融合机理与数据驱动的超超临界锅炉水冷壁温度预测模型OA
Fusion Mechanism and Data-Driven Prediction Model for Water Wall Temperature of Ultra Supercritical Boilers
针对锅炉炉内温度场高度非线性和强耦合性导致炉内水冷壁温度难以准确预测的问题,提出融合机理与长短期记忆网络(LSTM)建模方法,建立考虑炉内温度的水冷壁温度机理模型和水冷壁温度LSTM数据驱动模型;构建了机理与数据驱动模型可靠性评估指标,并作为模型切换条件.以某1 050 MW超超临界锅炉半年壁温数据作为测试集,结果表明:平均温度误差低于0.4％,最大误差不超过1.5％.提出的方法在精度和鲁棒性方面具有一定的优越性,为水冷壁温度的准确计算提供了一种新的解决方法.
The temperature field inside the furnace has a high degree of nonlinearity and strong coupling,which makes it difficult to accurately predict the temperature of water-cooled walls inside the furnace,therefore,a fusion mechanism and long short term memory(LSTM)modeling method is proposed to establish a water-cooled wall temperature mecha-nism model considering the temperature inside the furnace,and to establish a data-driven LSTM model for water-cooled wall temperature.The mechanism and data-driven model reliability evaluation indicators are constructed,and used as model switching conditions.The wall temperature data of a 1 050 MW ultra supercritical boiler during the half year are used as the test set,and the results show the average temperature error is less than 0.4％,and the maximum error does not exceed 1.5％.The method proposed in this article has certain advantages in accuracy and robustness,providing a new solution for accurate calculation of water-cooled wall temperature.
洪兵;孙波;伍玉祥;华山;王枢充;王迪;韩驰
国能浙江宁海发电有限公司,浙江宁波 315000国能浙江宁海发电有限公司,浙江宁波 315000国能浙江宁海发电有限公司,浙江宁波 315000国能南京电力试验研究有限公司,江苏南京 210000国能南京电力试验研究有限公司,江苏南京 210000东北电力大学自动化工程学院,吉林吉林 132000吉林工业职业技术学院,吉林吉林 132000
能源科技
机理模型水冷壁温度预测数据驱动
mechanism modelwater-cooled walltemperature predictiondata-driven
《锅炉技术》 2026 (3)
1-7,7
国家自然科学基金(52306004)
评论