A novel adaptive control design method for stochastic nonlinear systems using neural network

Abstract

This paper presents a novel method for designing an adaptive control system using radial basis function neural network. The method is capable of dealing with nonlinear stochastic systems in strict-feedback form with any unknown dynamics. The proposed neural network allows the method not only to approximate any unknown dynamic of stochastic nonlinear systems, but also to compensate actuator nonlinearity. By employing dynamic surface control method, a common problem that intrinsically exists in the back-stepping design, called “explosion of complexity”, is resolved. The proposed method is applied to the control systems comprising various types of the actuator nonlinearities such as Prandtl–Ishlinskii (PI) hysteresis, and dead-zone nonlinearity. The performance of the proposed method is compared to two different baseline methods: a direct form of backstepping method, and an adaptation of the proposed method, named APIC-DSC, in which the neural network is not contributed in compensating the actuator nonlinearity. It is observed that the proposed method improves the failure-free tracking performance in terms of the Integrated Mean Square Error (IMSE) by 25%/11% as compared to the backstepping/APIC-DSC method. This depression in IMSE is further improved by 76%/38% and 32%/49%, when it comes with the actuator nonlinearity of PI hysteresis and dead-zone, respectively. The proposed method also demands shorter adaptation period compared with the baseline methods.

Designing of 2‐degree of freedom load frequency controller for power system using novel improved pole clustering and genetic method of reduced‐order modelling

“…The compensator for this system is to be designed by using the proposed IPCG algorithm. The transfer function of the system obtained from state equation and output equation is given as shown in Equations ( 47) and (48). performing optimization, using the value of optimization parameters as given in Table 6 after 48 iterations.…”

“…There are some commonly noticed nonlinearities such as backlash or governor dead band, 47 time delay 48 and generation rate constraint 49 that may occurred at any instant of time and these may deteriorate the performance of the power system. So, these nonlinearities are necessary to tackle while implementing LFC.…”

An improved neuro‐adaptive‐optimal control for oscillation damping in wind‐integrated power systems

“…17 Various neural network-based controllers were used during the last decade and fairly good results were reported. [18][19][20][21] A diagonal recurrent neural network is proposed, 18 with one-hidden layer comprising of ten non-linear neurons. However, with these ten non-linear neurons and sixty-one connection weights, the control structure becomes complex and adds computation burden to the system.…”

“…Although, back-stepping technique is very useful, this control strategy faces "explosion of complexity" problem as the order of the system increases. 16,20 Neural-network-based control strategies have also been reported to replace the CPSS so as to effectively damp the low-frequency electromechanical oscillations under varying operating conditions of the system. [22][23][24][25][26][27][28][29][30][31] However, these control strategies lack some of the important attributes required for practical realizability of an adaptive controller.…”

“…• For on-line applications, as the number of parameters (equal to the number of connections) in the neural networks is increased, the parameters to be updated online in every time-step will grow and the learning process will be slow, thus reducing the real-time feasibility of the neuro-control. 20,32,33 • Slow convergence could lead the training algorithms to get trapped at local optima, resulting in suboptimal solutions. • Memory-bandwidth and floating-point operations as the resources are limited in practical applications of complex non-linear neural networks.…”