Water Quality Prediction Model
Objective
In recent years, environmental pollution has become a globally recognized important issue, and water quality pollution is one of the most important aspects. River is a common water body in people's life and production, and it is also one of the main carriers of environmental pollution. Therefore, the study of dynamic monitoring and prediction modeling of river water quality is particularly important.
Methodology and Process
This paper uses neural networks to predict the future degree of water pollution and the proportion of pollutant content, to analyze the regularity of information hidden behind the water quality data, to provide scientific decision-making basis for the prevention and control of water pollution, and to further play the function of bacterial and algal symbiosis for water purification.
In this paper, neural network modeling is carried out from two perspectives: water content analysis and water quality category analysis.
The basic principle of Applied Hierarchy Processing (AHP) is a decision-making method that decomposes a complex decision-making problem into multiple levels, and determines the optimal solution by comparing and ranking the criteria and factors at different levels.
For the water content analysis, after data preprocessing, a feed-forward neural network model is created using MATLAB and the model is trained according to the input data and target data, and the trained model is used to predict the input data and obtain the prediction results. In order to visualize the expression, the actual data and the predicted data are shown in the form of graphs. In addition, based on the prediction results and actual data, the model evaluation indexes such as root mean square error (RMSE), root mean square percentage error (MAPE) and mean absolute error (MAE) of the improved model are calculated.
The microbial embedding device is an experimental device that allows for the rapid acquisition of algal spheres of the desired diameter by adjusting the infusion rate and the caliber of the injection head, whereas the conventional preparation of algal spheres requires human control.
For water quality category analysis, a neural network model is constructed using Python's machine learning library (Keras), which builds a multi-layer neural network by means of the Sequential model. After the model is constructed, the loss function, optimizer, and evaluation metrics are specified using the COMPILE function, and the model is trained using the FIT function. After the training is completed, the model can be used for prediction. The predict function is used to predict the test data, predict future changes in each pollutant and visualize them.
In this paper, the neural network model is used to model the water body content analysis and water quality category analysis from two perspectives respectively, and the visualization and evaluation indexes are used to show the prediction effect and accuracy of the model.
Analysis of water content
The projections show trends in dissolved oxygen content, ammonia nitrogen content, chemical oxygen demand, total nitrogen, total phosphorus, nitrate content, and nitrite content for five cross-sections from 2023-06 to 2023-12.
Dissolved oxygen content analysis:
Ammonia and nitrogen levels were analyzed:
Similarly, the trends of chemical oxygen demand, total nitrogen, total phosphorus, nitrate content, and nitrite content are shown above.
Water quality category analysis
With the deep learning model, a set of known water quality features are input, the model learns the relationship between the features and water quality categories, and training is used to adjust the model parameters. The model was then used to predict new water quality features for possible water quality at each cross-section in May 2023, and the results are shown below.
Conclusion
Sum of the five sections of the future July data data, dissolved oxygen is relatively stable, ammonia nitrogen content decreased, total nitrogen and total phosphorus are slightly elevated, nitrate showed fluctuations, while the amount of nitrite is significantly higher. In the future, the pollution level of Sturgeon Po cross-section, Min'an cross-section, and Yunxiao Gaotang Ferry cross-section will increase, while the pollution level of Fuan Saiqi cross-section and Mulan Creek Sanjiangkou cross-section will be relatively lower.
Testing of the model
Compare the predicted results with the actual situation in May 2023.In May 2023, the water quality of Jinjiang River is Class III at Sturgeon Po cross-section. The water quality of Minjiang River at Min'an cross-section is Class II. The water quality of Zhangjiang River is Class III at Gaotang Ferry Crossing in Yunxiao. Fuan Saiqi cross-section, the water quality of Jiaoxi River is class III. The water quality of Mulan Creek at Sanjiangkou is Ⅳ.
The comparison shows that there are still some deficiencies due to the early establishment of the model. However, except for the Min'an and Fu'an Saiqi cross sections, the water quality categories of other cross sections were accurately predicted, and the probability of correctly predicted water quality categories was more than 77%. It can be seen that the model has a good prediction effect.
Summary
A neural network model was used to predict and analyze the water quality from June 2023 to December 2023.The prediction accuracy was higher in June-September compared to October-December, and the pollutant concentrations showed a seasonal trend of high in winter and spring and low in summer and fall. This is mainly due to the fact that the factors affecting water quality changes are not absolutely constant, and environmental problems and pollution generated in the upstream will be transmitted to the downstream watershed system, resulting in corresponding changes in water quality.
This paper helps the experimental group to design suitable biological apparatus to deal with pollutants by predicting future river pollutants, preventing excessive nitrogen and phosphorus content, preventing over-pollution of the river, providing scientific decision-making basis for the prevention and control of water pollution, and further giving full play to the function of bacterial and algal symbiosis for water purification, in an effort to safeguard the sustainable use of water resources, and to promote the process of environmental protection and sustainable development.
