针对大型机组主蒸汽流量计算模型复杂繁琐,存在众多限制条件且精度不高的现状,提出一种基于广义回归神经网络(GRNN)的主蒸汽流量软测量模型,并通过平均影响值(MIV)的评比筛选有效地减少模型变量,通过分布密度spread的优化调整改善模型的泛化能力.实例计算表明,优化后的GRNN主蒸汽流量测量模型简洁准确,具有良好的训练精度与泛化能力,可有效地用于主蒸汽流量的软测量. In view of the complex and cumbersome calculation model of main steam flow of main unit, which has many limitations and low precision, a soft sensor model of main steam flow based on generalized regression neural network (GRNN) is proposed. Through the average influence value (MIV) The comparison screening can effectively reduce the model variables and improve the generalization ability of the model through the optimization adjustment of the distribution spread.Examples show that the optimized GRNN main steam flow measurement model is concise and accurate and has good training accuracy and generalization ability, Used for soft metering of main steam flow.