摘要
实时准确监测河流水质是城市水管理战略的首要任务。溶解氧浓度是评价河流水质优劣的重要指标之一,也是维持水中高等生物生存的重要条件。因此获得准确可靠的溶解氧预测结果对于河流水体的管理和预警至关重要。首先,通过灰色关联度分析得到影响水体溶解氧含量的关键水质因子,即总磷、氨氮、高锰酸盐含量和pH值,进而卷积长短记忆神经网络(CNN;STM)的提出是为了提取溶解氧和其它水质因子之间深层复杂的相关特征,降低了不同信息之间特征的耦合,提高模型的预测精度并降低了模型训练时间。以实际采集京杭运河常州段溶解氧数据为研究对象进行了模拟实验,并从预测精度和训练时间两方面进行评价。实验表明:该模型的评价指标均方根误差(RMSE)和决定系数(R;)分别为0.429和0.953,其评价指标均优于其他对比模型。CNN;STM模型能够以较短的训练时间得到较高的预测精度,能为城市水管理提供技术支撑。
Real-time and accurate monitoring of river water quality is the primary task of urban water management strategy. Dissolved oxygen concentration is one of the important indicators for evaluating the quality of river water, and it is also an important condition for maintaining the survival of higher organisms in the water. Therefore, obtaining accurate and reliable dissolved oxygen prediction results is very important for river water management and early warning. First, the key water quality factors that affect the dissolved oxygen content of the water body, namely total phosphorus, ammonia nitrogen, permanganate content, and PH value,are obtained through gray correlation analysis. Then, as the convolutional long and short memory neural network(CNN_LSTM) is proposed for The extraction of deep and complex related features between dissolved oxygen and other water quality factors reduces the coupling of features between different information, improves the prediction accuracy of the model and reduces the model training time. A simulation experiment was carried out with the actual collection of dissolved oxygen data of the Changzhou section of the Beijing-Hangzhou Canal as the research object, and the evaluation was made in terms of prediction accuracy and training time.Experiments show that the root mean square error(RMSE) and coefficient of determination(R;) of this model are 0.429 and 0.953, respectively, and its evaluation indicators are better than other comparative models. The CNN_LSTM model can obtain higher prediction accuracy with a shorter training time, and can provide technical support for urban water management.
作者
袁金
徐宪根
Yuan Jin;Xu Xianggeng(Bell Tower Branch of Changzhou Ecological Environment Monitoring Center,Changzhou 210009;Changzhou Academy of Environmental Science,Changzhou 213022,China)
出处
《广东化工》
CAS
2022年第1期135-137,共3页
Guangdong Chemical Industry
基金
常州市科技支撑计划项目(CE20205037)。
