基于数据挖掘与建模的智能发电人机交互式系统辨识平台研究与应用  

Human-machine interaction system identification platform research and application of intelligent generation based on data mining and modeling

作　　者：田彬 张林 张东明 纪连恩[4] 杨云 崔芸菲 TIAN Bin;ZHANG Lin;ZHANG Dongming;JI Lianen;YANG Yun;CUi Yunfei(CHN Energy ZhiShen Control Technology Co.,Ltd.,Beijing 102211,China;Beijing Engineering Research Center of Power Station Automation,Beijing 102211,China;Guodian Hanchuan Power Generation Co.,Ltd.,Hanchuan 431614 Hubei,China;Department of Computer Science and Technology,China University of Petroleum,Beijing 102249,China)

机构地区：[1]国能智深控制技术有限公司,北京102211 [2]北京市电站自动化工程技术研究中心,北京102211 [3]国电汉川发电有限公司,湖北汉川431614 [4]中国石油大学(北京)计算机科学与技术系,北京102249

出　　处：《电力大数据》2021年第6期9-17,共9页Power Systems and Big Data

基　　金：国家重点研发计划(2017YFB0602105);北京市科技计划(Z181100005118005);国家能源集团2030重大项目(GJNY2030XDXM-19-08.2)

摘　　要：在传统的火电机组建模研究中多用阶跃扰动等实验设计方式,但由于工业过程的复杂性、环境的多变性和实验的局限性,往往很难获得精确的模型,而智能监控系统又对火电机组过程模型的依赖程度很高。本文为解决火电机组精确建模的难题,建立面向人机交互的可视化系统辨识平台,通过对历史数据挖掘得到满足系统辨识要求的数据,利用先验知识及专家分析结果选择模型类、建模算法以及合适的超参数到辨识平台进行计算,而后采用二次辨识算法将复杂、高阶、非线性模型转化为热工过程系统中常用的传递函数结构,最后通过实验从不同角度交互评估所建模型的性能,在工业过程上的应用结果表明人机交互平台辨识得到的模型能用于预测控制、自抗扰控制器参数预估、参数预警等场景,并取得了良好的应用效果。In classical thermal power unit,the current mostly-used modeling method is to design experiments such as adding step disturbances.Due to the complexity of industrial processes,variability of environment,and limitation of experiments,it is often difficult to get accurate model.Meanwhile,the intelligent monitoring system is more and more relied on thermal power unit process models.In order to solve the problem of accurate modeling in thermal power unit,this paper establishes a visual system identification platform for human-computer interaction,by using prior knowledge and expert analysis results,the model class,modeling algorithm and suitable super-parameters are selected and calculated on the identification platform,then the complex,high-order and nonlinear model is transformed into the transfer function structure commonly used in the thermal process system by the quadratic identification algorithm.Finally,experiments are conducted from different aspects to interactively evaluate the established model performance.The experimental results applied in industrial processes demonstrate that the model identified by the human-machine interaction platform is applicable to predictive control,active disturbance rejection controller parameter estimation and parameter warning,etc.,achieving good application results.

关 键 词：火电机组 系统辨识 数据挖掘 数据建模 人机交互平台 

分 类 号：TM621[电气工程—电力系统及自动化]