CAJ | 학술논문

汽轮机组特性随着机组老化而变化，传统上采用定期现场实测，需停机和采用专门的设备与系统，费用很高，因此基于现有自动化系统历史测量数据的特性曲线识别方法十分必要。一般汽轮机组汽耗量特性具有非凸和非连续等特点，常规的多元线性回归拟合不能适应。M5’模型树算法是一种多输入单输出系统的分段线性化的数据挖掘算法。提出采用M5’模型树的抽汽式机组汽耗量特性模型和其模型结构及参数识别算法，用于滚动利用最新的电厂测量历史数据获取最新的汽耗量特性。该方法简单、有效，逼近能力强，自动化程度高，在处理非凸形和非连续性的特性方程具有优势。通过多个热电厂的实时数据进行验证，具有很高的预测精度，效果优于多元线性回归拟合方程。
기륜궤조특성수착궤조노화이변화，전통상채용정기현장실측，수정궤화채용전문적설비여계통，비용흔고，인차기우현유자동화계통역사측량수거적특성곡선식별방법십분필요。일반기륜궤조기모량특성구유비철화비련속등특점，상규적다원선성회귀의합불능괄응。M5’모형수산법시일충다수입단수출계통적분단선성화적수거알굴산법。제출채용M5’모형수적추기식궤조기모량특성모형화기모형결구급삼수식별산법，용우곤동이용최신적전엄측량역사수거획취최신적기모량특성。해방법간단、유효，핍근능력강，자동화정도고，재처리비철형화비련속성적특성방정구유우세。통과다개열전엄적실시수거진행험증，구유흔고적예측정도，효과우우다원선성회귀의합방정。
The working condition characteristics of steam turbine unit will be changed due to aging, so routine shut down test in regular intervals is adopted with special equipment and system in costly expense, identification method using the of existed automation and therefore, the characteristics historical measurement records information system becomes necessary. The steam turbine characteristics normally has the property of non-convex and non-continuity, and multivariate linear regression model is not able to deal with. M5＇ model tree, a data mining method, is good for segmental linearization of single-output multi-output system, was introduced to model the working condition characteristics of extraction unit and to identify the model structure and its parameters. Such model identification method can be used to achieve the latest characteristics by a scroll way using latest measurement data, which can replace the traditional regular characteristics test. Such method has advantages of effective and efficient in implementation, good in approximation, and special property on processing of non-convex and non-continuous fimction of unit characteristics, and was approved by many CHP stations which shows its good performance and better than the general practiced multi-variables linear model.