摘要
针对实际化工生产过程中故障数据缺乏,采用适合小样本问题的支持向量机(SVM)对化工过程稳态故障进行诊断。为了保证在线故障诊断的实时性,消除高维监控数据以及系统噪声对故障诊断的干扰,提出了一种新的基于二进制量子粒子群优化(BQPSO)算法和SVM的故障特征选择方法。仿真实验表明:BQPSO算法具有良好的全局搜索能力,能够快速、准确地搜索到故障特征变量;而基于特征选择的SVM故障诊断方法能可靠地实现对复杂化工过程的在线故障诊断。
Considering fault data are absent in the real chemical production process, this paper utilizes support vector machines （SVM） which fits the small sample problems to diagnose the chemical process steady faults. To ensure the real-time capability of online diagnosis and eliminate the disturbances from higher dimensional monitored data as well as system noises, a novel fault feature selection method based on binary quantum particle swarm optimization （BQPSO） and SVM is proposed. The results of simulation prove that BQPSO can find the global optima effectively and select the fault features quickly and exactly. And, the fault diagnosis method based on SVM with feature selection can reliably diagnose the faults online in the complex chemical process.
出处
《华东理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第5期692-696,共5页
Journal of East China University of Science and Technology
