The invention discloses a characteristic parameter-based method for the condition monitoring and the fault identification of a planetary gearbox. According to the characteristic parameter-based method, two characteristic parameters, i.e. filtered root mean square (FRMS) and difference spectrum energy (DSE), are proposed, the running condition of the planetary gearbox is monitored respectively through a time domain and a frequency domain due to the two characteristic parameters, and the fault of the planetary gearbox is finally identified by combining the time domain with the frequency domain. The characteristic parameter-based method has the advantages that: the two characteristic parameters are proposed aiming at the problem of fault diagnosis of the planetary gearbox, the defects that the traditional parameters, such as root mean square, kurtosis, FM0, FM4 and the like, for the condition monitoring of a fixed-axle gearbox cannot used for effectively monitoring the running condition of the planetary gearbox, are overcome, and thus, the condition monitoring and the fault identification of the planetary gearbox can be accurately achieved.