K. S. Narendra and K. Parthasarathy, `Identification and control of dynamical systems using neural networks', IEEE Trans. Neural Networks, 1(1), 4--27, (1990).  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Implementation of Neural Network Learning Control of a Flexible Space Manipulator R. Todd Newton Yangsbeng Xu CMU RI TR 92 I 1 The Robotics Institute Carnegie Mellon Univmity Pittsburgh, Pennsylvania 15213 August 1992 1992 Carnegie Mellon University This work is supported by the Space Projects Office, Shimizu Corporation, Japan. Contents page i List of Figures Self Mobile Space Manipulator Concept . Photograph of SM 2 and Gravity Compensnfion System ....

....Implementation of Neural Network Learning Control of a Flexible Space Manipulator R. Todd Newton Yangsbeng Xu CMU RI TR 92 I 1 The Robotics Institute Carnegie Mellon Univmity Pittsburgh, Pennsylvania 15213 August 1992 1992 Carnegie Mellon University This work is supported by the Space Projects Office, Shimizu Corporation, Japan. Contents page i List of Figures Self Mobile Space Manipulator Concept . Photograph of SM 2 and Gravity Compensnfion System ....

[Article contains additional citation context not shown here]

K. S. Narendra and tC Parthasarathy, Identification and Control of Dynamical Systems Using Neural Networks, IEEE Transactions On Neural Networks, Vol 1 No. 1, 1992

....preset a number of delayed error corrections. Another di#erence is that ARIMA models include only linear components, whereas our approach is nonlinear. Our error correction system bears also some resemblance to the socalled Nonlinear AutoRegressive with eXogenous inputs recurrent networks (NARX) [NP90]. These networks describe a system by using a nonlinear functional dependence between lagged inputs, outputs and or prediction errors [MC01, p. 71] Hence, NARX models can be seen as a generalization of linear ARMA processes [Wei90, p. 56 7] NARX models are usually approximated by recurrent ....

Narendra K. S. and Parthasarathy K.Identification and control of dynamical systems using neural networks, in: IEEE Transactions on Neural Networks, Vol. 1, 1990, pp. 4-27.

.... N ( 0; 0] 1; 2] rotated over an angle of 3 (linear bivariate, LB ) and 2) generated by considering a Gaussian amplitude spectrum , adding random phase and computing the inverse FFT (linear complex, LC ) The two nonlinear sets were generated by a nonlinear system described in [7]: y k = fl x k Gamma1 x k Gamma2 (x k Gamma1 2:5) 1 x k Gamma1 x k Gamma2 x k ; 2) where fl is a parameter controlling the prevalence of the nonlinear over the linear part of the signal, which was set to fl = 0:6, unless stated otherwise. In the first nonlinear set (nonlinear ....

Narendra, K., Parthasarathy, K.: Identification and control of dynamical systems using neural networks. IEEE Trans. Neural Networks 1 (1990) 4--27

....of neural networks to model nonlinear dynamical systems has led to the development of numerous neural networks based control strategies. Most of these strategies are simply nonlinear extensions of existing linear techniques, such as direct inverse control [3] model reference adaptive control [9], predictive control [3] and internal model control [7] There are a number of successful applications of such ANN based controllers. However, there are still many unresolved issues relating to their use. Stability and robustness cannot be guaranteed in general for most ANN based controllers ....

K.S.Narendra, K.Parthasarathy, "Identification and control of dynamical systems using neural networks", IEEE Transactions on Neural Networks, vol. 1, no. l, 1990, pp. 4 - 27. 1267

....of the truss and robot were reduced to 1 3, while local dimcnsions (joints and grippers) were kept equal. This allows the testbed to be used in an average size laboratory, yet mechanisms are not un workably small. Each joint contains a rare earth magnet DC motor; a harmonic drive speed reducer (60:1 or 100:1 ratio) and an optical encoder on thc motor shaft to measure a joint angle. The motors and the drive components were selected and arranged to give maximum power and torque in a small, lightweight package. Presently. the robot is tethered to the computer hardware. The software has been ....

....moves around. In time, the neural network learns the correct network output to produce the desired refercnce output of the plant, and in turn the lbedback signal decreases. Several stimulating studies have shown that teaching neural networks by error feedback yiekts improved performance [3] 5] 6] However. there has been little work regarding real time implementation, and the advantages of the approach as well as various implementation issues cannot be revealed from simulation. As shown here, the approach of neural networks trained by feedback error learning 16 is implemented in a ....

