The abrupt change of train vibration energy is an important basis for judging track defects or deteriorations. Based on the continue wavelet transform’s detailed performance in time–scale mesh generation, and combined with the probability method, the time, scale and time-scale wavelet energy spectrum of train vibration response and track irregularity signal was established after the calculating and analyzing of track irregularity and train vibration energy, then the combined diagnosis and analysis was conducted. Through the track inspection data analysis for Shanghai-Nanjing Passenger Dedicated Railway and Shenchi-Huanghua Heavy Haul Railway, the results show that the method this paper proposed can effectively extract the type, damage degree and space position of small disease, and realize the evaluation of track irregularity’ s overall state. The proposed method offers a new technical approach for railway track disease diagnosis and state evaluation.