Abstract: Due to the on-site interference and system noise, conventional electrical measurements to detect partial discharge (PD) in gas insulated switchgear (GIS) will cause cases of missed and false detection. In order to improve the accuracy of PD detection and explore the distinction in characteristics of discharges initiated by different defects, this paper proposes a quadrifilar helical antenna optimization process based on particle swarm algorithm, designs a flexible power divider network, and forms an optic-electric integrated sensor based on ultra-high frequency antenna and silicon photomultiplier by combining flexible power divider network with silicon photomultiplier tube loaded with fluorescent fiber. On this basis, the typical discharge signals caused by defects in GIS are detected and statistically analyzed by combined sensors. The results show that the quadrifilar helical antenna attached a power division network maintains gain above 5.0dBi ranged from 0.5~2.0GHz, retaining small-size and high-gain; the discharge signals of typical defects detected by the integrated sensor show that the optical signals and the UHF signals maintain linear relationships with the discharge volume; optic-electric integrated sensing has distinctive sensing efficiencies for corona discharge, surface discharge and air-gap discharge, and the partial discharge interval time of different defect types also has different characteristics. The application of optic-electric integrated detection technology can provide a new and dependable solution for PD detection in GIS.