In recent days, supply chain and logistic industries have been going through a transformational wave of automation and digitization. Supply chain management (SCM) can involve machine learning (ML) abilities and prediction models to ensure that the demands are satisfied at a minimum cost. Intelligent models can be developed to determine whether adequate inventory is accessible to encounter the predictable rise in demands, and if not, the system spontaneously begins to adjust the orders with suppliers to source the raw materials for resolving the predicted future high demand. The conventional ways of SCM can be replaced by the design of recent artificial intelligence (AI) and deep learning (DL) techniques. By this motivation, this research presents a tree seed algorithm-based feature selection with optimum DL technique for supply chain management (TSA-ODLSCM). The proposed TSA-ODLSCM model involves the design of a feature subset selection approach using a tree search algorithm (TSA) algorithm. Besides, a new convolutional neural network with fuzzy cognitive maps (CNN-FCM) technique is designed for the classification process. Moreover, optimal parameter tuning of the CNN-FCM model was performed using the Henry gas solubility optimization (HGSO) technique. To exhibit the improved performance of the TSA-ODLSCM approach, a huge range of simulations were executed and outcomes were examined below several aspects. The experimental validation reported an enhanced 96.52% outcome of the TSA-ODLSCM approach over other methods.

最近几天，供应链和物流行业正在经历自动化和数字化的转型浪潮。供应链管理 (SCM) 可以涉及机器学习 (ML) 能力和预测模型，以确保以最低成本满足需求。可以开发智能模型来确定是否有足够的库存来满足可预测的需求增长，如果没有，系统会自发地开始调整与供应商的订单以采购原材料，以解决预测的未来高需求。 SCM 的传统方式可以被最新的人工智能 (AI) 和深度学习 (DL) 技术的设计所取代。出于这个动机，本研究提出了一种基于树种子算法的特征选择和供应链管理的最佳深度学习技术（TSA-ODLSCM）。所提出的 TSA-ODLSCM 模型涉及使用树搜索算法 (TSA) 算法设计特征子集选择方法。此外，还设计了一种新的具有模糊认知图的卷积神经网络（CNN-FCM）技术用于分类过程。此外，使用亨利气体溶解度优化（HGSO）技术对 CNN-FCM 模型进行最佳参数调整。为了展示 TSA-ODLSCM 方法的改进性能，执行了大量模拟，并在以下几个方面检查了结果。实验验证表明，TSA-ODLSCM 方法的结果比其他方法提高了 96.52%。