Adaptive split recalling-enhanced recurrent neural network based predictive control for the nano positioning of an electrostatic MEMS actuator

Increase in the micro manufacturing of tools has led to increased need for high precision in positioning devices especially those used in the Electrostatic Micro Electro Mechanical System (MEMS) actuators. There is a large utilization of MEMS actuators in different sectors due to the high efficiency and flexibility of MEMS. But certain limitations are inherent to these types of actuators, such as non-linear behavior and external noise which affect the possibility of reaching the desired function of nano-positions. One objective of this work is to improve the precision and robustness of nano-positioning in MEMS actuators by resolving the difficulties caused by non-linearity and interferences. The purpose of this present research is to design control model which leads to a better, effective and stable position control. In order to accomplish these aims, a new Adaptive Split Recalling Enhanced Recurrent Neural Network (ASRERNN) model is presented for the predictive control in nano-positioning. The ASRERNN model is aimed to precisely define the target position of a parallel plate actuator and in the same time predict and eliminate the impact of other disturbances. The model addresses the challenges posed by non-linearities, providing precise nano-positioning, thereby improving the performance of the MEMS actuator system. Furthermore, the application of Human Evolutionary Optimization was incorporated to come up with the most accurate placement of weights and loss function. Other tests were done in determination of the degrees of system resiliency and stability in conditions point to point and under disturbances. The simulation results do as well prove that with the help of ASRERNN model the positioning precision is increased considerably over than 50% increase in comparison with the conventional loops controllers. The proposed model is also more robust, response faster, and control input adjustments are more efficient under disturbances. The main advantage of the flow’s adaptive predictive mechanism is that the control of inputs can be added in iterations as needed for the optimization of system stability.