The Ching River Hydrological Research Based On Artificial Neural Network

Hydrologic research of river reach is an important subject in Hydrological Sciences, and flood-control project is the key target of hydrologic research, as a central part of hydrological studies-runoff forecasts, it is not only an important flood-control measure aimed at reducing the loss of the disaster areas, but also an method of scientific utilizing water resources, so more and more hydrological experts at home and abroad pay attention to it. Timely and accurate hydrological forecast, on one hand, it can provide an important reference for flood-controlling, scheduling and making decision, so as to reduce the risk of flood disaster; on the other hand, it can make rational utilization of water resources and bring considerable economic and social benefits to the region.At present, aiming at such complex "black box" issue, the majority of traditional ways are seeking linear relationships between input and output, But Artificial Neural Network is directly considering non-linear relationships between input and output by converting function, the latter is more suitable for movement characteristic of flood theoretically. Artificial Neural Network have non-linear and learning properties, but it also exits some disadvantages of slowly convergence and easily occurred concussion.Therefore, based on the Ching River as the research object, the main research contents are as follows:(1) Describe hydrological conditions of Ching River briefly, Compare traditional and new methods of hydrological forecast, then Artificial Neural Network was imported into the model, and introduce the current research situation of Artificial Neural Network model in hydrological study.(2) In this paper, we put forward some methods for further study based on the universal relevance of the BP Neural Network model, then Tabu Search was imported into the model, we established the model that global optimization can be achieved and improving the forecast precision of tabu network for global optimization, and we do the research on network structure design, algorithms and parameters setting of Tabu network for global optimization.(3) At the last procedure, the daily forecasting models of BP Neural Network and Tabu Global Optimization Network were constructed by simulation software MATLAB, learned and trained with hydrological data over the years from Ching River section. The results show that the forecast accuracy of Tabu Global Optimization Network is superior to BP Neural Network. Furthermore, the monthly forecast model was created with considering the mid-long term load forecasting plays an vital role in reservoir dispatching, flood control and relief, as well as water resources management, after analyzing, the forecast accuracy of Tabu Global Optimization Network is obviously better than BP Neural Network.(4) This paper has reasonable models, good effect of hydrological forecast, it is easily operability, the works have important reference value on hydrology of the Ching River Basin.