基于改进小波神经网络的大气质量评价模型

为了缩短人工神经网络的训练时间、减少迭代次数和提高输出结果的准确率，将小波基函数应用于人工神经网络，并用专家评分后归一化处理的方法对输入层的权值初始值进行优化，建立了优化的小波神经网络模型。将该模型对井冈山区域2012年大气监测数据进行评价，实验结果表明:经过优化的小波神经网络模型的评价精度较高。最后与将该模型与其它评价方法相比，该模型还具有计算快速、评价客观、可靠性强、效率更高的特点。

The Model for Atmospheric Quality Assessment Based on Wavelet Neural Network

In order to reduce the artificial neural network training time and the number of iterations,and improve the accuracy of output results,the wavelet function applied in artificial neural network,optimized with the method of ex-pert score normalization initial weights of the input layer values,a optimized wavelet neural network model was es-tablished . According to the atmosphere monitor data in the 2012 ,the model was used for assess the atmosphere qual-ity of the Ji'an city in Jiangxi Province. The assessment results shows that,the model had a higher assessment accu-racy. Finally,the model is compared with other assessment methods,the wavelet neural network has the characteris-tics of rapid calculation,objective assessment,high reliability and higher efficiency.