Experimental data set of a pneumatic levitator, PWM vs. level

The dataset used to train the GRU model was collected from the pneumatic levitation system under controlled experimental conditions. The data acquisition process involved applying a pseudo-random binary sequence (PRBS) to the fan's PWM control signal, which excited the system and generated observable variations in the levitated object's height. The dataset consists of 8,912 samples, recorded at a frequency of 10 Hz, spanning approximately 15 minutes of continuous operation.
Each sample includes two key variables: the PWM control signal (ranging from 0 to 255) and the corresponding height of the levitated object (measured in centimeters).