火力发电企业作为我国能源结构的重要组成部分,长期以来是我国碳排放的主要来源,在我国和全球加速推动低碳经济发展的宏观环境下,火电企业积极响应国家“能耗双控”向“碳排放双控”转变的战略部署。在此背景下,精确计量燃煤电厂的碳排放量变得至关重要。在燃煤电厂碳计量中,烟气流量影响燃煤发电中在线监测法的精度,而燃煤消耗量、燃煤元素碳含量以及飞灰碳含量共同决定核算法的可靠性。目前,大多数燃煤发电企业只对流量和燃煤消耗量进行实时监测,在现场恶劣的环境中对燃煤元素碳含量以及飞灰碳含量进行短周期、高频次的直接监测需要花费较大的人力以及物力,流量监测设备也易受烟道环境影响。而软测量技术以其高效和低成本的特点,可为传统碳排放计量过程中关键参数的监测提供一种替代方法。鉴于此,首先阐述了软测量模型的建立过程,包含数据预处理、辅助变量选择、软测量模型建立以及模型校正。数据预处理能够确保数据质量,提高建模效率;辅助变量选择是从大量潜在的变量中筛选出对目标变量的辅助变量,进一步提高建模效率;软测量模型建立主要是基于机理建模和数据驱动建模,是实现目标变量预测的核心;模型校正通过实际的离线或在线数据,对模型进行进一步优化,提高模型的预测精度。其次,针对碳计量相关参数,分析了烟气流量、燃煤消耗量、燃煤元素碳含量和飞灰碳含量监测存在的问题,论述了软测量技术在上述碳计量关键参数的国内外研究进展和应用,评估了机理建模和数据驱动建模技术的有效性、准确性和实用性。其中,机理分析建模主要基于电厂锅炉进出口的能量平衡以及烟风质量守恒等原理,有着确定的数学物理关系式,具有高度可解释性和稳定性,但是建模过程复杂,预测精度较低;数据驱动建模主要是利用各种机器学习方法,基于电厂分布式控制系统(Distributedcontrolsystem,DCS)丰富的运行数据,对碳计量关键参数进行“黑箱建模”,克服了机理分析建模复杂的过程分析,精度相对较高,但是建模过程不明确,且模型对于不同机组的泛化能力较差。最后,对于软测量技术在碳排放计量领域的发展应用进行了总结与展望。对电厂各参数之间的时序结构、电厂自身计算能力的限制以及机理分析融合数据驱动方法的发展提出相关建议,并对国外二氧化碳预测性排放系统结合软测量技术在国内外燃煤电厂的应用进行展望。

As a significant component of China′s energy structure, thermal power generation enterprises have long been the main sourceof carbon emissions in the country. With the global push for a low-carbon economy, those enterprisesare shifting from " dual control of energy consumption" to " dual control of carbon emissions." Under this backdrop, accurately measuring the carbon emissions of coal-firedpower plants has become crucial. In carbon measurement for coal-fired power plants, flue gas flow impacts the accuracy of the online monitoring method. In contrast, coal consumption, carbon content in coal, and carbon content in fly ash jointly determine the reliability ofthe calculation method.Currently, most coal-fired plants only perform real-time monitoring of flow and coal consumption. However, direct, high-frequency, short-cycle monitoring of carbon content in coal and fly ash in harsh plant environments requires significant humanand material resources and flow monitoring equipment is easily affected by the flue gas environment. Soft measurement technology, with itsefficiency and low cost, provides an alternative method for monitoring key parameters in traditional carbon emission measurements.Firstly,this study reviews the establishment of a soft measurement model, including data preprocessing, auxiliary variable selection, model establishment, and model correction. Data preprocessing ensures data quality and improves modeling efficiency. Auxiliary variable selection enhances modeling efficiency by filtering out useful variables. The soft measurement model, based on mechanism and data-driven modeling,is key to predicting target variables. Model correction optimizes the model with actual data, improving prediction accuracy.Secondly, thestudy analyzes issues in monitoring flue gas flow, coal consumption, coal carbon content, and fly ash carbon content. It discusses the research progress and application of soft measurement technology for these parameters. Mechanism modeling, based on energy balanceand mass conservation principles, has high interpretability and stability but is complex and less accurate. Data - driven modeling,using machine learning and data from distributed control systems (DCS) offers higher accuracy, but lacks transparency and generalizationability.Finally, this study summarizes and prospects the development and application of soft measurement technology in the field of carbonemission measurement. It provides suggestions for integrating the time-series structure of various plant parameters, the computational limitations of the plant itself, and the development of methods combining mechanism analysis and data-driven approaches. It summarizes theapplication scenarios of predictive CO emission systems abroad and anticipates the application of such systems combined with soft measurement technology in domestic and international coal-fired power plants.