To address the issues of frequent false alarms under dynamic operating conditions and the lack of interpretability in existing diagnostic network architectures for voltage-source inverters, an inverter open circuit fault diagnosis method based on parameter identification and statistical analysis is proposed. Firstly, a recursive least squares algorithm based on the exponential adaptive variable forgetting factor is adopted to identify load parameters, construct three phase current model and calculate current residuals. Secondly, the log-likelihood ratio of current residuals within each time subdomain is computed. The statistical distribution of the cumulative likelihood ratio is derived. Then, the alarm threshold corresponding to a prescribed false alarm rate is set to achieve fault diagnosis. Meanwhile, based on the circuit coefficient matrix, basis vectors in the residual space are selected. Current residual basis vectors that characterize each fault state are derived. The failure power transistors are located by combining basis vectors with residual polarities. Finally, based on the simulation and hardware-in-the-loop experimental results, the rapid responsiveness and robustness of the proposed diagnostic and localization scheme are verified.