Bloom Predication and Analysis Based on Principal Component Analysis and RBF Neural Network Model
Formation process of water bloom in urban landscape waters is complicated,time-varied and uncertain.To overcome shortages of bloom predication,such as low accuracy and excessively complicated prediction model,the dimension reduction ability of principal component analysis was combined with the self-learning ability of artificial neural network,and the PCA-RBF neural network model for bloom predication was proposed in this paper.The principal analysis results were used as the input matrixof RBF neural network and the main pollutants of landscape waters in urban areas were analysed.The results showed that the water quality prediction accuracy of the PCA-RBF neural network was 0.763,with an average relative error of 21.83％and the bloom prediction accuracy of 92.3％,which were much greater than those of a general RBF neural network model.The PCA-RBF neural network has strong generalization ability of bloom prediction model and high prediction accuracy,providing effective means for the prediction and early warning of water bloom and having a good guiding significance to the prevention and control of water bloom in urban landscape water.
Bloom predictionprincipal component analysisneural networklandscape water
万方数据
维普
