摘要
根据不同机组特性对电厂热电负荷在机组间进行优化分配,是机组节能的重要手段。对于径向基神经网络算法的隐藏层函数中心选取进行了优化,利用梯度下降法对连接权重进行初始化以及更新。利用改进后的径向基神经网络构建了关于全厂煤耗量与机组负荷之间的模型算法,对两台350 MW机组的实际运行数据进行了分析拟合。针对遗传算法可能陷入局部最优的缺点,对算法进行了改进,以得到全局最优解。采用改进后的遗传算法对供热机组进行了负荷优化分配,并与平均分配方案进行了比较,证明了该算法应用于负荷分配的有效性,为优化负荷分配提供了新思路。
According to the characteristics of different units, the optimal of thermal power load among units is an important means of unit energy saving.The selection of radial basis function(RBF) hidden layer function center is optimized, and the connection weight is initialized and updated by gradient descent method.The improved RBF neural network is used to construct the model algorithm between coal consumption and unit load of the whole plant, and the actual operation data of two 350 MW units are analyzed and fitted. The genetic algorithm is improved in order to avoid falling into locally optimal solutions and get the optimal solution.The improved genetic algorithm is used to optimize the load distribution of heating units, and compared with the average distribution method.The effectiveness of the algorithm in load distribution is proved, and a new way for optimizing load distribution is provided.
作者
徐朔文
王丹
XU Shuowen;WANG Dan(Equipment Management Department,Zheneng Aksu Thermoelectric Co.,Ltd.,Aksu 843000,China)
出处
《自动化仪表》
CAS
2022年第1期65-68,共4页
Process Automation Instrumentation
关键词
供热机组
径向基神经网络
能耗特性
遗传算法
负荷优化分配
梯度下降法
隐藏层函数中心
煤耗率
Heating units
Radial basis function(RBF)neural network
Energy consumption characteristics
Genetic algorithm
Optimal load distribution
Gradient descent method
Hidden layer function center
Coal consumption rate
