Resiliency Assessment of Power Systems Using Deep Reinforcement Learning

Table 2

Acronyms and notations used.

Category	Items/symbols	Description
Acronyms	LoR	Level-of-resilience
	PS	Power system
	DRL	Deep reinforcement learning
	DQN	Deep Q-network
	ML	Machine learning
	CIP	Critical infrastructure protection
	PV	Photovoltaic generator
	DDPG	Deep deterministic policy gradient
	FDA	False data injection
	Q value	State-action value
	(L–G)	Single line-to-ground fault
	(L–L)	Line-to-line fault
	(L–L–G)	Double line-to-ground
Notations	π(S)	Agent’s policy
	V(S)	Value function
	R	Reward
	Gt	Return
	γ	The discounting factor
	S	State
	A	Action
	ϵ	Probability of selecting an action
	,	Weights
		The value function target
		Gradient
		Parameterized function with respect to
		The advantage function
	μ(S)	Actor policy
	ST	Terminal state
	,	The learning rates
	Zh	The mode after hth fault and reconfiguration
	M	A set of attack scenarios
	N	Number of faults/attacks