Model-Based Recurrent Neural Network for Fault Diagnosis of Nonlinear Dynamic Systems
A model-based recurrent neural network (MBRNN) is presented for modeling dynamic systems and their fault diagnosis. This network has a fixed structure that is defined according to the linearized state-space model of the plant. Therefore, the MBRNN has the ability to incorporate the analytical knowledge of the plant in its formulation. Leaving the original topology intact, the MBRNN can subsequently be trained to represent the plant nonlinearities through modifying its nodes’ activation functions, which consist of contours of Gaussian radial basis functions (RBFs). Training in MBRNN involves adjusting the weights of the RBFs so as to modify the contours representing the activation functions. The performance of the MBRNN is demonstrated via several examples, and in application to the IFAC Benchmark Problem. The results indicate that MBRNN requires much shorter training than needed by ordinary recurrent networks. This efficiency in training is attributed to the MBRNN’s fixed topology which is independent of training In application to fault diagnosis, a salient feature of MBRNN is that it can be formulated according to the present model-based fault diagnostic solutions as its starting point, and subsequently improve these solutions via training by adapting them to plant nonlinearities. The diagnostic results indicate that the MBRNN performs better than ‘black box’ neural networks.
KeywordsRegularization Parameter Fault Diagnosis Extend Kalman Filter Recurrent Neural Network Radial Basis Function Network
Unable to display preview. Download preview PDF.
- Miller, W.T. (1989), “Real time application of neural networks for sensor-based control of robots with vision,” IEEE Trans. on Systems, Man, and Cybernetics, vol. 19, pp. 825–831.Google Scholar
- Parlos, A.G., Chong, K.T., and Atiya, A.F. (1994), “Application of the recurrent multilayer perceptron in modeling complex process dynamics,” IEEE Trans. on Neural Networks, vol. 5, no. 2, pp. 255266.Google Scholar
- Livstone, M.M., Farrell, J.A., and Baker, W.L. (1992), “A computationally efficient algorithm for training recurrrent connectionist networks,” ACC/WM2, pp. 555–561.Google Scholar
- Orr, M.J.L. (1996), “Introduction to radial basis function networks,” Tech. Rep., Centre for Cognitive Science, University of Edinburgh.Google Scholar
- Gelb, A. (1974), Applied Optimal Estimation, Cambridge, MA: MIT Press.Google Scholar
- Rumelhart, D.E. and McClelland, J.L. (1986), Parallel distributed processing: Explorations in microstructure of cognition, Cambridge, MA: MIT Press.Google Scholar
- Hagan, M.T., Demuth, H.B., and Beale, M. (1996), Neural Networks Design, Boston: PWS Publishing Company.Google Scholar
- Lewis, F.L., Abdallah, C.T., and Dawson, D.M. (1993), Control of Robot Manipulators,Macmillan Publishing Company.Google Scholar
- Patton, R.J. and Chen, J. (1996), “Neural networks in fault diagnosis of nonlinear dynamic systems,” Engineering Simulation, vol. 13, pp. 905–924.Google Scholar
- Patton, R.J. and Chen, J.A. (1991), “A review of parity space approaches to fault diagnosis,” No. 1, (Baden-Baden), pp. 239–255.Google Scholar