Assembly accuracy prediction and process parameter optimization of dual-mirror optical system considering in-service environment

As a critical component of high-precision photoelectronic equipment, the coaxial dual-mirror optical system’s imaging stability in service environments is of paramount importance. An assembly accuracy prediction and process parameter optimization method for dual-mirror optical systems is proposed, aiming at stable in-operation performance. First, a finite element model of the dual-mirror optical system incorporating threaded structures is established. The bolt tightening process is simulated, and the influence of coupled-field conditions—vibration, temperature, and microgravity—on mirror surface deformation is analyzed. Second, mirror surface deformation is fitted using Zernike polynomials, and Zemax optical simulation is integrated. A complete optomechanical coupling analysis workflow is thus constructed, linking assembly stress to imaging quality, which is characterized by the energy concentration ratio. An imaging quality prediction surrogate model is developed based on the light gradient boosting machine algorithm, achieving a prediction accuracy exceeding 95%, significantly outperforming models such as random forest and feedforward neural networks. Finally, with the dual objectives of maximizing the initial assembly energy concentration ratio and minimizing the performance degradation difference before and after service, a multi-objective genetic algorithm based on pareto optimality is employed to optimize the distribution of bolt tightening torques. The optimization results demonstrate that the obtained optimal process parameter combination improves the initial assembly energy concentration ratio by 4.37% on average and enhances the post-service energy concentration ratio by 11.2% on average. This effectively strengthens the optical system’s ability to maintain accuracy and imaging stability under harsh service conditions.