基于KPCA及改进蚁群遗传算法的齿轮箱故障诊断

Fault Diagnosis of Gearbox Based on KPCA and Improved Ant Colony Genetic Algorithm

鉴于齿轮箱系统的复杂性和齿轮箱故障信号的多样性,提出一种基于KPCA和改进蚁群遗传算法(LACG)相结合的齿轮箱故障诊断新方法。通过KPCA去掉原始故障参数集中的冗余信息,再利用IACG算法找出降维后参数的最优解。IACG算法改进了传统蚁群算法中的概率转移公式,通过增加区域目标函数值,提高了转移运算的效率和准确率;LACG算法增加了局部搜索功能,计算得到的蚂蚁解与遗传算法的均匀两点交叉算子相结合,减少了算法的搜索时间,扩大了搜索空间,使得收敛效果更趋近最优解。实验结果表明,KPCA与IACG相结合的算法可以有效识别齿轮箱故障,相对于传统的蚁群算法,其运算效率和准确率有很大提高。

Because of the complexity of gearbox system and the diversity of gearbox fault signal, a new method based on KPCA combined with the improved ant colony genetic algorithm （IACG） is proposed. Firstly, redundant information of the raw fault parameters is removed by using KPCA. Then the optimal solution of dimension-re- duced parameters has searched by IACG. IACG is an improved algorithm by increasing the regional objective function value to improve the probability-transfer formula. The algorithm improves the efficiency and accuracy of the transfer operation, further more, it increases the local search function which combines the ant solutions with the uniform two-point crossover operator algorithm. The improved IAGG can reduce the searching time, expand the searching space and make the convergence effect get close to the optimal solution. The experimental results show that the algorithm combined KPCA with IACG can effectively identify the gearbox fault. Compared to the traditional ant colony algorithm, the computational efficiency and accuracy of the new method are higher.