基于改进粒子群优化支持向量机的汽轮机组故障诊断

基于支持向量机（SVM）在核函数参数和惩罚因子人为选取的盲目性以及传统粒子群算法（PSO）后期易陷于局部最小值的不足,提出了一种改进的粒子群算法（MPSO）,建立了汽轮机组振动故障诊断模型并且利用故障数据进行了模式识别.结果表明：模型能够对SVM相关参数自动寻优,并且能达到较为理想的全局最优解;与PSO-SVM和GA-SVM算法相比,MPSO-SVM算法在收敛速度和准确率方面都有所提高.

Fault Diagnosis of Steam Turbine Based on MPSO-SVM Algorithm

To overcome the blindness of artificial selection for nuclear function parameters and penalty fac tors by Support Vector Machine （SVM） as well as the deficiency of easily falling into local minimum at lat er stage of traditional Particle Swarm Optimization （PSO）, a vibration fault diagnosis model has been es tablished for steam turbine based on a newly proposed modified particle swarm optimization （MPSO） algo rithm, with which pattern recognition is performed to realize automatic optimization on relevant SVM pared with PSO SVM and GA-SVM method, the and higher classification accuracy. using fault data. Results show that the model can help parameters and achieve global optimal solution. ComMPSO-SVM algorithm has a faster convergence speed