Study On Water Environment Assessment And System Simulation

Water pollution, which imperils the development and living conditions of human beings, has been a main subject in environment protection realm. The water quality monitoring and water quality assessment are one of the most important ways to manage and control the quality of water resources. As a first-phase work of water-quality improvement and water-resources management, they provide a practical scheme and scientific basis for the protection and integrated exploitation. Supported by the National Nature Science Fund project.(No. 60374033), the Neural Network models for water-quality assessment are established, based on Matlab, by analyzing the data monitored by multi-sensor on the ground and the multi-spectrum remote sensing image. The water-quality can be assessed completely from points (surveyed points) and lines (surveyed lines) to plane (water-area quality distribution).The main research contents in Thesis are listed as follows. First, the primary methods of water-quality assessment are introduced and their disadvantage and improvement are also showed here. At the same time, the theory of neural network and several Neural Network models of water quality assessment are studied deeply. Then, the BP network model is designed and established to analyze water quality, accordance with the ground survey data. Because it possesses the function of associative memory, self-organization, self-adaptation, self-learning and robustness et al, it has much more extensive adaptability, and the calculating result is much more dependable and objective, by compared with single factor assessment method and the method based on D-S theory. Another artificial Neural Network model is built and is succeeded in using gray degree values picked out from remote sensing image to inverse the concentrations of PH, BOD5, DO, CODMN. The remote sensing imagery data is made full use to complements ground survey data, so as to obtain more water quality information. Then, synthesizing the achievements in above research, a GRNN Network model is set to study water-quality space distribution. The concentration distribution of quota DO and its water quality degree space distribution in Tai lake are studied. The result suggests that applying generalized regression neural network model to research water quality space distribution could get perfect effect. Finally, the suggestions for further research content and method are put forward.