学术论文

      基于 OS-ELM 的 CCPP 副产煤气燃料系统在线性能预测

      Online performance prediction of CCPP byproduct coal-gas system based on online sequential extreme learning machine

      摘要:
      针对联合循环发电厂(combined cycle power plant, CCPP)煤气系统因工况变化频繁带来的模型与过程不匹配的问题,提出一种基于OS-ELM ( online sequential extreme learning machine)的CCPP副产煤气燃料系统在线性能预测方法。首先通过分析副产煤气系统各主要组成部件的工作原理,利用流体力学、质量守恒以及能量守恒等关系,建立起以离心压缩机、煤水分离器、冷却器等为核心部件的副产煤气系统机理模型。利用OS-ELM算法和滑动窗口技术对机理模型的输出误差进行修正,实现副产煤气系统出口参数的精确预测和模型的快速在线更新。仿真实验证明,该方法能够准确地预测副产煤气系统的输出压比和温比,并能够跟踪煤气系统工况的变化和特性的漂移,满足实际工业生产的需求。
      Abstract:
      Aiming at the problem of mismatch between the model and the process for a byproduct coal-gas system in a combined cycle power plant ( CCPP) due to frequent changes in working conditions, this article introduces a method for online performance prediction of the CCPP byproduct coal-gas system based on an online sequential extreme learning machine (OS-ELM). Firstly, by analyzing the working principle of each main component in the byproduct coal-gas system and using the fluid mechanics, energy conservation and mass conservation principles, a mechanistic model is established for performance prediction of the byproduct coal-gas system, which essentially consists of scrubbers, centrifugal compressors, and coolers. Further, the OS-ELM and the sliding window technique are also used to correct the error of the mechanistic model, thus we realize the accurate prediction of export parameters and the update of the model in time. Simulation results show that this method can accurately predict the pressure ratio and temperature ratio of the byproduct coal-gas system and track the change in coal-gas system working conditions and the characteristics drift, which meet the needs of actual industrial production.
      作者: 褚菲 [1] 叶俊锋 [1] 马小平 [1] 张淑宁 [2] 吴奇 [1]
      Author: CHU Fei [1] YE Jun-feng [1] MA Xiao-ping [1] ZHANG Shu-ning [2] WU Qi [1]
      作者单位: 中国矿业大学信息与电气工程学院,徐州,221116 鲁东大学信息与电气工程学院,烟台,264025
      刊 名: 工程科学学报 ISTICEIPKU
      年,卷(期): 2016, 38(6)
      分类号: TP273+.1
      机标分类号: X75 TQ5
      在线出版日期: 2016年10月14日
      基金项目: 国家自然科学基金资助项目,江苏省自然科学基金资助项目,江苏省博士后基金资助项目,中国博士后基金资助项目