MACHINE LEARNING-BASED HYDROLOGIC FORECASTING PRECISION EVALUATION METHOD AND SYSTEM

摘要

A machine learning-based hydrologic forecasting precision evaluation method and system, belonging to the field of hydrologic forecasting precision evaluation. The method comprises: obtaining a hydrologic forecasting result and a measured result at the same period of the same section, to calculate a floodpeak evaluation index, a floodvolume evaluation index and a runoff process evaluation index, to form a piece of evaluation index data; using a trained hydrologic forecasting precision evaluation model to identify a precision level to which the evaluation index data belongs, and using same as a hydrologic forecasting precision evaluation resu the hydrologic forecasting precision evaluation model being a machine learning classification model. The division of precision levels comprises: according to historical data, respectively calculating evaluation index data corresponding to each historical flood event as historical evaluation index data; clustering the historical evaluation index data to obtain C categories, which respectively correspond to C precision levels; and according to the evaluation index level within the categories, ranking the corresponding precision levels in order of preference. Said method and system are able to implement an accurate rating on hydrologic forecasting precision.