AN ANN EXPERT SYSTEM FOR ICE CONDITION FORECAST OF YELLOW RIVER INNER MONGOLIA REACH

摘要

An ANN, Artificial Neural Network, Expert System is presented for ice condition forecast of the Yellow River Inner Mongolia Reach. The system is based on GIS technique while its kernel is BP network arithmetic. Using the expert system, the ice conditions in the 2004-2005 ice period, such as the water temperature and the beginning dates of the ice run, freeze-up and break-up etc, were forecast on-line. The forecast period is in 15 days. All the forecasts are qualified, of which the forecast for the beginning dates of the ice runs and break-ups is excellent, the errors between the forecast results and observed ones are less than 2 days.