To address the limitations of traditional rolling bearing fault diagnosis methods, which suffer from insufficient fault feature extraction and limited diagnostic accuracy due to single-channel vibration data analysis, this study proposes a novel multivariate fault signal diagnosis approach that combines Multivariate Variational Mode Decomposition based on improved artificial bee colony algorithm (IABC-MVMD) with Refined Composite Multivariate Multiscale Fuzzy Entropy (RCMMFE). To overcome the challenges posed by the difficulty in self-tuning the decomposition layers and bandwidth balancing parameters during MVMD, an improved artificial bee colony algorithm (IABC) which integrated Chebyshev chaotic mapping, elite information guidance, and adaptive inertia weight is designed. The algorithm effectively satisfies the adaptive partitioning requirements of multivariate excitation signals in the frequency domain. On this basis, RCMMFE is utilized to construct a high-dimensional fault feature set that can comprehensively characterize bearing states, and then these features are input into a support vector machine for fault diagnosis. Through simulation validation on the CWRU dataset and comparative analysis with existing methods, the results show that the proposed method can extract and identify the multivariate fault signal features of rolling bearings efficiently and accurately, thus exhibiting significant theoretical value and practical implications.