An Improved Load Forecasting Method Based on the Transfer Learning Structure under Cyber-Threat Condition

Table 2

Parameter set of different forecasting methods.

Model	Meaning	Value
TCS-transfer learning	Hidden layers	2
	Number of input layer nodes	4
	Number of hidden layer nodes	32
	Activation function	Sigmoid, tanh
	Learning rate	0.01
	Optimization function	ReLU
	Epochs of training	400
	Distance function	MMD
	Data preprocessing methods	Optimal splitting
	Hidden layers	2
MMD-RNN	Number of input layer nodes	4
	Number of hidden layer nodes	32
	Optimization function	ReLU
	Learning rate	0.01
	Epochs of training	500
	Distance function	MMD
	Data preprocessing methods	None
	Hidden layers	2
	Number of input layer nodes	4
	Number of hidden layer nodes	32
LSTM	Activation function	Sigmoid, tanh
	Optimization function	ReLU
	Learning rate	0.01
	Epochs of training	500