基于BP神经网络的Kalina循环发电系统最佳运行参数预测  

Prediction of optimal operating parameters for Kalina cycle power system based on BP neural network algorithm

作　　者：杨月莹 王建永 王涛[1] 陈海峰[1] YANG Yue-ying;WANG Jian-yong;WANG Tao;CHEN Hai-feng

机构地区：[1]陕西科技大学机电工程学院,陕西西安710021

出　　处：《节能》2023年第11期1-5,共5页Energy Conservation

基　　金：国家自然科学基金项目(项目编号:51906131);中国博士后科学基金项目(项目编号:2020M673604XB);榆林市科技计划项目(项目编号:CXY-2020-089)。

摘　　要：Kalina循环发电技术是低品位热能回收利用的重要方式,而蒸发压力和蒸发温度是决定系统性能的关键参数。由于热源及工质氨水浓度的多样性,获取不同设计条件下最佳运行参数的过程复杂且耗时。通过遗传算法结合热力学模型的方式,以最大净输出功为目标,获取了1705组数据,基于BP神经网络算法建立Kalina循环的最佳运行参数预测模型。结果表明,选择单隐含层神经元数目为9、训练函数为“trainlm”时,神经网络的预测结果最好。验证集的预测误差均在1.5%范围内,表明建立的神经网络模型可以较好地预测Kalina循环发电系统的最佳蒸发压力和最佳蒸发温度。Kalina cycle power generation technology is an important technology for low grade energy recovery and utilization,and the evaporation pressure and temperature are key parameters that affect the system performance.Obtaining optimal operating parameters under different design conditions is complicated and time-consuming due to the diversity of heat sources and ammonia concentrations.By means of genetic algorithm combined with thermodynamic model,1705 sets of data were obtained with the goal of maximal net power output,and the prediction model of optimal operating parameter for Kalina cycle was established based on the BP neural network algorithm.The result showed that when the number of neurons in the single hidden layer was nine and the training function was trainlm,the prediction result of the neural network was the best.The prediction errors of the verification sets are all within 1.5%,which indicates that the neural network model can predict the optimal operating parameters of the Kalina cycle well.

关 键 词：KALINA循环 BP神经网络 预测模型 最佳运行参数 

分 类 号：TM617[电气工程—电力系统及自动化]