基于BP-RBF组合神经网络的废气监测盲区SO_2浓度预测

Prediction of SO_2 Concentration in the Blind Area of Exhaust Gas Monitoring Based on BP-RBF Combined Neural Network

工业园区大气管理中,监测盲区的废气浓度分析是现有监测系统需要解决的难点问题.本文提出一种组合神经网络,利用已知监测点信息对监测盲区的废气浓度进行预测.首先,根据BP与RBF神经网络的特点,提出二者组合的神经网络结构;其次,分析监测盲区废气浓度预测问题,并提出基于BP-RBF组合网络的预测模型算法;最后,运用工业园区SO_2实际监测数据对所提组合网络预测方法进行实验验证.实验结果表明:本文所提BP-RBF组合网络预测方法具有良好的性能,适用于监测盲区废气浓度预测问题.

In the management of industrial park atmospheric environment,it is a significant issue to analyze the exhaust gas concentration in the blind monitoring area by the current monitoring system.A combinatorial neutral network was proposed to predict the exhaust gas concentration in the blind monitoring area with the known monitoring information.Firstly,a neutral network structure was introduced to combine BP and RBF neural networks according to their characteristics.Secondly,the prediction problem of monitoring the exhaust gas concentration was analyzed in the blind area,and the algorithm of prediction model was presented based on BP-RBF combinatorial neural network.Finally,the experiment was conducted to validate the proposed prediction method with the practical monitoring data of SO2 concentration in an industrial park.The experiment result indicates that the prediction method in this paper has good performance.And it is suitable for monitoring the prediction of waste gas concentration in the blind area.