Terahertz communication has become a current research hotspot due to its extremely large bandwidth, and further improving the system capacity in terahertz systems has also become a problem worth exploring. This paper optimizes the terahertz system composed of all-electronic components using a hybrid Probability and Geometric Shaping (PGS) scheme. The Pairwise Optimization (PO) algorithm is employed to adjust the Probabilistically Shaped 16-ary Quadrature Amplitude Modulation (PS-16QAM) signal after probabilistic shaping to obtain the desired hybrid PGS-16QAM signal. By testing the optimization effects of PS-16QAM, Geometric Shaping (GS)-16QAM and PGS-16QAM on uniform 16QAM at different net rates, it is verified that the hybrid shaping has the best optimization effect. When the wireless transmission distance is set to 2 meters and the Normalized Generalized Mutual Information (NGMI) experimets have shown that threshold is set to 0.92, PGS-16QAM achieves a net transmission rate increase of 15.6%, 11.8%, and 3.8% compared to traditional 16QAM, PS-16QAM, and GS-16QAM, respectively.