The Prediction for BOD5 Based on Random GA-LM-BP Network
The BOD5 content is an important indicator of water quality; we took the BOD5 from Lun River of Xiaogan as the object of study.We established the random GA-LM-BP network to fit,test and predict the BOD5 content.LM-BP network has strong capability of study,but it is sensitive to its initial weight and threshold values.Thus its generalization capability is not strong.To overcome the shortcomings,we took genetic algorithm (GA) to optimize its initial weight and threshold values.To further improve the generalization ability of LM-BP network,we randomly generated the initial population of GA many times,and the GA optimized LM-BP network many times.The actual test indicated that the random GA-LM-BP network could 100% fit values.The tested errors were less than 5%,so the random GA-LM-BP network could predict the BOD5 content in the region,and it provided scientific basis for water quality forecast and water environment plan and management.The contrast experiments proved superiority of the method.
Genetic algorithm LM-BP network Testing error Water prediction
Xue-Mei CUI Hai-Ru GUO Ai-Qun ZHANG
School of Life Science and Technology,Hubei Engineering University,432000 XiaoGan,China;Hubei Key La School of Computer and Information Science,Hubei Engineering University,432000 XiaoGan,China
国内会议
杭州
英文
1-6
2014-10-18(万方平台首次上网日期,不代表论文的发表时间)