Accurate prediction of the remaining useful life (RUL) of cables can effectively enhance the reliability of power system. Therefore

this paper proposes a method for predicting the RUL of cables based on multi-feature fusion and multi-step estimation. Firstly

the time series data of multiple features are standardized

and important features are selected using the grey relational analysis method. Then

the weights of these features are calculated using the subjective and objective integrated weighting method to obtain a comprehensive feature. Secondly

the grey wolf optimizer and its improved forms are combined with the support vector regression model to form an augmented support vector regression model (ASVR). Thirdly

the ASVR is trained using the known comprehensive features and used for multi-step estimation of unknown comprehensive features. Finally

the RUL of the cable is determined based on the failure threshold of the comprehensive feature and the results of multi-step estimation. The example results show that the mean absolute errors of RUL prediction under the known datasets of 50%

60%

and 70% are 0.44 years

0.14 years

and 0.15 years

respectively. This indicates that the proposed prediction method has achieved accurate predictions of cable RUL

providing effective technical support for the health management of cables.