Research And Application Of Hydrological Time Series Similarity Model

Analysis to time series is of growing importance in data mining and similarity search is one of the important aspects in time series research. Although traditional time series problems are discussed in many literatures,the research on time series similarity from a view of data mining arises have a gradual development just in recent 10 years. Many different methods and techniques about time series similarity search have been proposed and successful applications have been made in some fields,such as stock analysis. While these researches have various characteristics of their own,there also have many difficulties in forming a uniform and applicable method because of the complexity of time series. