Research On Real-time Early Warnin Methods And System Of Rivers Accidental Water Pollution Events

In recent years, accidental pollution events of river occur frequently, which imposes severe threats on the lives, property and safety of people and the sustainable development of the society. As for coping with the accidental water pollution events, it is urgent to study the technology of water quality forecasting and early warning, and to establish early warning system.According to the topics on the projects supported by the National Water Pollution Control and Management Projects (National Water Special Projects) and the Natural Science Foundation of China (NSFC), the research of real-time early warning technologies have been carried out, which is of great importance in dealing with the accidental water pollution events. The rolling warning methods based on mechanism and non-mechanism models have been investigated, and applied to the early warning system of water quality. Moreover, the parallel computing and task dynamic allocation technology have been presented, which improves the timeliness and accuracy of the early warning computing. The main research work and innovative aspects of the thesis are as follows:(1) In order to improve the timeliness and accuracy of early warning models for water quality when coping with accidental water pollution events, a rolling warning technology has been put forward. The rolling warning method has the access to make real-time modification on the model parameters so as to improve the accuracy of forecasting and early warning. Besides, the rolling warning towards the accidental water quality is realized on the basis of mechanism model. Such technology mainly adopts the thought of rolling computation in the forecasting control. The early warning of accidental water pollution is completed through five steps:data trigger, hydrologic data forecasting, model computation, model modification, and early warning evaluation. Under the framework of rolling warning, the real-time warning model of accidental water quality is developed based on the Mike software and the Saint-Venant equation. Furthermore, in accordance with the actual demand, the analysis of water evolution trends has been carried out. And the results of forecasting and early warning have been proved to get better effects. (2) In order to compensate the limitation of modeling difficulty and to improve the model adaptability which the mechanism model may present under complicated boundary, it is necessary to study the rolling warning methods of accidental water events based on non-mechanism model. In addition, on the basis of rolling warning methods, the research on the forecasting methods is implemented based on the ARMA time sequence model, grey model and neural network model respectively. On that basis, a combined forecasting method is put forward. In this method, the results of each water forecasting method are combined through weighting scheme. By this way, better results and effects have been gained in the short-term, medium-term and long-term early warning on the water pollution.(3) In order to overcome the problem that it is too slow to solve the equation of water quality, the research of a parallel algorithm has been carried out, which can reach relatively high speed-up ratio. The key point of the parallel algorithm is based on the task dynamic distribution scheme, and the parallel algorithm service with relatively good load balance and fault-tolerant feature is fulfilled. On that basis, an SOA framework and adapter design mode are presented and researched, and the asynchronous task execution method is studied. What’s more, the system integration gearing to the computation service is investigated and completed, which significantly improves the concurrency and availability of early warning system.