Fault Diagnosis of Data-Driven Photovoltaic Power Generation System Based on Deep Reinforcement Learning

Algorithm 1

DQN algorithm.

(1)	All parameters of the Q network are initialized randomly with the corresponding value Q
(2)	Clear set D of experience playback
(3)	for episode = 1, M do
(4)	Initialization status , then get the eigenvector
(5)	for t = 1, T do
(6)	Use ε−greedy selection action
(7)	Execute action to get return value and next state
(8)	Sets and gets
(9)	Stores back to experience pool
(10)	Randomly collect a sample from the experience pool
(11)	Update:
(12)	Perform gradient descent steps:
(13)	End