摘要
为解决机组热力系统测点参数失效致使其性能计算模型不能作出正常评估的问题,基于相似性原理,利用某600MW机组热力系统历史正常运行数据建立了其动态超球预测模型,获得了该机组热力系统性能计算模型所需的预测值及相似度曲线,验证了性能计算模型预测值的可靠性,并以测点预测值作为挖掘数据替换其失效值.结果表明:合理选择和适当补入历史正常运行数据,可降低预测值的相对误差,并将其控制在工程允许范围内,提高其替代失效值的可靠性;当输入的热力系统测点参数因测量系统等引起随机波动甚至个别参数失效时,该方法仍可实现对机组热力系统运行性能的正常实时评估和故障诊断.
To solve the problem that the calculation model of a thermal system could not make a normal as- sessment on its performance caused by parameter failures on the measuring point, a dynamic hypersphere prediction model was established for the thermodynamic system of a certain 600 MW power unit based on its normal operation data, with which predicted values and similarity curves were obtained, while the relia- bility of the predicted values was verified, and the predicted values of measuring point were used as mining data to replace its failure values. Results show that the relative error of predicted values can be reduced and controlled within the allowable scope of actual engineering project, and the reliability of the failure data substitution can be improved by reasonable selection and proper supplementation of the normal historic da- ta. This method can help to realize real-time assessment and fault diagnosis on operation performance of the thermodynamic system, in the case of random fluctuation or even individual fault of input parameters on the measuring point caused by the measurement system.
作者
吕玉坤
李金岗
史久志
张庭钰
LUYukun LI Jingang SHI Jiuzhi ZHANG Tingyu(School of Energy, Power and Mechanical Engineering, North China Electric Power University, Baoding 071003, Hebei Province, China Shougang Jingtang United Iron g〉 Steel Co. , Ltd. , Tangshan 063200, Hebei Province, China)
出处
《动力工程学报》
CAS
CSCD
北大核心
2017年第3期242-248,256,共8页
Journal of Chinese Society of Power Engineering
关键词
热力系统
性能计算
失效
挖掘数据
相似度
thermodynamic system
performance calculation
failure
data mining
similarity
