Abstract
Traditional data mining algorithm had limited capacity at short-term hydrological forecasting with low accuracy, and made little use of the error between the data set and the results to correct the results. Considering of the traditional hydrology predictive algorithm combined only with the external associated factors, but had not fully excavated the predictive data itself, the BP neural network predictive algorithm with error feedback input was proposed. Because the algorithm makes full use of the relationship between the forecasting result and the system information entropy, it makes the forecasted results more accurate, and achieves a satisfied result.