论文信息 - Modeling dynamic engineering processes when the governing equations are unknown

Modeling dynamic engineering processes when the governing equations are unknown

Abstract The paper describes a method of modeling the dynamic behaviour of continuous engineering processes, using artificial neural networks. The technique is applicable to situations where the differential equations governing the behaviour of a system are nonlinear and poorly understood, such as is the case for frost-heave and thaw-settlement processes in soils. A means of modeling the unknown component of governing differential equations is first described. A method of discretizing the neural network models of these equations is then illustrated, and the way in which these networks can be used to simulate the behaviour of a process is discussed. The proposed approach is proven to provide highly accurate results in a series of experiments simulating the nonlinear thermal behaviour of translucent solid materials. The paper concludes with an identification of several on-going areas of further development and application of the proposed tool.

Ian Flood

[1] Dale E. Nelson,et al. Extrapolation of Mackey-Glass data using Cascade Correlation , 1992, Simul..

[2] Robert O. Shelton,et al. Controlling a truck with an adaptive critic CMAC design , 1992, Simul..

[3] S Arumugam,et al. RATE DEPENDENT BEHAVIOR OF CLAYS USING NEURAL NETWORKS , 1994 .

[4] Pedro Albrecht,et al. Weighing trucks in motion using Gaussian-based neural networks , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[5] John Moody,et al. Fast Learning in Networks of Locally-Tuned Processing Units , 1989, Neural Computation.

[6] Ian Flood,et al. Neural Networks in Civil Engineering. I: Principles and Understanding , 1994 .

[7] J. Konrad. Sixteenth Canadian Geotechnical Colloquium: Frost heave in soils: concepts and engineering , 1994 .

[8] Pedro Albrecht,et al. Computing truck attributes with artificial neural networks , 1994 .

[9] I. Flood. A Gaussian-based feedforward network architecture and complementary training algorithm , 1991, [Proceedings] 1991 IEEE International Joint Conference on Neural Networks.