基于支持向量机的水质检测数据融合研究-模式识别与智能系统专业论文.docxVIP

中国科学技术大学硕士论文 中国科学技术大学硕士论文 Abstract Abstract With the rapid development of society and economic，water pollution situation is becoming more and more serious．There strong requirements for water quality multi—parameter intelligent monitoring system Water quality multi—parameter detection，a interdiscipline，involves multi—sensor data fusion technology,computer technology,electrochemistry analysis and artificial intelligence，etc．，it has been widely applied in many fields The primary works ofthis dissertation summarized following： 1．An on—line weighted adaptive data fusion algorithm based least squares support vector machines： Multi—sensor data fusion(MSDF)has been used widely for its complementary and redundant information，and it greatly enhance system’s reliability Data fusion algorithm is key for MSDF In order to the defects oftraditional data fusion algorithm，it proposes on—line weighted adaptive data fusion algorithm based least squares support vector machines in this dissertation．This approach is used to detect water quality line，as result，the training time is shorten and the reliability and flexibility of data fusion are enhanced 2．A self-calibration of ion is proposed based least squares suppoa vector machines： Ion non—linear,drift and intercross-sensitivity impact its accuracy and reliability obviously,SO that，it is difficult for majority of ion to continuously detect in situ In order to detect dynamic environment line，self-calibration of NO；ion is investigated and its response properties are analyzed in term of the experimental data．The and time drift taken into account，a self-calibration of ion is proposed based least squares suppoft vector machines(LSSVM)，the description and analysis given out in detai 3 A method of Ion fault diagnosis based wavelet support vector machines feature extraction and support vector machines’classification The requirements of water on—line detection need high reliability of ion sensor，that is，ion fault be estimated in time