Title :
Identification of a nonlinear multivariable dynamic process using feed-forward networks
Author :
Isik, C. ; Çakmakci, A. Mete
Author_Institution :
Dept. of Electr. & Comput. Eng., Syracuse Univ., NY, USA
Abstract :
The practical aspects of identifying a nonlinear multi-input-multi-output dynamic system using feedforward neural networks (NNs) are discussed. By utilizing the measurements of 25 input and internal variables of the process, the primary process output is estimated with a network that has one hidden layer and partial connectivity. Two different connectivity patterns are compared, and problems encountered during the development are summarized. The accuracy of the estimate is demonstrated by comparing the NN output with the process output in time domain and frequency domain
Keywords :
feedforward neural nets; identification; multivariable control systems; nonlinear control systems; connectivity patterns; feed-forward networks; hidden layer; identification; multi-input-multi-output dynamic system; nonlinear multivariable dynamic process; partial connectivity; primary process output; process output; Acceleration; Control systems; Differential equations; Feedforward systems; Mechanical variables control; Neural networks; Nonlinear dynamical systems; Signal processing; System identification; Velocity control;
Conference_Titel :
Neural Networks, 1993., IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0999-5
DOI :
10.1109/ICNN.1993.298619