[Article contains additional citation context not shown here]

K.S. Narendra and K. Panhasarathy, "Identification and control of dynamical systems using neural networks," IEEE Trans. Neural Net., vol. 1, no. 1. 1992.

....of neural networks to model nonlinear dynamical systems has led to the development of numerous neural betworks based control strategies. Most of these techniques are simply nonlinear extensions of existing linear techniques such as direct inverse control [1] model reference adaptive control [6], predictive control [1] and internal model control of the ANN controller. These five operating points (i) P = 1.0 p.u, 0.85 lagging power factor (pf) Z= 0.02 j 0.4 p.u, ii) P = 1.0 p.u, 0.85 lagging pf, Z = 0.025 j 0.6 p.u, iii) P = 0.8 p.u, 0.85 lagging pf, Z = 0.025 j 0.6 p.u, iv) P = ....

Narendra KS, Parthasarathy K, "Identification and control of dynamical systems using neural networks", 1EEE Transactions on Neural Networks, Vol 1, No 1, 1990, pp 4-27.

....neural network features. Section 3 describes IMC strategy pointing out the ways to obtain model and control modules. Also FasBack adaptation capabilities to integrate it into an adaptive IMC strategy are commented. Section 4 presents experimental results on a theoretical MIMe plant proposed in [16]. Section 5 studies identification and control performance of a simulated [17] penicillin plant, showing results for realistic scenarios in which plant output measurements are taken at a lower rate than control action requires, there is noise present in output measurements and plant is time ....

....adaptation rate, which is 0.4 in all simulations. 4. Identification and control of a theoretical non linear MIMO plant To study the performance of the proposed IMC strategy using FasBack adaptive neurofuzzy system, the identification and control of the following non linear MIMe plant proposed in [16], and also used in [18] is studied: Yp l (k l)l = yp2 (k Although it is possible identify and control this system like two sIse systems and assuming it is known the additive effect of control signals, like in [16] we shall illustrate capabilities of neuro fuzzy systems in a more general ....

[Article contains additional citation context not shown here]

Narendra K.S. and Parthasarathy K., "Identification and Control of Dynamical Systems Using Neural Networks", IEEE Transactions on Neural Networks, vol. 1, no. 1, pp. 4-27, March 1990.

....The switch S2 is used to switch in out transmission lines and the switch S3 is used to switch in out a load. Figure 1: Multimachine power system 3 Online Trained Neuroidentifier The neuroidentifier is developed using the series parallel Nonlinear Auto Regressive Moving Average (NARMA) model [10]. This model output y at time (k D depends on both past n values of output and past rn values of input. The neuroidentifier output equation takes the form given byeq. 3) k 1) flY(k) y(k 1) y(k n 1) Y = u(k) u(k 1) u(k m l) 3) wherey(k) and u(k) represent the output ....

....equation takes the form given byeq. 3) k 1) flY(k) y(k 1) y(k n 1) Y = u(k) u(k 1) u(k m l) 3) wherey(k) and u(k) represent the output and input of the plant at time k respectively. This model has been chosen in preference to all other system identification models [10] because online learning is desired to correctly identify the dynamics of the turbogenerator and therefore avoiding a feedback loop in the model, which allows static backpropagation to be used to adjust the NN weights. This reduces the computational overhead substantially for online learning. ....

K.S.Narendra and K.Parthasarathy, "Identification and Control of' Dynamical Systems using Neural Networks", 1EEE Trans. On Neural Networks, vol. 1, no. I, Mar. 1990, pp. 4-27.

.... the turbogenerator optimally around one operating point; at any other operating point, the generator s performance is degraded [1] In recent years, there has been considerable research in the use of artificial neural networks (ANNs) for identification and control of nonlinear systems [2] [3]. An increasing demand in the performance specifications and the complexity of dynamic systems mandate the use of sophisticated information processing and control in almost all branches of engineering systems. The promise of fast computation, versatile represenManuscript received November 29, ....

K. S. Narendra and K. Parthasarathy, "Identification and control of dynamical systems using neural networks," IEEE Trans. Neural Networks, vol. 1, pp. 4--27, Mar. 1990.

....of neural networks to model nonlinear dynamical systems has led to the development of numerous neuralnetwork based control strategies. Most of these strategies are simply nonlinear extensions of existing linear techniques, such as direct inverse control [3] model reference adaptive control [14], predictive control [3] and internal model control [12] There are a number of successful applications of such ANNbased controllers (also called neurocontrollers) However, there are still many unresolved issues relating to their use. Stability and robustness cannot be guaranteed in general for ....

