Studying The Mechanical Properties Of Micro-Cantilever Beam Based On Neural Network

Artificial neural network is a cognitive understanding of human brain networks based on artificial structures to achieve some kind of neural network function. It is an information processing system that has certain performance characteristics in common with biological neural networks. Neural network possesses many of advantages: simple structure, easy implementation in hardware, the basic parallel computational architecture, and the model has features of study, memory, self adapting and diversity. In this paper, we study the mechanical properties of micro-cantilever beam based on BP network algorithm.BP network algorithm is widely used in nonlinear modeling, pattern recognition, prediction and so on. After pre-processing of experimental data of micro-cantilever beam load and deflection by wavelet multi-scale analysis, we establish the BP model of studying the relationship of load and deflection in the environment of MATLAB7.0. Then, we simulate and predict the relationship of load and deflection of different parameters micro-cantilever beam by analyzing the experimental data.Moreover, some modified BP algorithms are discussed in order to eliminate the disadvantages of standard back-propagation algorithm, such as slowly convergence speed and catching in a minimum point easily. The experimental results show that the deflection increases with increase in the beam length for a fixed beam width and thickness, decreases with increase in the beam width for a fixed beam length and thickness, and when deflections of specimens are about less than 600nm, the relationship of load and deflection are linear. For larger displacements, non-linear terms will appear in the force-displacement relationship. At the same time, the average effect of prediction is more than 0.997.The experimental results show that the artificial neural network is effective in studying the mechanical properties of micro-cantilever beam.