Structural Optimization Design of Electromagnetic Repulsion Mechanism Based on BP Neural Network and NSGA‐II

Abstract

The long‐stroke electromagnetic repulsion mechanism is subjected to large stresses during the driving phase, and the repulsion disc is prone to fracture due to excessive vibration amplitude up and down. Therefore, it is necessary to reduce the peak stress and vibration amplitude of the repulsion disc, improve the mechanical life, and improve the driving performance without decreasing the motion displacement, but it is not possible to do so at present. To address this problem, this paper takes the ‘coil‐repulsion disc type’ electromagnetic repulsion mechanism as the research object, and builds a simulation model of electromagnetic repulsion mechanism with structural strength and driving performance by using finite element simulation software, and analyzes the effects of four structural parameters, namely, the height of the repulsion disc, the inner diameter of the repulsion disc, the radius of the circular table and the radius of the chamfer, on the displacement and stress distribution of the model. After that, 39 sets of samples were output by BBD experimental design method, which were introduced into the BP neural network prediction model and combined with NSGA‐II algorithm to optimize the structure of the electromagnetic repulsion mechanism. The simulation results show that with the repulsion disc height of 14 mm, the repulsion disc inner diameter of 14 mm, the radius of the circular table is 29.78 mm and the chamfer radius of 7.06 mm, the peak stress of the repulsion disc of the electromagnetic repulsion mechanism is reduced by 57.83%, the amplitude of the repulsion disc is reduced by 3.24 mm, the peak electromagnetic repulsion force is increased by 7.52 kN and speed at the end of 3 ms increased by 9.8%. Under the condition of satisfying the constraints and increasing the overall displacement, the structural strength is increased and the driving performance of the mechanism is ensured to be improved, providing reasonable mechanism parameters for the opening and closing requirements of the fast mechanical switch. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC.

Structural Optimization Design of Three Phase Enclosed GIS Disconnector Based on MOGOA and ELM

“…However, most of the research on GIS disconnect switches at this stage is of split‐box structure, while GIS equipment of 110 kV voltage level widely adopts three‐phase enclosed structure [15], which has the characteristics of more compact space, more complex structure and higher insulation requirements, and there are fewer related studies, so this paper studies the electric field distribution of common box type GIS disconnect switches and carries out structural optimization design. Traditional optimization methods combine finite element simulation with optimization algorithms, which require several iterations of computation and cannot balance solution accuracy and speed [16], while the study of complex relationships between multidimensional variables and study objectives can be done with the help of machine learning algorithms such as Extreme Learning Machines (ELM). Ref.…”