Co-phasing of the James Webb Space Telescope-like space segmented telescope by neural network wavefront reconstruction.

High-precision piston detection over a large range is the key to the phasing of segmented optical systems. In this paper, a large-range piston error detection method based on an artificial neural network is proposed. By establishing a compound attention mechanism and introducing a multilayer perceptron convolution layer, the network can quickly and accurately learn the key features of high-throughput light-intensity images of dispersed fringe patterns and narrowband far-field spot patterns during training, thereby accurately mapping grayscale images to multi-piston error values. The test results in the simulation show that the method is simple and fast, substantially improving the piston error detection efficiency and sensing range, with the ability for highly sensitive fine phase correction under closed-loop conditions. This technique has a wide range of potential applications in simplifying wavefront sensing and modulation of large segmented telescopes.