为解决传统预测方法在涡扇发动机剩余使用寿命（remaining useful life, RUL）预测精度较低的问题，提出了一种基于卡尔曼滤波与时间卷积神经网络（temporal convolutional network, TCN）相结合的涡扇发动机寿命预测方法。首先，为减小传感器在采集数据时的噪声干扰，使用卡尔曼滤波（Kalman filtering, KF）算法对提取特征进行滤波处理。其次，构建KF-TCN模型，将滤波后数据输入到时间卷积网络中，捕捉时序数据中的时间依赖性并输出RUL预测值。最后，以FD001子集为例，使用商用模块化航空推进系统仿真（commercial modular aero-propulsion system simulation, C-MAPSS）航空发动机数据集对模型进行仿真试验验证。研究结果表明：与TCN、门控循环单元神经网络（gated recurrent unit, GRU）和长短期记忆人工神经网络（long short-term memory, LSTM）模型RUL预测结果相比较，本文所构建KF-TCN模型在发动机RUL预测过程中预测误差较小，为发动机寿命预测提供新思路。

In order to solve the problem of low prediction accuracy of RUL by traditional forecasting methods, a new forecasting method of turbofan engine life based on Kalman filter and time convolutional neural network (KF-TCN) was proposed. Firstly, in order to reduce the noise interference of the sensor when collecting data, Kalman filter algorithm was used to filter the extracted features. Secondly, the KF-TCN model was constructed, and the filtered data was input into the time convolutional network to capture the time dependence of the time series data and output the RUL predicted value. Finally, taking FD001 subset as an example, commercial modular aero-propulsion system simulation (C-MAPSS) aero-engine data set was used to verify the model. The results show that compared with the RUL prediction results of TCN, gated recurrent unit (GRU) and long short-term memory (LSTM) models, the prediction error of KF-TCN model constructed is smaller in the process of engine RUL prediction, which provides a way for engine life prediction.