In order to solve the problems of feature loss and limited prediction accuracy due to subspace segregation in traditional models in the field of aero-engine life prediction, a bidirectional temporal convolutional network model incorporating a multi-head self-attention mechanism was proposed. Firstly, a bidirectional architecture was introduced on the basis of temporal convolutional network to compensate for the defect that temporal convolutional network cannot capture bidirectional time series information. Subsequently, multiple attention heads were established, each independently performing bidirectional temporal convolutional network processing and feature fusion. The fused feature information was then utiliezd to enhance the feature interaction and adaptive weight assignment using the multi-head self-attention mechanism in order to capture different time series features. Finally, the C-MAPSS dataset was used for experimental validation. The validation results show that the model's remaining useful life prediction accuracy for aero-engines is higher than that of the classical methods.