Improving carbon efficiency is an effective way to save energy and reduce harmful emission for a sintering process. Optimizing carbon efficiency is an effective way to achieve that goal, and its precondition is to assess the performance of the carbon efficiency. However, there is seldom research about how to assess the carbon efficiency whether it needs to be optimized. To address this, this paper introduces a performance assessment method for evaluating the performance of the carbon efficiency. First, the sintering process and the key characteristics are analyzed, and the carbon efficiency indexes are defined. Second, the structure of the assessment method is presented. The method consists of a prediction model based on three NNs, and an assessment method based on the fuzzy synthetic evaluation method. Two-level combination strategy is proposed to improve prediction performance and assessment accuracy, with the using of bootstrap aggregating, linear combination, and majority voting. Finally, verification based on process data shows that the proposed method can assess the performance of the carbon efficiency with high accuracy. More specially, the prediction errors of the combination model for the CCR are basically in the range of [-2.738 kg/t, 3.442 kg/t], and for the CO/CO${}_2$ they are basically in the range of [-8.16 × 10⁻³, 4.828 × 10⁻³]. The combination models have better prediction performance than single NNs. Moreover, the assessment accuracy of the proposed method is 87%, which has higher accuracy than other models. This model lays the groundwork of improving the carbon efficiency for sintering process.

提高碳效率是烧结过程节能和减少有害排放的有效途径。优化碳效率是实现这一目标的有效途径，其前提是评估碳效率的表现。然而，关于如何评估碳效率是否需要优化的研究很少。为了解决这个问题，本文介绍了一种用于评估碳效率性能的性能评估方法。首先，分析了烧结过程和关键特性，并定义了碳效率指标。其次，介绍了评估方法的结构。该方法由基于三个神经网络的预测模型和基于模糊综合评价方法的评估方法组成。提出了两级组合策略，以提高预测性能和评估精度，使用引导聚合、线性组合和多数投票。最后，基于过程数据的验证表明，所提出的方法可以高精度地评估碳效率的性能。更具体地说，组合模型对 CCR 的预测误差基本在 [-2.738 kg/t, 3.442 kg/t] 范围内，而对于 CO/CO${}_2$它们基本上在[-8.16 × 10 ^-3 , 4.828 × 10 ^-3 ] 范围内。组合模型比单个 NN 具有更好的预测性能。此外，该方法的评估准确率为87%，比其他模型具有更高的准确率。该模型为提高烧结过程的碳效率奠定了基础。