Robust Optimization Of MEMS Bistable Acceleration Switch Based On Interval Theory

Due to the limitations of MEMS processing technology and the scale effect of forces,there are uncertainties in the structural parameters and material properties of MEMS bistable acceleration switch.These uncertain factors directly affect the robust of bistable switching performance,and even it may cause the failure of the switch function,so it is of great engineering significance to optimize the robust during the design phase of the MEMS bistable acceleration switch.Because MEMS bistable acceleration switch cannot make a large number of test samples during the design phase,the stochastic optimization method and fuzzy optimization method are limited.In view of the above problems,this thesis uses interval numbers to characterize the uncertainty factors in bistable switch.Based on the non-probabilistic robust optimization design at home and abroad,a robust optimization method for MEMS bistable acceleration switch is proposed.The main work of this thesis is as follows:First,the geometric structure of the MEMS bistable switch is introduced.The working principle and processing flow of the bistable switch are explained in detail,and the uncertainties of the structural parameters and material properties caused by the processing technology is analyzed to determine the effect of the bistable switch.Five kinds of uncertainty factors affecting the performance of the bistable switch and their range of uncertainties are determined.Secondly,the method of representation of interval numbers and its related theory are introduced,and a method of converting interval number optimization models to a multi-objective optimization model with penalty function is introduced by deterministic interval number ranking and degree-based interval number ranking.A two-level nested optimization algorithm based on genetic algorithm is used to solve the deterministic optimization problem after conversion.Next,an interval number robust optimization model for a three-segment long inclined beam is established.The BP neural network proxy model of three-segment long inclined beam is established,which effectively solves the problem of low calculation efficiency of finite element simulation.A two-layer nested genetic algorithm based on BP neural network model was proposed,and the optimal design size of the three-segment long inclined beam was obtained.The results show that after optimization,the maximum value of the bi-stable flexural force of the three-segment long inclined beam is significantly reduced,and the range of its uncertainty region is reduced by 49.3%.At the same time,the constraint satisfaction of the switch contact pressure is increased from 38.2% to 100%.Finally,a dynamic model of a bistable switch is established to simulate the dynamic response of the switch under acceleration excitation.Based on the design index of the bistable switch,an interval number robust optimization model of the overall structure of the switch is established.A two-level nested genetic algorithm based on a dynamic model is proposed,and the optimal design size of the overall structure of the switch is obtained.The results show that the optimized forward threshold of the switch is closer to the design target of 20 g,while the range of the uncertainty region of the forward threshold is reduced by 38%,and the sensitivity of the forward threshold to the uncertainty factor is significantly reduced.This thesis proposes a robust optimization method for MEMS bistable acceleration switch based on interval theory,which provides a reference solution for robust optimization of MEMS bistable acceleration switch,which is beneficial to better achieve the design and quality of MEMS bistable acceleration switch.