文献类型：期刊文献

英文题名：Prediction of Aquaculture Water Quality Based on Combining Principal Component Analysis and Least Square Support Vector Regression

作者：Liu, Shuangyin[1,2,3];Xu, Longqin[1];Li, Daoliang[2,3]

机构：[1]Guangdong Ocean Univ, Coll Informat, Zhanjiang 524025, Guangdong, Peoples R China;[2]China Agr Univ, China EU Ctr ICT Agr, Beijing 100083, Peoples R China;[3]China Agr Univ, Beijing ERC Adv Sensor Technol Agr, Beijing 100083, Peoples R China

年份：2013

卷号：11

期号：6-7

起止页码：1305

外文期刊名：SENSOR LETTERS

收录：SCI-EXPANDED(收录号：WOS:000328005300053)、、EI(收录号：20134817021285)、Scopus(收录号：2-s2.0-84887964961)、WOS

基金：This paper was supported by the National Science and Technology Supporting Plan Project No. 2011BAD21B01-1, Guangdong Science and Technology Plan Project No. 2012A020200008 and No. 2012B091100431, and Guangdong Natural Science Foundation Project No. S2012010008261.

语种：英文

外文关键词：LSSVR; PCA; Water Quality Prediction; Dimension Reduction

外文摘要：The traditional methods about the water quality prediction in the aquaculture can't resolve the data redundancy and complex characters, which makes the accuracy of forecast precision low. In order to improve the accuracy of model prediction and the response time, this study demonstrates a new prediction method based on combing principal component analysis and least square support vector regression (PCA-LSSVR). The principal component analysis (PCA) method is applied to dimension reduction which reduces redundant data between various factors and extracting the principal components, then the data pre-processing is finished. The least square support vector regression (LSSVR) method is for modelling and forecasting to the principal components. Comparative experiment of the water quality is conducted on the base of river crab aquaculture ponds in Yixing. The experimental results show that the model of PCA-LSSVR outperforms standard LSSVR in performance on average, and the dimension of the sample set not only can be reduced and the water quality prediction accuracy can also be improved effectively. The research is proved to have stronger practical value.