To solve the high noise

multidimensional and consistent attenuation with the remaining life of the aeroengine sensor detection data

the wavelet denoising (WD)

time convolution network (TCN) and multi head attention (MHA) mechanisms are used. A new network structure model

time convolution-multi head attention (TCN-MHA)

is presented to solve the mapping relationship among the engine attenuation characteristics to improve the prediction accuracy of aeroengine remaining usefull life (RUL). Firstly

the measurement data of multi-channel sensors are denoised by using WD to remove the white noise interference and reduce the error caused by many factors in the attenuation process of multi-sensor. Secondly

the TCN extracts the temporal characteristics of the processed multidimensional data and maps the degradation relationship among the system performance. Finally the MHA extracts the contribution degree of prediction of each dimension data

so as to assigns different weights to different dimensions data and effectively predicts the RUL of aeroengine. Comparing the present method with the TCN

MHA and long short-term memory network (LSTM) on the commercial modular aero-propulsion system simulation (C-MAPSS) dataset. The result shows that the present prediction method performs better than the other models

which verifies the effectiveness of the present method.