Flexible Integration Of Hydrological Time Series Basic Algorithm Components

There are different approaches to dynamically assemble different algorithms in the field of hydrological time-series data mining at present. The purpose and function of the past data mining system developed based on a certain type of algorithm is so single that it is very difficult to extend, to update and to transplant to different operation systems. What's more, its maintenance expense is also very high and it is difficult to make full use of those different algorithms by assembling them together. In a word, it is in great need of the flexible data mining system, which is flexible, extensible, transplantable, easily maintained and handled, to be used in the field of hydrological time-series data mining.Through the analysis on the process of hydrological time-series data mining, we can find that there are many similar parts in this process in spite of the complex composition of business logic application. Therefore, the process of time series comparability search is composed of three continuous phases, such as data pretreatment, characteristic mode mining and comparability measurement. There are a lot of different algorithms performing same function in every data mining phase. It is obviously illogical to develop a single system according to a single algorithm. Thus hi this paper, we introduce a blue-print about flexible integration of hydrological time-series data mining components based on the J2EE platform. After carefully researching and analyzing on the rationality and feasibility of the blue print, we also design a flexible integration system of hydrological time-series basic algorithm components (HTDMS V1.0), which realizes the aim of flexible integration of basic algorithm components and validates the correctness and reliability of the system.What we do in this paper goes as:1. We study and analysis the mathematic background of related basic algorithms about hydrological time-series data mining, including the typical algorithms of data pretreatment, characteristic mode mining and comparability measurement2. According to the specific of EJB 2.0, we introduce the abstract class interface and component specifics of basic algorithm components, design the algorithm component manager and describe detailed flexible integration mechanism of basic algorithm components.3. According to the characteristics in the field of water resource, we design a flexible integration frame and analysis its structure and function.4. Based on the J2EE platform, we integrate the above component technique andintegration frame to design a flexible integration system of hydrological time-series basic algorithm components about the water field. At last, we validate the validity and correctness of the flexible integration mechanism and frame.