最优加权组合预测法在水质预测中的应用研究

针对用再生水补给河流的水质预测问题,提出了基于最优加权法的组合预测模型.利用灰色预测模型、趋势外推法和指数平滑法3种方法分别进行预测,再以预测误差平方和最小为目标,将预测值的加权问题转化为优化问题,求解得到各种方法预测值的权值.然后,将3种方法所得的预测结果用最优加权法进行组合,得到组合预测值.最后,应用组合预测法对由再生水补给的永定河中的DO值进行预测,并与单一预测模型比较.仿真结果表明,组合模型可以平衡各种方法的偏差,模型的适用性和预测精度有所改善.

Application of weighted combination model on forecasting water quality

For water quality in the river recharged with reclaimed water, a combination forecast model was built based on the optimal weights to forecast the water quality. Three different models were developed, including gray forecast model, trend extrapolation model and double smoothing model. By setting minimizing the sum of squared prediction error as the goal, weighting the predictive values problems was transferred into an optimization problem. The weights of predictive values for three methods could be obtained by solving the optimization problem. Afterwards, the results were combined based on the optimal weights and thus the combination forecast values were obtained. Water quality of Yongdinghe River recharged with reclaimed water was forecast based on the combination forecasting model, compared with the three simple models. The results showed that the combination forecasting method could balance the bias of the single models, and the applicability and accuracy were improved to some extent.