Abstract: High voltage cable insulation aging threatens power safety, and frequency domain dielectric spectroscopy technology can detect aging problems in advance by analyzing the wideband dielectric response insulation materials, and effectively avoid fault risks. This paper focuses on the diagnosis of insulation aging of high voltage cables, and designs an experimental research scheme based on frequency domain dielectric spectroscopy (FDS). Through the establishment of a high-precision FDS testing system, key characteristic parameters such as the loss factor of medium and the complex dielectric constant are extracted. Combined the dielectric relaxation model fitting, principal component analysis and machine learning algorithm, the accurate diagnosis of the insulation aging state of the cable is realized. Experimental results show that the proposed method can distinguish different aging types and degrees, and provide reliable technical support for the state evaluation of high voltage cable insulation.