Title :
Input projection method for safe use of neural networks based on process data
Author :
Hämäläinen, Jari J. ; Järvimäki, IIpo
Author_Institution :
Tech. Res. Centre of Finland, VTT Autom., Finland
Abstract :
A principle is introduced for avoiding the use of a neural network model outside its validity domain defined by observations used for training and validation. The principle is based on clustering the regressor training data and projecting the input vector to the validity domain described by the cluster centers. The confidence level for the projection can be specified. The method is illustrated by two examples: a friction force model in an elevator simulator and a model of lime re-burning in a rotary kiln
Keywords :
chemical industry; feedforward neural nets; lifts; neurocontrollers; process control; clustering; confidence level; elevator simulator; feedforward neural networks; friction force model; input projection method; lime reburning model; process control; regressor training data; rotary kiln; validity domain; Automation; Electronic mail; Elevators; Feedforward neural networks; Friction; Kilns; Neural networks; Predictive models; Testing; Training data;
Conference_Titel :
Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence. The 1998 IEEE International Joint Conference on
Conference_Location :
Anchorage, AK
Print_ISBN :
0-7803-4859-1
DOI :
10.1109/IJCNN.1998.682261
