New technologies such as artificial intelligence (AI) play important roles in transportation emissions trading platforms. Due to the complexity and stochastically changing prices of emissions, an effective algorithm is needed in these platforms to optimize the use of emissions. By setting an upper bound for buying and a lower bound for selling, such an algorithm can reduce trading risk and ensure the stability of trading platforms. In this study, we developed an automatic emissions permit trading system using a dynamic programming approach with selling and purchasing bounds for transportation firms with non-negligible fixed transaction setup costs. We partially characterize the optimal transportation and permit trading policies by exploiting a new mathematical property that is suitable for a two-dimensional control system. We attempt to elucidate the optimal coordination of permit trading and permit consumption for a transportation firm facing both the Markov price process and random demand during a multi-period planning horizon. We prescribe an optimal trading policy and propose a well-performed heuristic policy and a tight lower bound for the platform. We also show that the easily implemented heuristic policy would not significantly increase emissions. Our findings contribute to the literature and provide guidance to help transportation firms using AI platforms.

人工智能（AI）等新技术在交通排放交易平台中发挥着重要作用。由于排放的复杂性和随机变化的价格，这些平台需要一种有效的算法来优化排放的使用。通过设置买入的上限和卖出的下限，这样的算法可以降低交易风险，保证交易平台的稳定性。在这项研究中，我们使用动态规划方法开发了一个自动排放许可交易系统，该系统具有不可忽略的固定交易设置成本的运输公司的销售和购买界限。我们通过开发适用于二维控制系统的新数学属性来部分描述最优运输和许可证交易政策。我们试图阐明在多周期规划范围内，面临马尔可夫价格过程和随机需求的运输公司的许可证交易和许可证消费的最佳协调。我们规定了一个最佳交易策略，并为平台提出了一个性能良好的启发式策略和一个严格的下限。我们还表明，易于实施的启发式政策不会显着增加排放。我们的发现对文献有贡献，并为帮助运输公司使用 AI 平台提供指导。我们规定了一个最佳交易策略，并为平台提出了一个性能良好的启发式策略和一个严格的下限。我们还表明，易于实施的启发式政策不会显着增加排放。我们的发现对文献有贡献，并为帮助运输公司使用 AI 平台提供指导。我们规定了一个最佳交易策略，并为平台提出了一个性能良好的启发式策略和一个严格的下限。我们还表明，易于实施的启发式政策不会显着增加排放。我们的发现对文献有贡献，并为帮助运输公司使用 AI 平台提供指导。