The development of large-area high-performance flexible photoelectronic synaptic devices has become a hot topic in the field of neuromorphic computing and artificial vision systems. In this work, we have successfully prepared a large-area, ultra-flexible semiconducting single-walled carbon nanotubes (sc-SWCNTs) photoelectronic synaptic thin-film transistors (TFTs) array (33×34) using solution-processable AlO_X thin film as the dielectrics by roll-to-roll gravure printing. Our photoelectronic synaptic TFTs exhibit excellent electrical properties with high switching ratio (≥10⁵), low subthreshold swing (73 mV dec^-1), excellent photoresponse properties over a wide wavelength range (from 270 nm to 650 nm), sustained photoconductivity effect (only 26.7% drop after removing light source for 36000 s) and remarkable mechanical reliability and flexibility (maintaining excellent electrical properties after bending more than 15000 cycles with a bending radius of 5 mm). In addition, concepts such as multimodal optoelectronic synaptic plasticity, optical writing speed perception simulation, and human eye self-recovery model have been successfully demonstrated using printed flexible sc-SWCNTs photoelectronic neuromorphic TFTs arrays. More importantly, we systematically investigated the response characteristics of these devices under deep ultraviolet light stimulation and, for the first time, successfully simulated bio-inspired visual perception self-recovery including the dynamic transition of the visual system from clarity to blurriness and their self-recovery over time. This work indicates that our photoelectronic neuromorphic TFT devices have great practical potential in human-computer interaction, environment perception, and visual simulation.