“…x; =W;,2Uk_i + y~(k -1) + y; (k -1) Neural networks are applied for inverse control of nonlinear systems in many works [22][23][24][25][26]. In that control, the NN model is applied to identify the plant input and then to generate the new input of the plant.…”
Section: Rbf Network Inverse Controlmentioning
confidence: 99%
“…Here, the identification MSE of u(t) is 8e-3 and tracking MSE is 0.18. where GO is the dynamic inverse of the plant [22]. In Fig.…”
I selami.bevhan@ege.edu.tr , 2 musa.alcirmege.edu.tr
Radial Basis Function NetworksRadial basis function neural networks (RBFNNs) are the one of the different functionalized type of NNs with high approximation and regularization capability [3]. The RBFs are preferred as the basic structure of neural networks because of their good local specialization and global generalization ability [4]. The design of a RBFN in its most basic form consists of three separate layers. The first layer is the input layer. The second layer is the hidden layer and it is structured with high dimension to provide better approximation. The last layer gives the output of the network. There exist nonlinear transformation between the input layer and hidden layer. However, from hidden layer to the output layer it is linear transformation [2]. There are used some radial basis functions as functions of RBFNNs such that Gaussian RBFs, multiquadratic RBFs, inverse multiquadratic RBFs, thin plate splines RBFs, cubic splines RBFs, linear splines RBFs. However, Gausssian RBFs are employed frequently, since it is bounded, strictly positive and continuous on 9t.[2]. Moreover, they are known with noise suppression properties [5]. So in this study, Gaussian RBFs are utilized in the network.In this paper, a novel radial basis function (RBF) neural network is proposed and applied successively for online stable identification and control of nonlinear discrete-time systems. The proposed RBF network has one hidden layer neural network (NN) with its all parameters being adaptable. The RBF network parameters are optimized by gradient descent method with stable learning rate whose stable convergence behavior is proved by Lyapunov stability approach. The aim of construction of the proposed RBF network is to combine power of the networks which have different mapping abilities. These networks are autoregressive exogenous input model, nonlinear static NN model and nonlinear dynamic NN model. In simulations, the proposed network is applied for the direct inverse control of one benchmark nonlinear functioned system and Van de Vusse reaction in a CSTR discrete system even there exist large disturbances. From simulations, it is seen that the RBF network with stable leaming rate identifies and controls nonlinear systems accurately.
“…x; =W;,2Uk_i + y~(k -1) + y; (k -1) Neural networks are applied for inverse control of nonlinear systems in many works [22][23][24][25][26]. In that control, the NN model is applied to identify the plant input and then to generate the new input of the plant.…”
Section: Rbf Network Inverse Controlmentioning
confidence: 99%
“…Here, the identification MSE of u(t) is 8e-3 and tracking MSE is 0.18. where GO is the dynamic inverse of the plant [22]. In Fig.…”
I selami.bevhan@ege.edu.tr , 2 musa.alcirmege.edu.tr
Radial Basis Function NetworksRadial basis function neural networks (RBFNNs) are the one of the different functionalized type of NNs with high approximation and regularization capability [3]. The RBFs are preferred as the basic structure of neural networks because of their good local specialization and global generalization ability [4]. The design of a RBFN in its most basic form consists of three separate layers. The first layer is the input layer. The second layer is the hidden layer and it is structured with high dimension to provide better approximation. The last layer gives the output of the network. There exist nonlinear transformation between the input layer and hidden layer. However, from hidden layer to the output layer it is linear transformation [2]. There are used some radial basis functions as functions of RBFNNs such that Gaussian RBFs, multiquadratic RBFs, inverse multiquadratic RBFs, thin plate splines RBFs, cubic splines RBFs, linear splines RBFs. However, Gausssian RBFs are employed frequently, since it is bounded, strictly positive and continuous on 9t.[2]. Moreover, they are known with noise suppression properties [5]. So in this study, Gaussian RBFs are utilized in the network.In this paper, a novel radial basis function (RBF) neural network is proposed and applied successively for online stable identification and control of nonlinear discrete-time systems. The proposed RBF network has one hidden layer neural network (NN) with its all parameters being adaptable. The RBF network parameters are optimized by gradient descent method with stable learning rate whose stable convergence behavior is proved by Lyapunov stability approach. The aim of construction of the proposed RBF network is to combine power of the networks which have different mapping abilities. These networks are autoregressive exogenous input model, nonlinear static NN model and nonlinear dynamic NN model. In simulations, the proposed network is applied for the direct inverse control of one benchmark nonlinear functioned system and Van de Vusse reaction in a CSTR discrete system even there exist large disturbances. From simulations, it is seen that the RBF network with stable leaming rate identifies and controls nonlinear systems accurately.
“…Along with the development of artificial neural networks, they are applied for inverse control of nonlinear systems in many works [4][5][6][7]. Since Albus proposed the cerebellar model articulation controller (CMAC) in 1975, it has earned widespread interest because of its rapid learning convergence.…”
Abstract. In order to solve the difficulty in complicated system control, a new direct inverse model control strategy is proposed based on a new improved CMAC (Cerebellar Model Articulation Controller) neural network to control a kind of nonlinear system with strong hysteresis i.e. continuous-stirred tank reactor (CSTR). The idea of credit is introduced to help design a new Improved Credit Assigned CMAC (ICA-CMAC) with fast learning speed, which is helpful in real time control of CSTR. Simulation results show that the ICA-CMAC based method performs faster than conventional CMAC, and is strong in self-learning and helpful for improving the nonlinear control performance.
“…This advantage is combined with the ability of fuzzy system to describe a system with linguistics variable which is easier to be understood by human. This integration is very advantageous for nonlinear plants where its mathematical model is very difficult to derive and creates a powerful tool for identification process as well as for control system designs [1][2][3][4][5][6].…”
Section: Introductionmentioning
confidence: 99%
“…The integration between virtual sensor and a controller enables a development of an on-line control scheme involving the immeasurable variable. The selected controller is a neuro-fuzzy based controller, namely Adaptive Neuro-Fuzzy Inference Systems (ANFIS) controller with on-line learning [1,4,5,7]. ANFIS has the ability to deal with complex, nonlinear, and time varying systems with least numerical information.…”
Abstract. In many industrial plants, some key variables cannot always be measured on-line and for the purpose of control, an alternative of sensing system is required. This paper is concerned with a development of an alternative intelligent control strategy, which is an integration between the neuro-fuzzy based controller and virtual sensing system. This allows an immeasurable variable to be inferred and used for control. The virtual sensor is composed of the Diagonal Recurrent Neural Network (DRNN) for plant modeling and the Extended Kalman Filter (EKF) as the estimator with inputs from DRNN. The integration between virtual sensor and the controller enables a development of an on-line control scheme involving the immeasurable variable. The real-time implementation demonstrates the applicability and the performance of the proposed intelligent control scheme, especially in dealing with nonlinear processes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.