利用Adam算法优化后的BP神经网络训练预测P掺杂ZnO后的各体系的缺陷形成能，分析得出最易形成的体系是ZnO∶PZn和ZnO∶PZn(2VZn)体系，反之是ZnO∶PO和ZnO∶PZn(1VZn)体系，之后在第一性原理的基础上研究各体系光电特性，分析可知ZnO∶PZn体系呈n型导电，带隙0.78 eV，大于本征体系。ZnO∶PZn(2VZn)体系呈p型导电，带隙和本征体系相似，电导率与ZnO∶PZn体系相近且都远高于ZnO∶PZn(1VZn)体系，反射率、吸收率和光透率都优于本征ZnO体系。

The defect formation energy of each system after P doped ZnO was predicted using the BP neural network optimized by the Adam algorithm, the analysis shows that the most easily formed systems are the ZnO∶PZn and ZnO∶PZn (2VZn) systems, otherwise the ZnO∶PO and ZnO∶PZn (1VZn) system, and the photoelectric characteristics of each system on the basis of first principles were studied. The analysis shows that the ZnO∶PZn system is n-type conductive, with a band gap of 0.78 eV, which is larger than the intrinsic system. ZnO∶PZn (2VZn) system is p-type conductive, and the band gap is similar to the intrinsic system. The electrical conductivity is similar to the ZnO∶PZn system, which is much higher than the ZnO∶PZn (1VZn) system, and the reflectivity, absorptivity, light transmittance are better than intrinsic ZnO system.