【作者】 何欣；

【导师】 易善桢；

【作者基本信息】 华中科技大学 ， 水利工程（专业学位）， 2022， 硕士

【摘要】 水质预测作为水资源保护的基础性手段,能够反映水体水质的变化趋势,为做好高质量水资源统筹规划提供技术支持,有助于缓解水资源危机,推进生态环境修复。水环境是一个极其复杂的灰色系统,内在变化规律未完全可知,水体中各水质参数的变化具有非线性和不确定性的特点。长短期记忆神经网络(LSTM)以其出色的时序处理能力在水质预测中有着广泛的应用。为了进一步提升水质预测的精确性和适用性,本文基于深圳市茅洲河监测站点的水质数据,提出了一种将变分模态分解(VMD)与LSTM相结合的水质预测方法,以实现对河流水质的高精度预测。本文的主要工作如下:(1)在深入了解当前水质预测工作所用方法的基础上,基于对茅洲河水质预测的适用性分析,设计并构建了基于遗传算法优化变分模态分解的LSTM水质预测模型,即GA-VMD-LSTM模型。主要通过VMD算法对复杂的水质数据进行分解,得到更为简单有序的子序列以降低后续预测难度,并使用遗传算法(GA)优化VMD参数以获得最优分解效果;对分解得到的子序列根据平稳性特征进行不同模式的LSTM预测,以最大利用子序列的特征信息,最后将所有子序列的预测结果相加得到最终的预测值。(2)在深圳市茅洲河共和村水质监测断面对本研究所提的GA-VMD-LSTM模型进行了实验验证。选择水体中的溶解氧、p H、高锰酸钾指数和氨氮作为预测指标,实验结果显示,以上四种水质指标均有较高的预测精度,表明本文所建立的模型在深圳市茅洲河的水质预测上具有适用性。(3)为进一步验证模型性能,将本文所建立的模型与其他模型进行对比验证分析。结果表明,无论是与BP、RNN、LSTM等单一模型相比,还是与EMD-LSTM、EEMD-LSTM、GA-VMD-BP、GA-VMD-RNN等组合模型相比,本文所建立的模型整体的预测精度更高,预测性能更好,证明了本文模型的有效性。更多还原

【Abstract】 As a basic means of water resources protection,water quality prediction can reflect the change trend of water quality,provide technical support for the overall planning of highquality water resources,and help alleviate the water resources crisis and promote the restoration of ecological environment.Water environment is a very complex grey system,and its internal variation law is not completely known.The variation of water quality parameters is nonlinear and uncertain.Long and short-term memory neural network(LSTM)has been widely used in water quality prediction due to its excellent time sequence processing ability.In order to further improve the accuracy and applicability of water quality prediction,based on the water quality data of Maozhou River monitoring station in Shenzhen city,a water quality prediction method combining variational mode decomposition(VMD)and long and short-term memory neural network(LSTM)is proposed to achieve high precision prediction of river water quality.The main work of this paper is as follows:(1)Based on the analysis of applicability of water quality prediction of Maozhou River,a LSTM water quality prediction model based on genetic algorithm optimized variational mode decomposition(GA-VMD-LSTM)was designed and constructed.The complex water quality data were decomposed by VMD algorithm to obtain more simple and ordered subsequence to reduce the difficulty of subsequent prediction.Genetic algorithm(GA)was used to optimize VMD parameters to obtain the optimal decomposition effect.LSTM prediction of different modes is performed for the decomposed sub-sequences according to the stationarity characteristics to maximize the feature information of the sub-sequences.Finally,the prediction results of all sub-sequences are added to obtain the final predicted value.(2)The GA-VMD-LSTM model proposed in this study was verified experimentally in the water quality monitoring section of Maozhou River,Shenzhen City.Dissolved oxygen,p H,potassium permanganate index and ammonia nitrogen in water were selected as the prediction indexes.The experimental results show that the above four water quality indexes have high prediction accuracy,indicating that the model established in this paper is applicable to the water quality prediction of Maozhou River in Shenzhen city.(3)To further verify the performance of the model,the model established in this paper is compared and verified with other models.The results show that compared with the single model such as BP,RNN and LSTM,or the combined model such as EMD-LSTM,EEMDLSTM,GA-VMD-BP and GA-VMD-RNN,the prediction accuracy of the proposed model is higher and the prediction performance is better,which proves the validity of the proposed model.更多还原