In the context of the “dual carbon” strategy and sustainable development, the green supply chain marketing strategy has become an essential means to guide enterprises and consumers to form awareness of energy conservation and emission reduction. This article proposes a Bidirectional Long Short-Term Memory (BiLSTM) prediction model structure that integrates evidence theory and K-means clustering to address the complex and dynamic modeling of energy awareness in multi-agent participation, and constructs an energy awareness prediction model for green supply chains. Firstly, utilizing evidence theory to fuse multi-source information enhances the ability to handle data uncertainty and decision robustness. Subsequently, the K-means algorithm was used to cluster the historical behavior data of different entities in the supply chain and identify typical energy behavior patterns. Finally, based on the clustering results, a BiLSTM prediction model is constructed to explore temporal behavior characteristics and predict and analyze the energy awareness evolution trends of different entities under green marketing strategy intervention. The experimental results show that compared with models such as CNN BiLSTM, the method proposed in this article has reduced Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), while the upper limit of relative error does not exceed 8%.
Menghan Chen (Thu,) studied this question.