CAJ | 학술논문

针对遗传算法(GA)和粒子群优化(PSO)算法优化支持向量机(SVM)存在容易陷入局部最优解、诊断精度相对较低、鲁棒性较差的问题,提出了一种结合 GA、PSO、模拟退火算法的 GAPSO 优化算法,利用这种算法对SVM 的参数进行了优化,优化后的算法能够较好地调整算法的全局与局部搜索能力之间的平衡.通过对航空发动机典型故障的诊断研究表明,该方法不仅能够取得良好的分类效果,诊断精度高于 BP 神经网络、自组织神经网络、标准SVM、GA-SVM,而且有较好的鲁棒性,更适合在故障诊断中应用.
침대유전산법(GA)화입자군우화(PSO)산법우화지지향량궤(SVM)존재용역함입국부최우해、진단정도상대교저、로봉성교차적문제,제출료일충결합 GA、PSO、모의퇴화산법적 GAPSO 우화산법,이용저충산법대SVM 적삼수진행료우화,우화후적산법능구교호지조정산법적전국여국부수색능력지간적평형.통과대항공발동궤전형고장적진단연구표명,해방법불부능구취득량호적분류효과,진단정도고우 BP 신경망락、자조직신경망락、표준SVM、GA-SVM,이차유교호적로봉성,경괄합재고장진단중응용.
Genetic algorithm(GA)and particle swarm optimization(PSO)algorithm optimized support vector ma-chine(SVM)has such disadvantages as the tendency to fall into local optimal solution,relatively low diagnostic accu-racy and poor robustness. To solve the problems,an GAPSO algorithm was proposed in this paper,which combines GA,PSC and simulated annealing algorithm together and is used to optimize the parameters of SVM. It is proved that the optimized algorithm can well balance the overall search ability and the local search ability. A typical aircraft engine fault diagnosis shows that the method can achieve good classification effects,with greater diagnostic accuracy than BP neural network,adaptive neural network,the standard SVM and GA-SVM,and it has good robustness. There-fore,it is verified that the proposed algorithm is more suitable for fault diagnosis.