Aiming at the intersection with large fluctuation of random traffic, an optimal induction control strategy was proposed, and the orthogonal test method was used to obtain the optimal combination of control parameters. The maximum queuing length was used as the traffic demand threshold to optimize the induction control logic, and the three phase switching mechanisms (priority queuing, priority delay and fixed order) were added to the induction control parameter combination. In the SUMO simulation, the intersection environment of Beichen West Road and Kehui South Road in Beijing was simulated, and the optimal parameter combination of induction control under each traffic flow was selected by using the orthogonal test method. A comparative experiment was designed to verify the effectiveness of the optimal parameter combination, and the optimal parameter combination was applied to the deep Q-network (DQN) algorithm to further optimize the induction control. Results show that the optimal parameter combination can be obtained quickly and effectively by using the orthogonal test method. Under the low and the medium traffic flow, compared with the DQN algorithm without optimal parameter combination, the convergence speed of the DQN algorithm using the optimal parameter combination increase by 48.14% and 38.89% respectively, and the average cumulative vehicle delay decrease by 8.45% and 7.09% respectively.