基于贝叶斯统计推理的故障定位实验研究.pdfVIP

计算机研究与发展 ISSN N 111777TP J ournal of Comp ut er Research and Develop ment 47 (4) : 7077 15 , 20 10 基于贝叶斯统计推理的故障定位实验研究 1 1 1 1 1 2 柳永坡 　吴 　际 　金茂忠 　杨海燕 　贾晓霞 　刘雪梅 1 (北京航空航天大学计算机学院　北京 　100 19 1) 2 (北京城市学院　北京 　100083) (liuypo @sei . buaa . edu . cn) Experimentation Study of BBNBased Fault Local ization Liu Yongpo 1 , Wu J i1 , J in Maozhong1 , Yang Haiyan1 , J ia Xiao xia1 , and Liu Xuemei2 1 ( ) S chool of Comp ute r S cience an d Technology , B ei hang Uni ve rs i ty , B eij i ng 100 19 1 2 ( ) B eij i ng Ci ty Uni ve rs i ty , B eij i ng 100083 Abstract 　Fault localization t echnique s help p ro grammer s find out t he location s and t he cau se s of t he f ault s an d accelerat e t he debu ggin g p roce ss. The relation bet ween t he f ault and t he f ailure i s u sually co mp licat ed , making it har d to deduce how a fault cau se s t he f ailure . At p re sent , analy si s of variance i s broadly u sed in many recent correlative re searche s. In t hi s p ap er , a Baye sian belief net wor k (BBN) for f ault rea soning i s con st ruct ed ba sed on t he su spiciou s p at t er n , w ho se no de s con si st of t he su sp iciou s p at t er n and t he caller s of t he met ho d s t hat con stit ut e t he su sp iciou s p at t er n . The co n st r ucting algorit hm of t he BBN , t he cor relative p robabilitie s , an d t he for mula for t he co nditional p robabilitie s of each arc of t he BBN are defined . A rea sonin g algorit hm ba sed on t he BBN i s p rop o sed , t hrough