In order to improve the efficiency and accuracy of fault diagnosis of power supply and distribution system in intelligent buildings, a fault diagnosis method based on Bayesian network and wavelet transform was proposed. Firstly, the topological structure of power supply and distribution network in intelligent buildings was analyzed in detail in theory. Secondly, the switching and electrical quantities in fault information were filtered and reorganized by wavelet transform. Finally, the fault information after the reorganization was modeled and analyzed by Bayesian network, and the fault diagnosis results were obtained. In this paper, the process of extracting electrical and switching quantities from fault information was introduced in detail. According to the fault characteristics of the existing intelligent building power supply and distribution system, the corresponding recovery strategy was given. IEEE⁃39 multi⁃node complex power fault system is taken as an example, the simulation results show fault diagnosis result of the proposed method is fast and accurate. The research results have important reference value for fault diagnosis research of intelligent building power supply and distribution network.