第10期 陈强等：基于人工免疫的故障诊断模型及其应用 。1045。 【习位耀光，郑德玲，付冬梅，等.基于生物免疫系统克隆选择机理 6结论 和免疫网络理论的免疫算法.北京科技大学学报，2005,27 仿真实验结果表明：本文提出的基于免疫原理 (2):245 I6 Costa Branco P J.Dente JA.Vilela Mends R.Using immunolo 的齿轮箱故障在线检测与诊断系统体系结构和特征 gy principes for fault detection.IEEE Trans Ind Electr,2003, 量提取方法合理、有效，对不同种类抗原进行分别处 50(2):362 理的方法可有效地帮助系统区分自己和非己模式， [7 Ishida Y.An immune net work model and its appli cations to pro 提出的进化学习机制使系统获得了令人满意的学习 cess diagmsis.Syst Comput.1993,24(6):38 L图杜海蜂，王孙安.基于AT一人工免疫网络的多级压缩机故障 效果.免疫进化学习算法表现出收敛性.所提出的 诊断.机械工程学报.2002.384)：88 利用进化学习结果和系统故障信息库知识区分和标 「9刘树林，黄文虎，夏松波，等.基于免疫机理的往复压缩机气阀 记不同故障的故障诊断方案是可行的. 故障检测方法.机械工程学报，2004.40(7：156 【10陈强，郑德玲.一种基于人工免疫的数据模式进化学习模型 参考文献 及其应用研究.计算机工程与应用，2005,41(20)：40 [1]Forrest S.Javornik B.Smith R E,et al.Using genetic ako- [11]Ishida Y.Mizessyn F.Learning algorithms on an immune ret rithms to explore pattern wecognition in the immune system.Evol work modek application to sensor diagnosis//Proceedings of In- Comput,.1993,1(3):191 temational Conference on Neural Netw orks.Wuhan,1992:33 [2]Timmis J,Neal M,Hunt J.An artificial immune system for data [12]de Castm L N.Zuben F.An evolutionary immune retwork for analysis Biosystems,2000.55(1/3):143 data clustering /Proceedings of the IEEE SBRN on Artificial [3]Forrest S,Perelson A,Sllen L,et al.Self-nonself discriminat ion Neural Netw orks.Brazil:IEEE Computation,2000:84 in a compute //Pmceedings of IEEE Symposium on Research I Se 13 Ruolph G.On a multiobective evolutionary algorithm and its curity and Privacy.Oakland:IEEE,1994:202 convergence to the pareto set //Proceedings of the 1998 IEEE [4付冬梅，郑德玲，位耀光.具有记忆与积分能力的双因子免疫 Interrational Conference on Evou tiomry Computation.Piscat- 控制器及其特性的仿真研究.北京科技大学学报。2006.28 aww:1EEE,1998:511 (2):190 Faults diagnosis model based on artificial immunity and its application CHEN Qiang,ZHENG Deling,LI Xiangping 1)Informat ion Engineering School.University of Science and Technology Beijing.Beijing 100083.China 2)Machinery and Powergenerating Equipment Engineering School Jiangxi University of Science and Technology,Ganzhou 341000.China 3)Continue Educat ion Colege,Jiangxi University of Science and Technobgy,Ganzhou 341000,China ABSTRACT A sort of system for faults detection and diagnosis based on the immunology principle was present- ed.Initial detectors were produced at random combining negative selection of selfpatterns which response nor- mal working situation of detecting objects.The learning and memory of non-selfpatterns w hich response abnor- mal working situation of detecting objects were realized using the mechanism of evolution leaning based on the artificial immune theory.The comesponding zones of different faults on states space were distinguished and marked using the results of evolution leaming and information w arehouse of faults.Regarding the set of each era antibodys mutated in the system learning as a random series,the condition of convergence of the series and a proof were presented.The algorithm's astringency was proved.Appling the method in detection and diagnosis for faults of gear case of machine tools,the ex perimental results indicate that the method is effective. KEY WORDS artificial immunity;evolution and learning:anomaly detection;fault diagnosis6 结论 仿真实验结果表明:本文提出的基于免疫原理 的齿轮箱故障在线检测与诊断系统体系结构和特征 量提取方法合理 、有效, 对不同种类抗原进行分别处 理的方法可有效地帮助系统区分自己和非己模式, 提出的进化学习机制使系统获得了令人满意的学习 效果.