Aerosol optical thickness retrieval from satellite observation using support vector regression
CIARP'10 Proceedings of the 15th Iberoamerican congress conference on Progress in pattern recognition, image analysis, computer vision, and applications
| Hi-index | 0.00 |
According to the limitations of traditional BP neural network algorithm, the method of adding momentum factor and changing learning rate is used to improve the traditional BP neural network algorithm and establish the new model of BP neural network which is applied to the urban air quality prediction. Practical application shows that improved BP neural network algorithm overcome the shortcomings like slow convergence speed, bad generation ability and easily falling into local minimum values. The model established for urban air quality prediction has characteristics of representative and predicting ability so that it has a broad application prospect in future urban air quality assessment.