摘要
运用人工神经网络对东湖沉积物营养状况进行了模拟研究。以东湖三个样点柱状沉积物的TN、TP、TOC、SRP、TSL、pH 6个因子作为网络模型的输入变量,建立了训练样本和测试样本,同时采用有机指数值作为输出变量,对东湖沉积物营养状况进行评价分析。通过模型分析,结果显示东湖沉积物处于肥污染状态。在2008年6月采得的数据样本中,抽取6个样本作为检验网络模型的确证集,对网络模型进行验证,误差结果符合要求,与传统分析法所得评价结果一致,结果显示模型可靠性较好,能够对沉积物营养状况作出正确评价。
An artificial neutral network model was developed to evaluate the nutrient states of sediment sample in Lake Donghu, Wuhan. With six factors such as TN, TP, TOC, SRP, TSL and pH chosen as the input, the training sample and testing sample were established, and results of organic assessment were chosen as output to evaluate the nutrient. Through analysis of model, results of fertility assessment showed that the organic index in the whole Lake is higher than standard level of fertility in sediment. Six samples which were obtained at June of 2008 were used to test the model, the error reached the standard, and testing results kept in accordance with data which were analyzed by traditional method. The efficiency of the network showed that the model performance was fine, and the faction can make correct assessment for nutrient states of the sediment.
出处
《环境科学与技术》
CAS
CSCD
北大核心
2011年第12期110-113,155,共5页
Environmental Science & Technology
基金
国家水专项课题资助(2008ZX07103-001)
关键词
人工神经网络
BP模型
营养状况
肥力评价
artificial neutral network
BP network model
nutrition
fertility assessment
