Optimization of active support layout for mirrors based on an RBF neural network and a hybrid optimization algorithm.

Area of Science:

• Optical Engineering
• Computational Mechanics
• Artificial Intelligence

Background:

• Optimizing active support layout is critical for optical system performance.
• Traditional methods face challenges with efficiency, long cycles, and local optima.

Purpose of the Study:

• To develop an efficient and accurate method for optimizing the active support layout of a 450 mm meniscus thin mirror.
• To overcome limitations of traditional optimization techniques in optical system design.

Main Methods:

• Coupling Radial Basis Function (RBF) neural networks with hybrid optimization algorithms.
• Establishing a parametric finite element model for the mirror.
• Generating a dataset via design of experiments to train the RBF neural network.
• Employing multi-objective optimization for surface accuracy (RMS and PV values).

Main Results:

• RBF neural network demonstrated excellent fitting performance and predictive accuracy (correlation coefficients > 0.9).
• Optimized mirror surface achieved reduced peak-to-valley (PV) to 5.689 nm and root mean square (RMS) to 1.1775 nm.
• Achieved improvements of approximately 43.78% (PV) and 43.11% (RMS) compared to the initial scheme.
• Optimization efficiency increased by 96% compared to single optimization methods.

Conclusions:

• The proposed RBF neural network and hybrid optimization strategy significantly enhances mirror surface accuracy.
• This method surpasses existing techniques in both accuracy and optimization efficiency.
• The approach holds substantial engineering value for improving optical system performance.