To enhance the performance of multi-stage reluctance coil emitters and resolve the issue of electromagnetic emission speed influenced by triggering timing, this study proposes a deep reinforcement learning-based multi-stage coil triggering control strategy. Using coil voltage, current, and temperature as state factors, these parameters serve as input data for the time-based decision network model constructed by the deep reinforcement learning algorithm. The decision factors are determined by the triggering timing of each stage in the multi-stage coil system. An error function is established using the power output of the multi-stage reluctance coil emitter as the evaluation reward, which determines the optimal triggering timing actions. Through iterative optimization, the best triggering timing control strategy is derived. Experimental verification on a fifteen-stage reluctance coil emitter demonstrates the effectiveness of this method, providing valuable references for optimizing electromagnetic coil emitter parameters in multi-stage systems.