人民黄河
人民黃河
인민황하
Yellow River
2014年
11期
20-21,25
,共3页
凌文韬%谢利云%何玉琛%张伟
凌文韜%謝利雲%何玉琛%張偉
릉문도%사리운%하옥침%장위
Mann-Kendall法%BP神经网络%趋势预测%径流%黑河流域
Mann-Kendall法%BP神經網絡%趨勢預測%徑流%黑河流域
Mann-Kendall법%BP신경망락%추세예측%경류%흑하류역
Mann-Kendall method%BP neural network%trend prediction%runoff%Heihe River basin
根据黑河1956—2012年各水文站径流量数据,应用Mann-Kendall非参数检验法对黑河流域径流变化规律进行了分析,并利用BP神经网络对未来黑河径流演变趋势进行了预测。结果表明:①黑河干流及主要支流汛期来水量占年径流量的比例较大,为63.18%~94.56%,出山口径流量未来总的趋势为增加;②Mann-Kendall趋势检验结果表明莺落峡、祁连、新地及嘉峪关站未来径流预测趋势与实测序列趋势较为一致,而札马什克站未来径流预测趋势和实测序列趋势相反,但该站M-K值的绝对值逐渐变小并趋于0,可以推测未来该站径流量趋势将由减少转为增加,这与BP神经网络预测的结果一致。
根據黑河1956—2012年各水文站徑流量數據,應用Mann-Kendall非參數檢驗法對黑河流域徑流變化規律進行瞭分析,併利用BP神經網絡對未來黑河徑流縯變趨勢進行瞭預測。結果錶明:①黑河榦流及主要支流汛期來水量佔年徑流量的比例較大,為63.18%~94.56%,齣山口徑流量未來總的趨勢為增加;②Mann-Kendall趨勢檢驗結果錶明鶯落峽、祁連、新地及嘉峪關站未來徑流預測趨勢與實測序列趨勢較為一緻,而札馬什剋站未來徑流預測趨勢和實測序列趨勢相反,但該站M-K值的絕對值逐漸變小併趨于0,可以推測未來該站徑流量趨勢將由減少轉為增加,這與BP神經網絡預測的結果一緻。
근거흑하1956—2012년각수문참경류량수거,응용Mann-Kendall비삼수검험법대흑하류역경류변화규률진행료분석,병이용BP신경망락대미래흑하경류연변추세진행료예측。결과표명:①흑하간류급주요지류신기래수량점년경류량적비례교대,위63.18%~94.56%,출산구경류량미래총적추세위증가;②Mann-Kendall추세검험결과표명앵락협、기련、신지급가욕관참미래경류예측추세여실측서렬추세교위일치,이찰마십극참미래경류예측추세화실측서렬추세상반,단해참M-K치적절대치축점변소병추우0,가이추측미래해참경류량추세장유감소전위증가,저여BP신경망락예측적결과일치。
According to the runoff data of each hydrological station of Heihe River basin in 1956—2012,non-parametric Mann-Kendall method was applied to analyze the trends of runoff change,and the future runoff trend was predicted by using BP neural network. The results show that a) the proportion that water to flow of main stream and main tributaries in flood season accounted for the annual runoff is larger,which is 63. 18% ~94. 56%,and the runoff which is out of the mountains in Heihe River basin will increase;b)the test results of Mann-Kendall trend show that the forecasting future trends are coincident well with the measured results in the stations of Yingluoxia,Qilian,Xindi and Jiayuguan. However,the forecasting trends and the measured results are opposite in the stations of Zhamashike,and the absolute value becomes smaller and tends to 0,which can speculate the future runoff will increase in this station,and that is in line with the prediction of BP neural network.