Dimension optimization design of an under-restrained 6-DOF four-cable-driven parallel manipulator based on least square-support vector regression

For the design of a work platform for outside wall cleaning (WPOWC), a four-cable under-restrained parallel manipulator with 3 rotations and 3 translations (URPM4-3R3T) is adopted. The purpose of this paper is to optimize dimensions of the URPM4-3R3T to meet its workspace requirement of constraint condition in terms of cable tension. Accordingly, Pseudo-Curve model simplification is adopted in order to set up the cable tension equilibrium equations; preliminarily the workspace under a set of different dimension parameters is analyzed. As the dimensions of WPOWC are different at different local conditions, it is necessary and popular to obtain many feasible optimal dimensions instead of only one optimal solution. Therefore, the global optimization method, Least Square-Support Vector Regression (LS-SVR), is presented in this paper. In this way, many optimal solutions are provided, which allows the end user to choose the best design compromise for the problem at hand.