摘要
针对扩展置信规则库(extended belief rule base,EBRB)系统在不一致的激活规则过多时推理准确性不高的问题,引入带精英策略的快速非支配排序遗传算法(NSGA-Ⅱ),提出一种基于NSGA-Ⅱ的激活规则多目标优化方法。该方法首先将激活权重大于零的规则(即激活规则)进行二进制编码,把最终参与合成推理的激活规则集合的不一致性以及激活权重和作为多目标优化问题的目标函数,通过带精英策略的快速非支配排序遗传算法求解不一致性更小的激活规则集合,从而降低不一致激活规则对于EBRB系统推理准确性的影响。为了验证本文方法的有效性和可行性,引入非线性函数和输油管道检漏实例进行测试。实验结果表明,基于NSGA-Ⅱ的扩展置信规则库激活规则多目标优化方法能够有效提高EBRB系统的推理能力。
To address the low reasoning accuracy of extended belief rule-base(EBRB)systems with too many inconsistent activated rules,this paper introduces a fast elitist non-dominated sorting genetic algorithm(NSGA-II)and proposes a rule activation multi-objective optimization approach based on the NSGA-II algorithm.In this approach,binary coding is carried out for the activated rules whose activation weights are greater than zero.The inconsistent set of activated rules following synthetic reasoning and the sum of activation weights are taken as the objective function of the multi-objective optimization problem.Using the fast elitist non-dominated sorting genetic algorithm,the problem of a set of activation rules with a small inconsistency is solved,reducing the effect of the inconsistent activated rules on the reasoning accuracy of EBER systems.To validate the efficiency and feasibility of the proposed method,this paper introduces a nonlinear function and the proposed method was tested against the leak detection of an oil pipeline.The experimental results show that the rule activation multi-objective optimization approach based on NSGA-II can effectively improve the reasoning performance of EBRB systems.
作者
林燕清
傅仰耿
LIN Yanqing;FU Yanggeng(College of Mathematics and Computer Science,Fuzhou University,Fuzhou 350116,China)
出处
《智能系统学报》
CSCD
北大核心
2018年第3期422-430,共9页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(71501047,61773123)
福建省自然科学基金项目(2015J01248).
关键词
扩展置信规则库
不一致性
激活规则
多目标优化
NSGA-II算法
extended belief rule base(EBRB)
inconsistency
activation rules
multi-objective optimization
NSGA-II algorithm
