Research on Operation Efficiency Prediction and Optimization of Biological Retention System Based on GA-BP Neural Network Model

Bioretention systems are an important measure in sandy city construction to remove pollutants. This study collected all the operating data samples of bioretention ponds currently operation China and used Mantel test Redundancy Analysis (RDA) optimize main factors that affect changes pollutant-removal efficiency. Genetic Algorithms (GA) were Back Propagation (BP) neural network model, a predictive index system was established predict efficiency removing typical pollutants (ammonia nitrogen nitrite nitrogen) by ponds. The evaluated predicted based on GA-BP model. results indicated highest positive correlation with X3 (rainfall duration), coeffi-cient 0.79, while negative X7 (percentage strong percolating soil) X12 (the proportion mineral fillers), coefficients −0.89. overall regression model for predicting pollu-tants 0.9397 0.9303, respectively, indicating high accuracy representativeness. is 0.9298, removal rates blank area 78.72%, 70.31%, 86.92%, re-spectively. can provide basis reference planning bio-retention areas lacking them.