阅读
为解决置信规则库中现有参数学习方法主要是串行算法且不适用于求解大数据下参数优化模型的问题,结合群智能算法中的差分进化算法和集群系统中分布式方法,提出了基于消息传递接口的并行参数学习方法。以输油管道检漏问题为例,对比分析了本算法与现有参数学习方法在收敛时的误差,并在不同结点数的集群系统中分析了本算法的加速比和效率。实验结果表明,并行的参数学习方法是有效可行的。
In order to solve the problem that the existing learning methods in the rule base of the confidence rule are mainly serial algorithms and are not suitable for solving the parameter optimization model under big data, combining the differential evolution algorithm in the group intelligent algorithm and the distributed method in the cluster system, Message passing interface parallel parameter learning method. Taking the problem of oil pipeline leak detection as an example, the error of this algorithm and the existing parameter learning method in convergence are comparatively analyzed. The speedup and efficiency of this algorithm are analyzed in the cluster system with different node numbers. The experimental results show that the parallel parameter learning method is effective and feasible.