Dexterous workspace optimization for a six degree-of-freedom parallel manipulator based on surrogate-assisted constrained differential evolution

Parallel manipulators are increasingly applied because of their high stiffness, load capacity, and motion precision. However, the workspace of a parallel manipulator is relatively small compared to that of a serial manipulator. Therefore, appropriate structural designs of parallel manipulators for maximizing the workspace are essential. This paper studies the dexterous workspace maximization problem of a six-degree-of-freedom (DOF) parallel manipulator. First, a constrained optimization model is formulated, in which, the objective is to maximize the dexterous workspace, and the constraints include several kinematic and dynamic indicators, and some general structural limitations. To address the highly constrained, computationally-expensive problem, a surrogate-assisted two-phase constrained differential evolution method is proposed. Its effectiveness and efficiency are verified by the superior performance over two surrogate-free evolutionary methods: differential evolution (DE) and genetic algorithm (GA), and two surrogate-assisted state-of-the-art evolutionary methods: mod-ified constrained expected improvement (CEI) and the lower confidence bounding approach (PCLCB), on ten benchmark functions and the established real-word 6-DOF manipulator design optimization problem. & COPY; 2023 Elsevier B.V. All rights reserved.