BP神经网络模型在空气质量级别评价中的应用

Application of BP neural network model on air quality rank appraisal

为了方便广大市民及时准确的了解空气质量状况,利用环境评价问题建立多层前向神经网络数学模型,以上海市2007年12月份的空气质量状况指标作为训练样本,对网络模型进行训练,使模型不断学习样本中存在的内在模式,并将训练好的网络用于空气质量状况评价。将评价结果与实际结果进行分析比较后发现,该网络模型具有较高的评价精度、较低的误差率。采用Matlab软件进行实验,评价准确度达95.83%。

In order to facilitate the general residential to understand air quality condition promptly and accurately, using environment appraisal question to establish multi-layer front neural network mathematical model, taking air qualitative index of Shanghai on December, 2007 as the training sample, the training to the network is carded on, the model unceasingly to study the intrinsic pattern which exists in the sample, and the trained network in the air quality appraisal is used. After carrying on the analysis comparison between evaluating the result and the actual result, we discover that, this network model has the higher appraisal precision, the lower error coefficient, the Matlab software is also used to carry on the experiment, the appraisal accuracy reaches 95.83%.