Aiming at frequent malfunctions of the relay protection system caused by the oscillation current in DC traction network of urban rail transit

we propose a method for identifying the oscillation current and short-circuit fault current in DC traction network based on improved local mean decomposition（ILMD） and multi-scale time-frequency entropy. Firstly

the ILMD method is used to analyze the feeder current signal in DC traction network so that the time-frequency distribution of the feeder current signal can be obtained. Secondly

the Shannon entropy theory is introduced into the time-frequency distribution

and the time-frequency plane is divided into some frequency channels. Then the time-frequency entropy of each frequency channel is figured out. Finally

the overall multi-scale time-frequency entropy of the time-frequency plane is acquired

which can be used to quantitatively express the uniformity of the distribution of the feeder current signal’s energy in the plane. Differences in uniformity can indicate differences among operating states of DC traction network

which makes it possible to distinguish the oscillation current and short-circuit fault current in DC traction network by means of multi-scale time-frequency entropy. On the basis of simulation analysis and field data calculation

availability of the suggested approach is certified.