Effective machine learning pull-in instability estimation of an electrostatically nano actuator under the influences of intermolecular forces

Considering the importance of approximating the pull-in instability occurrence through determining its respective threshold voltage in nano-structures, this work proposes the use of an effective Multi-Layer Perceptron (MLP). Neural Network (NN) and Support Vector Regression (SVR) methods, both having excellent capabilities in estimating data and its respective regression, are considered. To estimate the pull-in voltage of nanostructures 500 data points are used for training, validation, and test procedures where the pull-in voltage and nanostructure characteristics are set as the target as inputs. The pull-in voltage values are determined using the Step by Step Linearization Method (SSLM) and Galerkin modal expansion method. The MLP employs a feed-forward back-propagation approach with two hidden layers containing 10 and 8 neurons. SVR with a Radial Basis Function (RBF) kernel is also utilized. Comparing the two methods, MLP demonstrates good capability in estimating pull-in voltage, with NN showing effective performance in determining nanostructure pull-in voltage. Also, the capability of the MLP method has been evaluated by comparing with the presented results of previous studies, which indicated the competence of the MLP method in predicting the pull-in voltage of nano-beam switches.Author details: Kindly check and confirm whether the corresponding author is correctly identified.It is correct.