Optimization of multi-bolt assembled precision optical mirrors considering non-ideal mating surfaces

Non-ideal surface mating with multiple bolts is an important assembly structure in precision optical mirrors. However, determining the optimal torque for each bolt under the non-ideal contact of structural mating surfaces remains challenging. This paper develops an integrated optimization strategy that synergizes experimental analysis, finite element modeling, and intelligent algorithms to address this challenge. Mirror assembly experiments show that both non-ideal morphology and bolt torque significantly affect assembly accuracy. Simulations reveal non-uniform contact pressure arising from non-ideal surface morphology during assembly. A Genetic Algorithm-Backpropagation (GA-BP) neural network mapping bolt torque to assembly accuracy is integrated with a genetic algorithm to determine the optimal torque configuration. Experimental and numerical validations confirm the strategy's efficacy, achieving an average reduction in assembly-induced surface-figure Root Mean Square (RMS) of 36.09 %. The proposed approach provides a practical solution for the precision assembly of multi-bolt structures, effectively mitigating the adverse effects of non-ideal mating surfaces.