基于已建立的有机朗肯循环(ORC)人工神经网络(ANN)模型，将其与热源进行耦合，从而在不同烟气工况下对ORC进行循环性能预测及工质优选。为了分析与热源耦合的ORCANN模型精度，基于初选的10种工质，比较了该模型与REFPROP软件对基本ORC和回热ORC的计算结果，比较结果表明：该ORCANN模型对大部分循环参数的平均相对偏差都小于5%。在此基础上，针对不同烟气热源温度（523.15，488.15和453.15 K），以最大净输出功为目标，分别优化循环的蒸发温度，优化结果显示：3种热源温度对应的最佳工质分别为R1336mzz(Z)，R600a和R236fa。

The established organic Rankine cycle(ORC)artificial neural network(ANN)model is coupled with the flue gas heat source, so as to predict ORC performance and select the optimal working fluid under different flue gas conditions.In order to analyze the precision of the ORCANN model coupled with heat source, the calculation results of basic and regenerative ORCs by the model and the REFPROP software are compared, based on the initially selected 10 working fluids. The results show that the average relative deviations of the ORCANN model are less than 5% for most of cycle parameters. On this basis, at different flue gas heat source temperatures of 523.15 K, 488.15 K and 453.15 K, the evaporation temperatures of ORC are optimized respectively with the maximization of net output work as target.The optimal working fluids corresponding to three heat source temperatures are R1336mzz(Z), R600a and R236fa.

TK115

国家自然科学基金(52106037)；湖南省自然科学基金(2021JJ40755)；横向课题研究项目,超临界CO2循环优化及动态仿真的关键技术研究（202003024）