Online electronic transactions in the big data era form the basis of the application environment for realtime dynamic pricing, and automated dynamic pricing for e-commerce platforms has become a trend. In this paper, the dynamic pricing strategy optimization problem for e-commerce platforms is deeply explored based on deep reinforcement learning methods. The study defines the dynamic pricing problem and transforms this problem into a Markov decision process model, based on which the A3C algorithm is applied to decouple the model to realize the dynamic pricing strategy optimization of e-commerce platform. Experiments show that. The dynamic pricing algorithm for e-commerce platforms based on the A3C method can show better gain results, with an average gain as high as 4033, and is more stable compared to other benchmark algorithms, which can adapt to the complex demand level as well as a large state space. In addition, the effects of gain threshold and loss threshold on the ordering decision of e-commerce platform mainly occur in the low initial inventory region, and the effects on the pricing decision are more significant in the high initial inventory region. This paper has strong application value for pricing strategy optimization and inventory control decision of e-commerce platforms.
