计算机技术与发展
計算機技術與髮展
계산궤기술여발전
COMPUTER TECHNOLOGY AND DEVELOPMENT
2014年
4期
235-238,242
,共5页
郭庆春%郝源%李雪%杜北方%张向阳
郭慶春%郝源%李雪%杜北方%張嚮暘
곽경춘%학원%리설%두북방%장향양
神经网络%水质%化学需氧量%溶解氧%氨氮
神經網絡%水質%化學需氧量%溶解氧%氨氮
신경망락%수질%화학수양량%용해양%안담
neural network%water quality%COD%DO%NH3-N
水质变化具有非线性、突变性,且含有噪声,传统线性预测模型不能全面反映其变化规律,预测精度低,误差大。针对水质变化规律复杂,影响因素间非线性程度高的问题,为了提高水质预测精度,将改进算法的BP神经网络引入化学需氧量( COD)预测预报领域,以pH、溶解氧( DO)、氨氮( NH3-N)为输入向量,以COD为输出向量,建立了COD的预测模型并对效果进行检验。结果表明:检验样本中COD的预测值与实测值的线性相关系数为0.991。 BP神经网络模型预测精度高,收敛速度快,具有良好的泛化能力,能较好地反映COD和影响因子的变化规律。
水質變化具有非線性、突變性,且含有譟聲,傳統線性預測模型不能全麵反映其變化規律,預測精度低,誤差大。針對水質變化規律複雜,影響因素間非線性程度高的問題,為瞭提高水質預測精度,將改進算法的BP神經網絡引入化學需氧量( COD)預測預報領域,以pH、溶解氧( DO)、氨氮( NH3-N)為輸入嚮量,以COD為輸齣嚮量,建立瞭COD的預測模型併對效果進行檢驗。結果錶明:檢驗樣本中COD的預測值與實測值的線性相關繫數為0.991。 BP神經網絡模型預測精度高,收斂速度快,具有良好的汎化能力,能較好地反映COD和影響因子的變化規律。
수질변화구유비선성、돌변성,차함유조성,전통선성예측모형불능전면반영기변화규률,예측정도저,오차대。침대수질변화규률복잡,영향인소간비선성정도고적문제,위료제고수질예측정도,장개진산법적BP신경망락인입화학수양량( COD)예측예보영역,이pH、용해양( DO)、안담( NH3-N)위수입향량,이COD위수출향량,건립료COD적예측모형병대효과진행검험。결과표명:검험양본중COD적예측치여실측치적선성상관계수위0.991。 BP신경망락모형예측정도고,수렴속도쾌,구유량호적범화능력,능교호지반영COD화영향인자적변화규률。
Water quality change is of nonlinear and dynamicity,it is a kind of complex time series data,therefore,the traditional linear pre-diction model cannot reflect the variation rule,and the prediction accuracy is low. For the problems of complex water quality change rule and high degree of nonlinear between factors,in order to improve the water quality prediction accuracy,introduce the BP neural network of improved algorithm into a model of COD,with pH,DO,NH3-N as input and COD as output,the prediction model of COD is estab-lished and tested. The research results show the linear correlation coefficient of COD between forecasting and the monitoring in the test samples is 0. 991. BP neural network has high forecast precision,fast convergence rate and the good generalization ability,which can bet-ter reflect the change rule between COD and impact factors.