摘要
研究火电厂供电煤耗优化问题,火电厂供电煤耗是衡量电厂能经济运行的一项重要指标,由于热耗与锅炉平衡效率计算需要大量离线数据,很难在电厂监控系统上实时显示供电煤耗。为解决上述问题,利用可变学习速率的BP算法建立了供电煤耗仿真的三层神经网络模型,并用C语言编制了相应的计算程序。对比不同输入参数以及不同隐层节点数情况下进行仿真,结果表明仅用16个强关联参数和22个隐层节点数的BP神经网络模型就能够在线仿真出精度较高的供电煤耗,并且动态响应好,具有很强的鲁棒性。
Power supply coal consumption is an important economic indicator for a coal-fired power plant.It is difficult to display real-time power supply coal consumption on monitor system because the calculation needs a large number of off-line data.In this paper,the variable learning rate of BP algorithm is established for the power supply coal consumption calculation of neural network model,and the corresponding computer program using C language is also made,then comparison between simulation errors in case of different input parameters and different hidden-layer nodes are also conducted.The results show that the three-layer BP neural network with 16 strong associated parameters and 22 hidden layer nodes is capable of on-line and high precision power supply coal consumption simulation,and has good dynamic response and strong robustness.
出处
《计算机仿真》
CSCD
北大核心
2011年第4期313-315,344,共4页
Computer Simulation
基金
上海市科委科技攻关项目(08DZ1204200)
关键词
供电煤耗
神经网络
输入参数
Power supply coal consumption
Neural network
Input parameters
