Vertical force calibration of smart force platform using artificial neural networks

Abstract

INTRODUCTION: The human body may interact with the structures and these interactions are developed through the application of contact forces, for instance due to walking movement. A structure may undergo changes in the dynamic behaviour when subjected to loads and human bodies. The aim of this paper is to propose a methodology using Artificial Neural Networks (ANN) to calibrate a force platform in order to reduce uncertainties in the vertical Ground Reaction Force measurements and positioning of the applied force for the human gait. METHODS: Force platforms have been used to evaluate the pattern of applied human forces and to fit models for the interaction between pedestrians and structures. The designed force platform consists in two force plates placed side by side in the direction of walking. The reference voltages applied to the Wheatstone bridge were used for calibration as the input data to the ANN, while the output data were the estimated values of the standard weights applied to the force platform. RESULTS: It was presented a framework to enhance traditional calibration methods for force platforms (vertical component) using an ANN. The use of ANN shows significant improvements for the measured variables, leading to better results with lower uncertain values that are smaller than those using a simple traditional calibration. CONCLUSION: The results suggest that the calibration with the ANN method may be useful in obtaining more accurate vertical Ground Reaction Forces and positioning measurements in a force platform for human gait analysis.