Based on the current von Neumann bottleneck problem in computer systems, this group proposes a flexible photoelectric bionic synaptic neural network system to solve this problem. This study started with a needs analysis, and conducted device prototype development, performance optimization, and feasibility testing. In this study, flexible high-performance electrical bionic synapses and photoelectric bionic synapses based on the electron capture mechanism were developed. In addition to having excellent memory characteristics, this type of device also has good flexibility and enriches possible application scenarios. In addition, for the photoelectric bionic synapse, this research uses gate control to make the device work in two different working modes. Based on the developed photoelectric bionic synapse and electrical bionic synapse, this study also proposes a working protocol that satisfies the communication of these two parts, and builds a complete flexible photoelectric bionic synapse neural network system.