Sparse deconvolution methods can not only adaptively design the filter to counteract the negative effect of the transmission path but also exhibit superior performance compared to traditional deconvolution methods based on kurtosis. Yet, determining the optimal filter length remains challenging, and there is still further study to be done on how to make sparse deconvolution methods work better under heavy noise and harmonic interference. To address these issues, this paper proposes a novel sparse deconvolution technique named the maximum generalized Gini squared envelope optimized with the sparrow search algorithm (MGGSED-SSA). First, a new sparse index is constructed based on the generalized Gini index, which is less sensitive to filter length and more robust to interference. Second, the MGGSED-SSA is derived by considering the new sparse index as a deconvolution objective function. Third, the sparrow search algorithm is applied to optimize filter coefficients in the generalized spherical coordinate transformation. Simulation cases and experimental cases concerning bearing with inner and outer race faults are used to verify the effectiveness of the proposed method. It can be seen that the proposed method is less sensitive to filter length and holds better diagnosis performance even when bearings suffer from heavy noise and harmonic interference, compared with traditional methods.

中文翻译：

---

MGGSED-SSA：一种改进的用于滚动轴承诊断的稀疏反卷积方法

稀疏反卷积方法不仅可以自适应地设计滤波器来抵消传输路径的负面影响，而且与传统的基于峰度的反卷积方法相比，表现出优越的性能。然而，确定最佳滤波器长度仍然具有挑战性，如何使稀疏反卷积方法在重噪声和谐波干扰下更好地工作还有待进一步研究。为了解决这些问题，本文提出了一种新颖的稀疏反卷积技术，称为使用麻雀搜索算法优化的最大广义基尼平方包络（MGGSED-SSA）。首先，在广义基尼指数的基础上构造了一种新的稀疏指数，该指数对滤波器长度不太敏感，对干扰具有更强的鲁棒性。其次，通过将新的稀疏索引视为反卷积目标函数来导出 MGGSED-SSA。第三，应用麻雀搜索算法来优化广义球坐标变换中的滤波器系数。通过针对内外圈故障轴承的仿真案例和实验案例验证了该方法的有效性。可以看出，与传统方法相比，该方法对滤波器长度不太敏感，即使在轴承遭受严重噪声和谐波干扰的情况下也能保持更好的诊断性能。