【摘要】 依据开河机理,结合松花江某水文站的历史资料,建立了该站开河日期预报的BP神经网络模型.应用Levenberg-Marquardt算法改进了BP算法,采用循环递进的方法建立了实用预测模型.经实际流凌开河日期预报的验证表明,循环递进方法建立的实用预测模型能充分利用预报因子的信息和神经网络方法的非线性映射能力,预测精度高,合格率为100%.更多还原

【Abstract】 A river ice condition forecasting model was established based on BP neural network,whose training method was Levenberg-Marquardt algorithm.The model was to forecast the break-up date of the station on the Songhua River,based on the mechanism of the break-up and the ice condition statistics of the station.The practicable model was also founded with the iterative method,and the test results of the model show that the pass rate was 100%.This practical model was stable and precise due to its adequate utilization of relevant information of predicated factors and nonlinear mapping ability of the neu-ral network method.更多还原