This project is born from the collaboration between Unibg and BFT company. BFT is a company that develops innovative systems for the automation of residential, commercial and urban contexts. The aim of the project is to automatically estimate the mass of a sliding gate from measured signals by onboard sensors. The problem is well motivated: infact, for safety reasons, regulations restrict the speed of the gate according to its mass. A heavier gate is allowed to reach lower opening and closing speeds than a lighter one. Currently, this setting is performed by a operator at the moment of the gate installation. If the system would be able to automatically detect the mass of the gate, this will translate to an improved user experience (who can set the gate by his own) and money saving by the manufacturing company, that can employ the worker on other tasks. In order for a solution to be useful, it has to be simple and fast. The methodologies should not require the installation of additional sensors, relying only on variables already measured.
To solve the problem two alternative solutions has been tested, a data-driven one based on Machine Learning classificators and a model-based one based on system identification of the parameters of a mathematical model of the motor-gate system.
A set of experimental tests have been performed on a 1:1 scale gate to validate the mass estimation algorithms.
The procedure has to be able to classify the system as belonging to one of four weight categories, namely 300kg, 400kg, 500kg, 600kg.