基于混沌神经网络模型的水库叶绿素a浓度短期预测

Chaos Neural Network Model for Short-term Predicting on Time Series of Reservoir Chlorophyll-a Concentration

通过混沌理论对水库叶绿素a浓度时间序列进行分析计算,得到最大Lyapunov指数为0.0218(正数),表明该时间序列具有混沌特性,可进行短期预测。同时,利用相空间重构的方法计算出时间延迟τ和嵌入维数m,并由此构建了可用于水库叶绿素a浓度短期预测的混沌神经网络模型。将该模型对于桥水库的叶绿素a浓度时间序列进行短期预测,平均预测相对误差为7.85%,取得较为满意的预测效果。该方法对水库的水环境管理具有一定的参考价值。

Time series of reservoir chlorophyll-a concentration were analyzed and calculated by chaos theory, with calculation results of the maximal Lyapunov exponent as 0.0218 for positive number, which indicated that the time series had chaos characters and could be short-term predicted. Time delay and embedding dimension were calculated through phase space reconstruction method, based on which the chaos neural network model for short-term predicting on the time series of reservoir chlorophylt-a concentration was constructed and applied to predicting on chlorophyll-a concentration of Yuqiao Reservoir. A reasonable forecasting result was achieved, with mean relative error as 7.85%. The method is instructive to water environmental management of reservoir.