论文部分内容阅读
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 Machine (SVM) A New Statistical Learning Method. However, as a classification algorithm, it has the disadvantages of large computation time and long running time. In order to solve the parameter selection problem of LSSVM, the concept of black hole in physics is introduced and a black hole model is established. Combined with simulated annealing algorithm, a black hole PSO (Simulated Annealing Algorithm) is proposed. This algorithm can increase the diversity of particles and overcome the problem of falling into local extremum during the optimization of PSO algorithm, which improves the optimization performance and improves the convergence property. The parameters of LSSVM are optimized and selected by using BHPSO-SA algorithm. The data of UCI database are used for classification verification. Compared with LSSVM optimized by CV parameters, the classification speed and accuracy are improved. Finally, the BHPSOSA-LSSVM algorithm is applied to the fault diagnosis of fan gearbox, and achieved good results.