Series arc fault (SAF) is one of the main causes of electrical fire. Aiming at the problem of SAF detection in the frequency converter load circuit under complex working conditions in the industrial field, a feature extraction method of the SAF based on the quantum well and polynomial fitting was proposed. First, the SAF experiments under different conditions of power supply harmonics, carrier frequency and the operating frequency of the frequency converter and current level were carried out by using the frequency converter load. Second, the quantum well was used for energy transmission after normalizing the experimental data, then the energy transmission coefficients were sorted according to the energy size, and the fault feature vector was constructed by the polynomial fitting. Finally, the SAF features were tested by utilizing a support vector machine (SVM) identification model. The results indicated that the proposed method can effectively detect the SAF in the frequency converter load circuit under complex working conditions, and the detection accuracy can reach 98.92%.