Reducing CO₂ emissions is a hot topic, and an important policy to achieve this target is carbon tax. When an enterprise is subject to a carbon tax, it has to pay this extra fee for the long-term if it does not upgrade its production technology. It needs to pay a certain upgrade fee in the short-term if it chooses to upgrade its plant. Thus, it has been an important problem for enterprises seeking to balance the trade-off between the ‘long-term tax fee’ and the ‘short-term upgrade fee’. This paper explores how to optimise an enterprise’s production technology upgrade strategy based on existing low-carbon technologies, to minimise the total upgrade cost subject to an expected total cost per product. An integer programming model is proposed to formulate the problem, and a ‘multi-agent system – genetic algorithm’ method is presented for its solution. The model is applied to a numerical example and the results indicate that the proposed method is feasible. The impacts of carbon tax and enterprise’s expected cost on its technology upgrade strategy are further discussed.

减少CO ₂排放是一个热门话题，实现这一目标的重要政策是碳税。当企业要征收碳税时，如果不升级生产技术，则必须长期支付这笔额外费用。如果选择升级工厂，则需要在短期内支付一定的升级费用。因此，对于寻求在“长期税费”和“短期升级费”之间进行权衡的企业来说，这已经成为一个重要的问题。本文探讨了如何在现有低碳技术的基础上优化企业的生产技术升级策略，以使总升级成本降至最低，并以预期的每种产品总成本为准。提出了一个整数规划模型来解决这个问题，并提出了一种“多智能体系统-遗传算法”的解决方案。将该模型应用于数值算例，结果表明该方法是可行的。进一步讨论了碳税和企业预期成本对其技术升级策略的影响。