Design of ball-beam balance control system using neural-fuzzy algorithm

Design and implementation of adaptive neuro-fuzzy inference system for the control of an uncertain ball and beam apparatus

“…For example, a driver can intuitively state the rules for implementing a cruise control in a car. Alternatively, numerical data (inputs and outputs) are collected while a human expert controls the system in real-time or using simulations [16]. Then, input-output mapping is used to establish the heuristic rules for FIS.…”

“…For discrete-time approximation of Eq. 2, we used backward difference rule (for derivation), and right-side rule (for integration) [16]. The discrete-time approximation thus obtained is given by Eq.…”

“…Obtaining an admissible mathematical model is viable for a simple system, but this is not always possible for complex cases. To bridge this gap, non-model based control-strategies like neural and fuzzy techniques were later developed [5] and implemented in a number of control applications like, cruise control [6][7][8][9], industrial processes [10][11][12], robotics [13][14] and ball and beam/plate systems [15][16][17][18][19], etc.…”

“…In this study, we tested the effectiveness of ANFIS to control a real-time system, i.e., ball and beam, similar to the earlier studies [15][16][17]. A step further, we used different balls to tune and test the controller; hence, introducing an uncertainty in the system.…”