Mechanism Design Optimization through CAD-Based Bayesian Optimization and Quantified Constraints

TL;DR

The main research results demonstrate that the proposed CAD-based Bayesian optimization framework can effectively identify optimal design parameters that minimize the root mean square (RMS) torque while adhering to specified static and dynamic constraints.

Abstract

This research delves into optimizing mechanism design, with an emphasis on the energy efficiency and the expansive design possibilities of reciprocating mechanisms. It investigates how to efficiently integrate Computer-Aided Design (CAD) simulations with Bayesian Optimization (BO) and a constrained design space, aiming to enhance the design optimization process beyond the confines of traditional kinematic and dynamic analysis. The study sets out to create a novel optimization framework that merges CAD simulations with a BO strategy. Initially, the feasibility of a mechanism design is assessed through CAD-motion simulations, which gauge its practicality. Upon deeming a design feasible, an evaluation via CAD-motion simulations is conducted to ascertain the objective value. This research proposes utilizing non-parametric Gaussian processes for crafting a surrogate model of the objective function, considering the design space's static and dynamic constraints. The findings reveal that the introduced CAD-based Bayesian Optimization framework adeptly identifies optimal design parameters that minimize root mean square (RMS) torque while complying with predetermined constraints. This method markedly diminishes the complexity seen in analytical approaches, rendering it adaptable to intricate mechanisms and practicable for machine builders. The framework evidences the utility of integrating constraints in the optimization process, showing promise for attaining globally optimal designs efficiently. A case study on an emergency ventilator, with three design parameters, demonstrates a 71% RMS torque reduction after 255 CAD-based evaluations, underscoring the approach's effectiveness and its potential for refining mechanism design optimization.

Figures (18)

• Figure 1: The use case within this research is an emergency ventilator developed by Gear Up Medical VZW Herregodts2019.
• Figure 2: The mechanism shown in its ending $\delta_e$ (left) and starting position $\delta_i$ (right) of its point-to-point movement.
• Figure 3: The emergency ventilator performs the same task, being the end-effector movement $\delta(t)$ (right), for different combinations of design parameters (DPs) $|OA|$, $|AB|$, and $|BC|$ (left). Resulting in a difference in the motor torque profile $T_m(t)$ (right).
• Figure 4: Three essential motion simulations and measurements are required to determine the driving torque for a specific design.
• Figure 5: Choosing design parameters lengths ($|OA|$, $|AB|$, and $|BC|$) that cause bars OA and AB to align in a straight line from point B to O creates an unsolvable simulation when driving the end-effector from point C.