基于综合优化方法的风力发电机故障诊断  

Parameters Optimization of LSSVM and Application in Fault Diagnosis of Wind Power Gearbox

作　　者：焦斌[1] 徐志翔[1] 

机构地区：[1]上海电机学院电气学院,上海200240

出　　处：《控制工程》2012年第4期681-686,共6页Control Engineering of China

基　　金：上海市基础研究重点项目(10JC1405800);上海市教委重点学科(J51901);上海市经委科研项目(09A118)

摘　　要：支持向量机(SVM)一种新型的统计学习方法。但是作为分类算法,它存在计算量大、运行时间长的缺点。针对LSSVM的参数选择问题,引入物理学中的黑洞概念,建立黑洞模型,结合模拟退火算法,提出了黑洞粒子群-模拟退火算法(BH-PSOSA)。该算法可以增加粒子的多样性,克服PSO算法优化过程中陷入局部极值的问题,提高了优化性能,改善了收敛特性。利用BHPSO-SA算法对LSSVM的参数进行优化选择,用UCI数据库的数据进行分类验证,相比CV参数优化的LSSVM,提高了分类速度和精度。最后把BHPSOSA-LSSVM算法应用到风机齿轮箱的故障诊断中,取得了良好的效果。Support Vector Machines （SVM） is a kind of novel statistics learning method. SVM＇ s classification algorithm is an important application of it. But as a classification algorithm, it has the weakness such as a large mount of calculation and long running time. In allusion to the parameter optimization of LSSVM, the paper introduces the concept of black hole, builds the model of black hole and combines it with SA, and proposes the BHPSO-SA algorithm. This new algorithm can increase the particles diversities and overcome it stacking in the local optimum, so it can improve the optimization performance. We optimize the parameters of LSSVM using BHPSOSA, and carry out the experiments on the UCI database with it. Experiment results indicate that this method can efficiently solve the problem of LSSVM parameters optimization. Compared with CV-LSSVM, BHPSOSA - LSSVM can improve precision of classification and get a better classification performance. Finally, we use BHPSOSA-LSSVM algorithm in the fault diagnosis of wind power gearbox and have a good result.

关 键 词：LSSVM 参数优化 BH—PSOSA 故障诊断 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]