免疫进化学习算法表现出收敛性.所提出的 利用进化学习结果和系统故障信息库知识区分和标 记不同故障的故障诊断方案是可行的. 参 考 文 献 [ 1] Forrest S, Javornik B, Smith R E, et al.Using geneti c algo￾rithms t o explore patt ern recognition in the immune syst em .Evol Comput, 1993, 1( 3) :191 [ 2] Timmis J, Neal M, Hunt J.An artificial immune syst em for dat a analysis.Biosystems, 2000, 55( 1/ 3) :143 [ 3] Forrest S, Perelson A, Sllen L, et al.Self-nonself discrimination in a compute ∥Proceedings of IEEE Symposium on Research I S e￾curit y and Privacy .Oakland:IEEE, 1994:202 [ 4] 付冬梅, 郑德玲, 位耀光.具有记忆与积分能力的双因子免疫 控制器及其特性的仿真研究.北京科技大学学报, 2006, 28 ( 2) :190 [ 5] 位耀光, 郑德玲, 付冬梅, 等.基于生物免疫系统克隆选择机理 和免疫网络理论的免疫算法.北京科技大学学报, 2005, 27 ( 2) :245 [ 6] Cost a Branco P J, Dente J A, Vilela Mends R.Using immunolo￾gy principles f or fault det ection.IEEE Trans Ind Electr, 2003, 50( 2) :362 [ 7] Ishida Y.An immune network model and its appli cations to pro￾cess diagnosis.Syst Comput, 1993, 24( 6) :38 [ 8] 杜海峰, 王孙安.基于 ART-人工免疫网络的多级压缩机故障 诊断.机械工程学报, 2002, 38( 4) :88 [ 9] 刘树林, 黄文虎, 夏松波, 等.基于免疫机理的往复压缩机气阀 故障检测方法.机械工程学报, 2004, 40( 7) :156 [ 10] 陈强, 郑德玲.一种基于人工免疫的数据模式进化学习模型 及其应用研究.计算机工程与应用, 2005, 41( 20) :40 [ 11] Ishida Y, Mizessyn F.Learning algorithms on an immune net￾w ork model:application t o sensor diagnosis∥Proceedings of In￾ternational C onf erence on Neural Netw orks.Wuhan, 1992:33 [ 12] de Castro L N, Zuben F.An evolutionary immune netw ork for data clustering ∥Proceedings of the IEEE SBRN on Artificial Neural Netw orks.Brazil:IEEE Computation, 2000:84 [ 13] Ruolph G .On a multiobjecti ve evolutionary algorithm and its convergence to the pareto set ∥Proceedings of the 1998 IEEE International C onf erence on Evolu tionary Computation.Pisscat￾aw ay :IEEE, 1998:511 Faults diagnosis model based on artificial immunity and its application CHEN Qiang 1, 2) , ZHENG Deling 1) , LI X iangping 3) 1) Information Engineering School, University of S cience and Technology Beijing, Beijing 100083, China 2) Machinery and Pow er-generating Equipment Engineering School, Jiangxi University of Science and Technology, Ganzhou 341000, China 3) Continue Education College, Jiangxi University of S cience and Technology, Ganzhou 341000, China ABSTRACT A so rt of system for faults detection and diag nosis based on the immunology principle was present￾ed .Initial detectors w ere produced at random combining negative selection of self-patterns w hich response nor￾mal w orking situation of detecting objects .The learning and memory of non-self-patterns w hich response abnor￾mal w orking situation of detecting objects w ere realized using the mechanism of evolution leaning based on the artificial immune theory .The co rresponding zones of different faults on states space were disting uished and marked using the results of evolution learning and info rmation w arehouse of faults .Regarding the set of each era antibodys mutated in the system learning as a random series, the condition of convergence of the series and a proof were presented .The algorithm' s astringency w as proved .Appling the method in detection and diagnosis for faults of gear case of machine tools, the ex perimental results indicate that the method is effective . KEY WORDS artificial immunity ;evolution and learning ;anomaly detection ;fault diagnosis 第 10 期 陈 强等:基于人工免疫的故障诊断模型及其应用 · 1045 ·