학술논문
Backpropagation Neural Network Based Design of a Novel Sit-to-Stand Exoskeleton at Seat-Off Position for Paraplegic Children
Document Type
Conference
Source
2020 6th International Conference on Control, Automation and Robotics (ICCAR) Control, Automation and Robotics (ICCAR), 2020 6th International Conference on. :546-552 Apr, 2020
Subject
Language
Abstract
In this work, a novel sit-to-stand exoskeleton for paraplegic children is modeled in SolidWorks having a provision of different heights with constant body weight (40kg). A simplified mathematical formulations are, further, presented to find knee joint and foot torques at seat-off position during sit-to-stand motion. The centre of mass distances from knee joint and foot are calculated using SolidWorks model by changing children heights (95cm-180cm). Thereafter, two backpropagation neural network models (BPNN-I and BPNN-II) are designed to predict the centre of mass distances and joint torques in case of knee and foot. From the results, it is observed that both neural network models show potential conformity of predicted outputs as compared to simulated ones. The absolute percentage error for the predicted outputs is found to be minimal (