State Evaluation Method of Distribution Terminal Based on Deep Reinforcement Learning

Table 4

Prediction performance of the traditional CNN network.

Learning rate	Training parameters	ER
0.050	10000	337.36
0.100	10000	312.72
0.150	10000	278.87
0.050	20000	305.25
0.100	20000	299.87
0.150	20000	268.78