As for the rapid expansion of high-speed electrified railway industry, the current collection quality of electric locomotives becomes one of the crucial technical issues for the stable operation of trains. The key measurement index that can characterize the quality of current collection is the contact force between the catenary and pantograph on the locomotive roof. One of the major factors that affect the contact force is the contact wire irregularity, which appears to be very difficult to be measured or detected at present. In recent years, the frequency-domain indicators of the pantograph–catenary contact force (PCCF) have been gaining importance to address this issue. In this paper, a recent quadratic time–frequency distribution is selected to describe the time–frequency representation (TFR) of PCCF signal based on its nonstationarity. The Zhao–Atlas–Marks distribution (ZAMD) shows high time–frequency resolution and cross-term suppression for PCCF signals. To investigate the deterioration of the PCCF under contact wire irregularity, a pantograph–catenary interaction model is presented and verified. Some synthetic contact wire irregularities are generated and added to the interaction model. Based on the characteristics of quadratic TFRs of PCCF signals, a detection and localization approach for contact wire irregularity is proposed in this paper. The simulation and real-life experimental results indicate that the different forms of contact wire irregularity could be effectively detected and localized with meter accuracy using the ZAMD.