A critical review of the most popular types of neuro control

Abstract

In this review article, the most popular types of neural network control systems are briefly introduced and their main features are reviewed. Neuro control systems are defined as control systems in which at least one artificial neural network (ANN) is directly involved in generating the control command. Initially, neural networks were mostly used to model system dynamics inversely to produce a control command which pushes the system towards a desired or reference value of the output (1989). At the next stage, neural networks were trained to track a reference model, and ANN model reference control appeared (1990). In that method, ANNs were used to extend the application of adaptive reference model control, which was a well-known control technique. This attitude towards the extension of the application of well-known control methods using ANNs was followed by the development of ANN model-predictive (1991), ANN sliding mode (1994) and ANN feedback linearization (1995) techniques. As the first category of neuro controllers, inverse dynamics ANN controllers were frequently used to form a control system together with other controllers, but this attitude faded as other types of ANN control systems were developed. However, recently, this approach has been revived. In the last decade, control system designers started to use ANNs to compensate/cancel undesired or uncertain parts of systems' dynamics to facilitate the use of well-known conventional control systems. The resultant control system usually includes two or three controllers. In this paper, applications of different ANN control systems are also addressed.Key Words: Control, neural network, adaptive, feedback linearization, predictive, model reference, perceptron, radial basis. I. ARTIFICIAL NEURAL NETWORKS USABLE IN CONTROLArtificial neural networks (ANNs) are special mathematical models inspired by human neural networks. An ANN usually includes neurons, connections and biases. Neurons are arranged in 'layers'. There are a variety of neural networks suitable for different purposes [1,2]. In neuro control, it is a difficult and unsolved problem to find the best ANN structure for each specific application; thus, a fairly large ANN is usually employed to deal with relatively complex approximation problems [3]. In this paper, the most common types of artificial neural networks in the area of control are introduced: multi layer perceptrons (MLPs) and radial basis function networks (RBFNs). Both of these are known as universal approximators of systems [4,5]. The advantage of RBFNs is their quick training process compared with MLPs; however, for complicated systems with many inputs, the number of neurons in RBFNs are usually considerably higher than MLPs [1,2]. Regardless of the main structure, if the output of an ANN depends not only on the current q

Estimate of the Head Produced by Electrical Submersible Pumps on Gaseous Petroleum Fluids, A Radial Basis Function Network Approach

“…The reason to choose an RBFN as the head-estimating model is the fact that RBFNs are mathematically proven universal approximators [21]. Inspired by existing empirical models, a single output of H m and triple inputs of p in , q m and α were opted for the RBFN model.…”

Development of a Radial Basis Function Network to Estimate the Head Generated by Electrical Submersible Pumps on Gaseous Petroleum Fluids

“…The reason to choose an RBFN as the head-estimating model is the fact that RBFNs are universal approximators with significant mathematically proven modelling capabilities [21]. Inspired by existing empirical models, a single output of H m and triple inputs of p in , q m and α were opted for the RBFN model.…”

Intelligent modeling of a piezoelectric tube actuator

“…3). The inputs to a neuron are summed and the result passes through a function, namely 'activation function' [23]. In this research, two forms of perceptrons are utilized.…”