Time Series Data Mining Technology And Its Applied Research In The Prediction Of Water Quality

Time series data mining is is the search for underlying and useful information in large volumes of time-series data set and using the information to predict the time-series future. In this dissertation, the research of the time series data mining technique and its application in forecasting the water-quality, taken as a part of the projections of the Decision and Support System of water-environment (the Science and Technical Study Projection of Guangdong) and the Modeling and Control of time-delay large system in the water-pollution (the National Natural Science Foundation of China), is proposed. The main contents of this dissertation include the study of the trending predictive technique based on time-series patterns and association rules, the research of the time-series mining technique based time-series trending structure series, the research of the time-series mining technique based rough set and time-series trending structure series, and the application of time series data mining technique in the water-quality prediction. The main work and the primary results and hard core of this dissertation are summarized as following:(1) In the chapter 1, the background and the studying actualities of the time series are summarized, and the valuable and significance about the researches of the doctor dissertation is dissertated. Finally, the structure and the study contents of the dissertation are given.(2) In the chapter 2, the conception and the process of data mining are introduced. Finally, the conception and the discovery of association rule, and the meaning and the approaches of classifying discovery are summarized.(3) In the chapter 3, the conceptions of the rising time sub-series and the falling time sub-series are first defined, and then a new trend forecasting approach based on time-series rules and patterns is presented. For the giving time series data, the technique is that the author converts the time-series data into a time sub-series data set first, and then investigates this time sub-series data set, dig the rising time sub-series data set and the falling time sub-series data set mainly, and extract the valuable rules and patterns from them. Finally, the time-series trending predicting techniques based on rules or patterns and based on support or confidence are proposed. And the mining algorithm is given.(4) In the chapter 4, the conceptions of the time-series trending structure series, the latest time sub-series, the trending structure being same, the trending prediction based on confidence and support are defined. The author regards the latest time sub-series as the information collection of the time series. Basing on this point of view, the author proposes a new time-series data mining approach based on its time-series trending structure series. The mining technique is that the time series set waiting for mining is first converted into its trending structure series set, and then the information underlying in the latest time sub-series of the time-series trending structure series is...