To address the low computational efficiency and accuracy for reliability optimization of flexible mechanisms, the mean-probability decomposition-coordination-based extreme support vector machine regression (MPDC-ESVR) method based on the dynamics and uncertainty is proposed. Firstly, a dynamical model is built to calculate the maximum deformation. Then, the ESVR is used as a surrogate model to study the reliability of the flexible mechanisms. Finally, the MPDC are adopted to optimize the weight of the flexible mechanisms considering multiple failure modes. It is proved that the computational efficiency and accuracy of the present methodology are superior to those of Monte Carlo simulation (MCS) and quadratic polynomial-Monte Carlo simulation (QP-MCS) through exemplification with a two-link flexible manipulator. The results indicate the two-link flexible manipulator optimized by current MPDC is lighter than that by the equal distribution probabilistic method and the original manipulator, which is essential for low cost-effectiveness without compromise to safety.

针对柔性机构可靠性优化计算效率低、精度低的问题，提出了一种基于动力学和不确定性的基于均值-概率分解-协调的极限支持向量机回归(MPDC-ESVR)方法。首先，建立动力学模型来计算最大变形。然后，将 ESVR 作为替代模型来研究柔性机构的可靠性。最后，考虑多种故障模式，采用 MPDC 来优化柔性机构的权重。以双连杆柔性机械手为例，证明该方法的计算效率和精度优于蒙特卡罗模拟（MCS）和二次多项式-蒙特卡罗模拟（QP-MCS）。