<p>Aiming at the challenge of difficulties in the feature extraction and the fault detection of bearings in strong noise environments, a bearing fault detection method based on improved sparrow search algorithm (ISSA) optimized variational mode decomposition (VMD) and maximum correlation kurtosis deconvolution (MCKD) is proposed in this paper. The VMD method is adopted to decompose the vibration signal of the bearing to obtain the optimal modal components. The MCKD method is used to enhance the fault impact components in the decomposed optimal modal components, and the envelope analysis is utilized to achieve precise detection of bearing faults. However, the values of the key parameters of VMD and MCKD have a significant impact on the signal processing and the effect of fault feature enhancement. To address the issues of the standard SSA being prone to entrapment in local optima and exhibiting a slow convergence rate, an ISSA is proposed by integrating Chebyshev chaotic mapping with a uniform distribution and a random walk strategy. Comparative experiments revealed that the proposed method effectively avoids local optima and achieves superior performance in terms of convergence speed compared to the standard sparrow search algorithm (SSA), particle swarm optimization (PSO), and genetic algorithm (GA). The key parameters of VMD and MCKD are determined using the proposed ISSA. Both simulation and experimental results demonstrate that the proposed method is capable of accurately extracting and enhancing fault features under varying signal-to-noise ratio (SNR) conditions. The obtained envelope spectra clearly reveal that the characteristic frequencies of the fault and their multiples are quite obvious. Furthermore, comparative experiments validate the superiority of the proposed method in different noise environments across a range of SNRs, enabling effective fault detection of bearings even in high-noise scenarios.</p>