基于参数联合优化VMD-SVM的工业机器人旋转部件故障诊断方法

针对因工业机器人旋转部件故障诊断模型最优参数难以自适应确定导致故障识别率低的问题，提出了一种参数联合优化的VMD-SVM的工业机器人旋转部件故障诊断方法；提出了一种基于遗传变异的改进灰狼算法，该算法采用Logistic混沌映射进行种群初始化，将非线性因子引入位置更新公式，并利用遗传变异策略解决算法陷入局部最优时的停滞现象；基于该算法对VMD和SVM进行参数联合优化；利用参数优化的VMD对故障信号进行分解，对所得的本征模态函数计算改进样本熵以构成特征向量，再输入至参数优化的SVM完成工业机器人旋转部件的故障诊断；仿真和实验结果表明，本文方法能够准确地进行故障诊断，在信号无噪和含噪的条件下准确率最高均达100%，较EMD、LMD、DTCWT、VMD等四种方法具有更优的指标。

Fault Diagnosis for Rolling Parts of Industrial Robot Based on Parameter Collaborative Optimization VMD-SVM

Aiming?at?the?low?diagnostic?recognition?rate?caused?by?the difficulty in determining optimal parameters of fault diagnosis model adaptively for industrial robot rolling parts, a fault diagnosis method based on parameter collaborative optimization variational?mode?decomposition?(VMD) - support?vector?machine (SVM) is proposed. An improved grey wolf optimization based on genetic variation is proposed. In this algorithm, a logistic chaotic map is adopted in population initialization, a nonlinear convergence factor is introduced in updating the location of grey wolf, and a genetic variation strategy is used to solve the stagnation phenomenon when the algorithm is stuck in the local optimum. The algorithm is used to optimize the parameters of VMD and SVM collaboratively. Fault?signals are decomposed into intrinsic mode functions (IMF) by?the parameter?optimization?VMD method, and the improved sample entropy of these IMFs are calculated to form feature vectors, which are then brought to SVM for fault diagnosis for rolling parts of an industrial robot. The?simulation?results?show that the proposed method is effective in fault diagnosing, with the accuracy up to 100% under the condition of both noised and noiseless signal, which is superior than the accuracy of other methods such as empirical mode decomposition (EMD), local mean decomposition (LMD), Dual-tree?complex?wavelets (DTCWT) and VMD.