<p>The wet shift clutch is a key component of the automotive transmission system. The slipping torque during its working process is an important parameter of its performance, affecting the power transmission capacity and operational stability of the transmission system directly. However, the slipping torque is variable with uncertainty during the engagement of the wet shift clutch, affected by multiple uncertain parameters. It is difficult to predict the slipping torque of the wet shift clutch accurately based on the traditional mechanical methods. The paper proposes a new prediction method of shift clutch slipping torque with uncertainty based on the TCN (temporal convolutional network: a deep learning architecture based on convolutional neural networks). The slipping torque of the wet shift clutch is simulated considering the thermalmechanical coupling. The Sobol method is utilized to identify the uncertainty parameters having a greater impact on the slipping torque. And the uncertainty slipping torque is quantified by using the Bootstrap method. The neural network method is utilized to establish the uncertainty slipping torque prediction model based on the MLP (multilayer perceptron network: a feedforward neural network structure composed of multilayer neurons). The prediction results at each time point on the test set are evaluated. The relative error range is 0.74 %-3.88 % at each time point. Moreover, the accuracy of the slipping torque prediction model is validated by test method. The results supply the theoretical basis for the optimal design of the wet shift clutch.</p>