论文部分内容阅读
在分析影响火电厂氮氧化物浓度检测精度的多种原因基础上,提出建立RBF-BP神经网络模型进行改进的方法,并详细说明了样本数据、神经网络构成、训练函数、回归因子、动量因子等关键技术。以2015年实际火电厂检测数据为依据,对建立的RBF-BP神经网络进行了仿真实验,并与单一RBF仿真实验进行对比分析,实验表明平均相对误差为0.3%~1%,证明了该方法的有效性。
Based on the analysis of many reasons that affect the detection accuracy of NOx concentration in thermal power plant, a method to establish RBF-BP neural network model is proposed. The sample data, neural network structure, training function, regression factor, momentum factor And other key technologies. Based on the actual thermal power plant test data in 2015, the RBF-BP neural network is established and compared with the single RBF simulation experiment. The experimental results show that the average relative error is 0.3% ~ 1%, which proves that this method Effectiveness.