CSAPSO-BPNN-Based Modeling of End Airbag Stiffness of Nursing Transfer Robot

Abstract

The use of nursing transfer robots is a vital solution to the problem of daily mobility difficulties for semi-disabilities. However, the fact that care-receivers have different physical characteristics leads to force concentration during human–robot interaction, which affects their comfort. To address this problem, this study installs an array of double wedge-shaped airbags onto the end-effector of a robot, and analyses airbag mechanical properties. Firstly, this study performed the mechanical testing and data collection of the airbag, including its external load and displacement, at various gas masses. Then, the performance of the Back Propagation (BP) neural network is improved using chaos (C) theory and simulated annealing particle swarm optimization (SAPSO), resulting in the establishment of the CSAPSO-BP neural network. By this method, a fitting model is developed to determine the mechanical parameters of the wedge-shaped airbag stiffness, and the fitting relation of external load–displacement is obtained. Data analyses show that the wedge-shaped airbag stiffness increases quadratically, linearly, and with a constant rate as the gas mass increases. The airbag stiffness regulation and model describe its three distinct phases with quadratic, linear, and linear invariant characteristics as the gas mass changes. These findings contribute to the structural optimization of airbags.