论文部分内容阅读
以北京顺义汉石桥湿地自然保护区中水处理厂的潜流湿地为例,选取2014~2015年的水质监测数据,以电导率、溶解性固体总量、氧化还原电位、p H、水温和总输入氮含量为输入层,比较遗传算法优化的BP神经网络模型和广义回归神经网络模型对多处理单元潜流湿地出水中的总氮含量预测能力。研究结果表明,遗传优化的BP神经网络模型的拟合优度R2可达到0.835,平均相对误差百分比为12.89%,说明其对出水中的总氮含量有一定的预测能力,但精度较差;广义回归神经网络模型的平均相对误差百分比为4.46%,精度较高。利用广义回归神经网络模型对潜流湿地出水中的总氮含量进行预测较适宜。
Taking the potential wetland in the water treatment plant of Hanshiqiao Wetland Nature Reserve in Shunyi, Beijing as an example, the monitoring data of water quality from 2014 to 2015 were selected. The conductivity, total dissolved solids, redox potential, p H, water temperature and total The nitrogen content was input as input layer. The BP neural network optimized by genetic algorithm and the generalized regression neural network model were compared to predict the total nitrogen content in the sub-flow wetland effluent of multi-treatment unit. The results show that the goodness of fit (R2) of genetic optimization BP neural network model can reach 0.835, the average relative error percentage is 12.89%, indicating that it has some predictive ability to the total nitrogen content in effluent, but the accuracy is poor. The average relative error percentage of regression neural network model was 4.46% with high accuracy. Using generalized regression neural network model to predict the total nitrogen content in sub-flow wetland effluent is more suitable.