FOA-WPT降噪和PSO-SVM在滚动轴承故障诊断中的应用

在轴承故障诊断中，为了进一步提高诊断方法的自适应性和分类准确率，提出果蝇优化小波包降噪和粒子群支持向量机相结合的方法。利用果蝇算法对小波包降噪的阈值进行优化，结合粒子群算法在GCV算法下的错误率最低，得到SVM的最优惩罚参数和核函数参数，建立PSO-SVM分类模型，对4种工况下滚动轴承的10类故障进行分类。实验结果表明，使用FOA-WPT降噪后，信号有着更高的信噪比和更低的均方误差（MSE）；和粒子群支持向量机相结合的分类方法准确率达到89%，与未使用粒子群算法优化的SVM相比，提高了约8%，进一步证明了该方法可以实现滚动轴承的多分类故障诊断。

Applications of FOA-WPT and PSO-SVM in Faults Diagnosis of Rolling Bearing

In order to improve the self-adaptability and accuracy of faults diagnosis of rolling bearings, a method combined FOA-WPT and PSO-SVM was proposed. Fruit Fly Algorithm to optimize Wavelet Package Transformation, together with Particle Swarm Optimization with lower error-rates under GCV algorithm, gains the best punish parameter and kernel function parameter which are used to build SVM classification model to classify 10 kinds of rolling bearing faults. The experimental results show that the signal after FOA-WPT has higher Signal-Noise Ratio (SNR) and lower Mean Square Error (MSE); Besides, the accuracy of classification methods by using PSO-SVM can reach 89%, which is about 8% higher than that of traditional SVM without particle swarm optimization. All above prove that FOA-WPT combined with PSO-SVM can be applied to the faults diagnosis of rolling bearings