论文部分内容阅读
利用支持向量回归机(SVR)对航空发动机滑油中金属含量进行预测时,通常利用粒子群算法优化支持向量回归机中的参数。而随着迭代的深入,可能出现粒子陷入局部最优的情况。通过建立粒子散射模型对这部分粒子进行重定位,使之快速跳出局部最优。引入松弛系数p,对惯性参数ω进行调节,使整个算法快速收敛。仿真实验表明,算法有助于粒子收敛于全局最优点,提高了滑油中金属含量的预测精度。
When using SVR to forecast the metal content of aeroengine oil, the particle swarm optimization algorithm is usually used to optimize the parameters in the support vector regression machine. With iterations going deeper, it may happen that particles fall into local optimum. Through the establishment of particle scattering model of this part of the particle relocation, so that it quickly jump out of the local optimum. The relaxation coefficient p is introduced to adjust the inertial parameter ω so that the whole algorithm converges rapidly. Simulation results show that the proposed algorithm can help the particle converge to the global optimum and improve the prediction accuracy of the metal content in the oil.