摘要
人工湿地技术被广泛应用于污水处理㈣，但其去污机理复杂，影响因子众多，呈高度非线性，缺乏长期运行效果准确可靠的模拟手段。而人工神经网络能在数据样本较大的情况下很好地逼近复杂的非线性函数，因此利用人工神经网络构建模型模拟系统的处理效果可为人工湿地系统的运行管理提供参考。通过构建人工湿地基质系统进行为期4个月的试验，监测得到56组COD去除率数据，经Matlab小波去噪后利用Elman网络构建动态神经网络模型，模拟该人工湿地基质系统对生活污水中COD去除效果。结果表明所构建模型的均方根误差为0．0163，精度较高，对具有非线性时间序列特征的人工湿地复合基质系统去污效果模拟具有较好的应用前景，利用Elman神经网络模型模拟该人工湿地复合基质系统运行后期的COD去除率为49．5％-56．4％。
Constructed wetland has been widely applied in wastewater treatment. There is lack of accurate and reliable simulation methods on constructed wetland treatment because of its complex mechanism, influencing factors and nonlinear. However, the artifi- cial neural network can approximate the nonlinear function under the large data sample. Therefore, the artificial neural network can be used to simulate constructed wetland treatment, which can provide reference for operation management. In 4 months＇ experiment, 56 groups of data of COD removal rate were obtained by constructed wetland simulation device. To simulate COD removal rate of constructed wetland with different compound substrates, the model based on the Elman neural network was presented after wavelet de-noising under the environment of Matlab. The results show that the RMS error of the measurements is 0.016 3, which means the precision of the model is high. There is a good application prospect of Elman neural network to simulate the nonlinear constructed wetland system. COD removal rate of constructed wetland simulation device is 49.5 %-56.4 %.
出处
《净水技术》
CAS
2012年第6期61-64,共4页
Water Purification Technology
基金
云南省应用基础研究面上项目(2010CD066)
国家自然科学基金(41171074
40771013)
