基于深度学习及混沌优化的燃机电站机组热电负荷优化分配  

Load optimal distribution of combined heat and power in gas turbine power plant using deep learning and chaotic optimization method

作　　者：刘钢 金轶群 曹旭[2] 赖菲[2] 柴胜凯[2] 吴涛[2] 何新[2] 王智微[2] 褚贵宏[2] LIU Gang;JIN Yiqun;CAO Xu;LAI Fei;CHAI Shengkai;WU Tao;HE Xin;WANG Zhiwei;CHU Guihong(SPIC Sihui Cogeneration Co.,Ltd.,Zhaoqing 519000,China;Xi’an Thermal Power Research Institute Co.,Ltd.,Xi’an 710054,China)

机构地区：[1]国家电投四会热电有限公司,广东肇庆519000 [2]西安热工研究院有限公司,陕西西安710054

出　　处：《热力发电》2022年第2期178-182,共5页Thermal Power Generation

摘　　要：提出了一种基于深度学习及混沌优化的燃机电站机组热电负荷优化分配的新方法。采用深度学习理论中长短时记忆(LSTM)神经网络算法建立机组能耗模型,通过对模型中机组能耗和电负荷、热负荷、环境参数之间非解析函数关系训练学习,并采用混沌优化算法对LSTM神经网络算法训练得到的模型进行负荷优化分配,得到机组最优负荷分配下最小气耗量。实际算例计算结果表明,本文方法计算结果有效,可提高机组运行的经济性。A new method for solving the optimal distribution of thermoelectric load in gas turbine power plants is presentedon the basis of deep learning and chaos optimizationmethod.Long-and short-time memory(LSTM)neural network algorithm in deep learning theory is used to establish the energy consumption model for the unit.The non-analytical functional relationship between unit power consumption and electrical load,thermal load and environmental parameters in the model is trained and studied,then the load distribution of the model trained by the deep learning LSTM neural network algorithmis optimized by the chaotic optimization algorithm,and the minimum gas consumption of the unit under the optimal load distribution is obtained.The actual calculation results show that the method is effective and can improve the economy of unit operation.

关 键 词：燃机电站 能耗特性 负荷分配 深度学习 LSTM神经网络 混沌优化 

分 类 号：TM73[电气工程—电力系统及自动化]