Sensor fusion approach for shape estimation of a base-excited cantilever beam

Abstract

In smart structures, achieving a reliable set of measurement signals to monitor the system’s performance is critical. Also, the essence of using the optimum batch of sensors and an efficient algorithm to process these signals is significant for active vibration control of these structures. This paper primarily introduces a method of sensor fusion using the Kalman filter as an observer to gain the proper position signal from both an accelerometer and an ultrasonic sensor mounted on the tip of a cantilever beam. The main goal of this procedure is to eliminate both sensors’ shortcomings. Also, we present a novel approach to estimate the overall shape of the beam, using only the tip position signal. To this end, a high-speed camera is used to capture the motion of three markers on the beam under different excitation frequencies. Then, three long short-term memory networks are trained by deep learning methods, using a finite sequence of beam tip position, to act as observers for estimating the shape of the beam. The proposed methods are simulated and then validated by experiments.