Generative Adversarial Network-based Missing Data Handling and Remaining Useful Life Estimation for Smart Train Control and Monitoring Systems

Table 8

Detailed architecture and relevant parameters of the GAN-based RUL estimation (case for fault code ID 31).

Classification	Detailed architectures
General learning parameters	(i) Learning epoch: 1000; (ii) Learning rate:
Discriminator	(i) Number of Layers:10; (ii) Number of nodes in each layer; =(1, 50, 100, 200, 500, 1500, 2500, 3000, 3500, 2653); (iii) Used activation functions; =(leaky RU for final layer, Sigmoid for layers 1−9)
Generator	(i) Number of layers: 7; (ii) Number of nodes in each layer; =(2653, 2700, 2800, 3000, 3200, 3500, 2653); (iii) Used activation functions: Sigmoid function for each layer