K. S. Narendra and K. Parthasarathy, "Identification and control of dynamical systems using neural networks," IEEE Trans. Neural Networks, vol. 1, pp. 4--27, Mar. 1990.

....trade off, accuracy. 1. Introduction Internally static networks can be trained by the simplest algorithms: for the buffered Multi Layer Perceptron (MLP) with only input buffer (with no recursion) the standard Back Propagation (BP) should be used, while for the NarendraParthasarathy network [18], i.e. buffered MLP with input and output buffers, the so called open loop approximation of the standard BP is usually employed. It consists in opening the loop during the backward phase, feeding the network with the desired outputs instead of the true network outputs. In Infinite Impulse ....

....the case when the recurrent network behaviour relaxes to a fixed point. However, if a general temporal processing is needed, two main gradient based learning approaches exist for recurrent networks [15,16,20] Back Propagation Through Time (BPTT) 6,1,27,15] and Real Time Recurrent Learning (RTRL) [2,18,22,27,15]. BPTT is a family of algorithms which extends the BP paradigm to dynamic networks. There are two major points of view to understand what BPTT is. The first is an intuitive one: time unfolding of the recurrent network, i.e. for single layer single feedback delay fully recurrent networks one can ....

K.S. Narendra, K. Parthasarathy. Identification and control of dynamical systems using neural networks. IEEE Trans. on Neural Networks, vol. 1, pp.4-27, March 1990.

No context found.

K. S. Narendra and K. Parthasarathy, `Identification and control of dynamical systems using neural networks', IEEE Trans. Neural Networks, 1(1), 4--27, (1990).

No context found.

K. S. Narendra and K. Parthasarathy, "Identification and control of dynamical systems using neural networks," IEEE Trans. Neural Networks, vol. 1, pp. 4--27, 1990.

No context found.

K. Narendra and K. Parthasarathy, "Identification and control of dynamical systems using Neural Networks", IEEE Trans. Neural Networks, 1, 4-27 (1990).

No context found.

K. S. Narendra and K. Parthasarathy. Identification and control of dynamical systems using neural networks. IEEE Transactions on Neural Networks, 1:4--27, 1990.

No context found.

K. S. Narendra and K. Parthasarathy, "Identification and Control of Dynamical Systems Using Neural Network," IEEE Transactions on Neural Networks, v1, n1, pp 4-27, March 1990.

No context found.

K.S. Narendra, and K. Parthasarathy, "Identification and Control of a Dynamical System using Neural Networks," IEEE Transactions on Neural Networks, 1(1), 427, 175

No context found.

K. S. Narendra and K. Parthasarathy, "Identification and Control of Dynamical Systems Using Neural Networks," IEEE Trans. On Neural Netwroks, vol.1, no. 1, pp. 4-27, March 1990.

No context found.

Narendra, K.S., & Parthasarathy, K. (1990). Identification and Control of Dynamical Systems Using Neural Networks, IEEE Trans. Neural Networks,, 1(1), 4--27.

No context found.

K. S. Narendra and A. M. Annaswamy, "Identification and control of dynamical systems using neural networks," IEEE Trans. Neural Networks, vol. 1, pp. 4--27, Mar. 1990.

No context found.

Narendra K.S., Parthasarathy K., "Identification and control of dynamical systems using neural networks," IEEE Transactions on Neural Networks, Vol.1, No.1, pp. 4-27, 1990.

No context found.

Narendra, K. S., Parthasarathy, K., "Identification and Control of Dynamical Systems Using Neural Networks," IEEE Trans. on Neural Networks, Vol. 1, No. 1, pp. 4-27, 1990.

No context found.

Narendra, K. S., & Parthasarathy, K. (1990). Identification and control of dynamical systems using neural networks. IEEE Transactions on Neural Networks, 1, 4-27.

No context found.

K. S. Narendra and K. Parthsarathy,"Identification and Control of Dynamical Systems Using Neural Networks", IEEE Trans. Neural Networks,Vol 1(1):4-27. 39

No context found.

K.S. Narendra and K. Parthasarathy, "Identification and control of dynamical systems using neural networks", IEEE Transactions on Neural Net,vorks, vol. 1, no. 1, pp. 4-27, March 1990.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC