Designing the Topology of Microelectromechanical Systems by Machine Learning Methods

The article reviews and studies specialized methods for the topological optimization of micro electromechanical systems. The article analyzes the optimal topology and searches for an algorithm to optimize the physical characteristics of the microelectronic speaker membrane based on metamaterials. We implemented an optimization method that uses machine learning technologies for practical calculations. The topology of the membrane is optimized to increase the efficiency of the micro speaker. Lattice structures based on individual cells are used to model the membrane. Topological optimization of the complex physical characteristics of the micro speaker was realized by a combination of optimization methods and machine learning algorithms. We obtained high uniaxial load and short-term physical impact based on general topological optimization. That is why the lowest level of resistance to external influence is about 40% of composite materials, and the most optimal result is close to composite structures and standard materials.