摘要
提出基于小波变换和支持向量机的水质预测模型。该模型运用小波变换得到水质时间序列在不同尺度下的变化特性,并用改进后的粒子群算法优化回归支持向量机的三个参数,提高了模型预测精度。运用该模型对王江泾自动监测站测得的溶解氧浓度进行了1步预测及2步预测,10组测试样本最高MAPE为4.54%,并用基于BP神经网络的预测结果进行了比较。结果表明,该模型性能良好、预测精度高、简便易行,比基于BP神经网络的模型具有更好的预测效果,为水质预测提供了一种有效的方法。
The paper proposes a prediction model of water quality based on wavelet transform and support vector machine（SVM）.It uses wavelet to obtain characteristics of water quality time-series at different scales,also uses improved particle swarm optimization（PSO） to optimise three parameters of regressive SVM,which improves the prediction accuracy.The model is applied to 1-step and 2-step predictions of dissolved oxygen concentration measured at Wangjiangjing automatic monitoring station.The maximum MAPE of 10-group test samples is 4.54%,and this is compared with the prediction of BP neural network model.The results show that the model is of good performance,high precision,easy to use and has better prediction effect than the BP neural network model＇s,so it is an effective method for water quality prediction.
出处
《计算机应用与软件》
CSCD
2011年第2期83-86,共4页
Computer Applications and Software
基金
浙江省重大科技专项(2008C13017-2)
