Neural Network-Based Adaptive Dynamic Surface Control for Inverted Pendulum System

Intelligent robust PI adaptive control strategy for speed control of EV(s)

“…Note that the free parameters of LS-SVR systems (17) and (18) are adapted from the beginning, whereas the free parameters of the PI structure are adapted when the state enters the boundary layer region. The asymptotical stability of the control scheme is demonstrated and proved in Theorem 2.…”

“…Theorem 2: If the control law in (23) is applied to the uncertain non-linear system in the form of (1) or equivalently (18), which considered the Assumptions (1)-(3) with the following adaptation laws…”

“…Despite the positive features of fuzzy systems, large number of fuzzy rules emerged when fuzzy systems are used for approximating high-order systems, i.e. real applications, and in multiple input multiple output systems [7,10,18,24]. Hence, the control scheme based on fuzzy approximation approach will be complex and high computation analyses are consumed.…”

“…Hence, the control scheme based on fuzzy approximation approach will be complex and high computation analyses are consumed. On the other hand, neural networks are alternative approaches to deal with the uncertainties, due to their universal approximation capability [6,18,25]. In neural adaptive control, direct and indirect forms, neural networks are used to discover the unknown uncertainties to save the time and effort spent on modelling complex systems.…”

“…Adaptive fuzzy controllers which use fuzzy systems as models of the plants are indirect adaptive fuzzy controllers, i.e. they use fuzzy IF-THEN rules describing the plant [14,17,18]. Wang [17] proposed a general control approach based on Lyapunov theory to design a stable adaptive controller with the supervisory controller.…”