The utility model discloses a transformer fault diagnosis system based on wavelet transform. The transformer fault diagnosis system comprises a transformer load current collection sensor, a voltage collection sensor, a cooling fan pressure sensor, a temperature sensor, a machine body vibration signal sensor, a plurality of collection modules and a background control center, wherein the transformer load current collection sensor, the voltage collection sensor, the cooling fan pressure sensor, the temperature sensor and the machine body vibration signal sensor are connected with the background control center through corresponding collection modules. According to the transformer fault diagnosis system disclosed by the utility model, the sensors monitor the state of a transformer in real time, frequency band-energy characteristic based on wavelet transform is used as standard evaluation of transformer vibration in combination with external environmental factors and the current operation conditions, the diagnosis rule is selected to effectively analyze the main energy frequency band of a vibration signal, so as to accurately evaluate mechanical vibration and health conditions of a transformer iron core and a winding.