DEVELOPMENT OF DYNAMIC MODEL FOR MR ELASTOMERS USING ARTIFICIAL NEURAL NETWORKS

Abstract

The fundamental step is to design MRE based devices for the adaptation of vibration over range of frequency, which is also known as the characterization of magnetorheological elastomers. The properties of shear mode for MREs are characterized over a wide range. Whereas in the squeeze mode, these properties are addressed less often in several studies. The reason behind this being complexities in design of the experiment. The properties of MREs are such that they can be very beneficial for changing stiffness property. These materials can be used for the sake of base isolation in buildings as well as in bridges. They can be used to resist unpredictable loading due to earthquakes and vibrations. The purpose of this research is to study the effect of nano particles of iron in the matrix of silicon rubber. The material was utilized to make samples of MRE elastomers and samples were prepared. The dynamic shear testing is done over these samples using different particle percentages and data is obtained. This data is then utilized in MATLAB to train a neural network. The artificial network model is created, and the data obtained from the neural network (this as results) can be compared with the actual results. The difference in the actual results and results obtained from training of neural network can be compared, which upon conclusion came out to be close.