摘要
针对装备诊断与测试实际过程中普遍存在的不确定性问题,通过引入测试不确定概率,建立基于贝叶斯网络的测试性分析模型。在此基础上获取测试不确定条件下的故障-测试相关性矩阵,经贝叶斯推理计算测试性指标参数,建立测试项目集优化模型,并利用混合二进制粒子群-遗传算法进行求解。案例验证表明,该分析与计算过程由于考虑了测试不确定性,使得结果与实际情况更加吻合,与传统的确定性优化方法相比具有更高可信度。
Synthetically considering the ubiquitous uncertainties of the process of materiel diagnosis and testing,a model of testablitiy analysis was founded referring to the Bayesian network by introducing test probabilities.Based on the corresponding dependency matrix between faults and tests under the condition of test uncertainties,testability indices were computed via Bayesian reasoning,and then the test set was optimized using hybrid BPSO(binary particle swarm optimization) and GA(genetic algorithm).The experiments show that the conclusions are consistent with the actual situations because of the test uncertainties considered,consequently have higher confidence comparing with traditional methods.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2011年第4期379-384,共6页
China Mechanical Engineering
基金
"十一五"部委预研基金资助项目(51317040102)
关键词
测试性方案
测试选择
测试不确定性
贝叶斯网络
优化
testability scheme
test selection
test uncertainty
Bayesian network
optimization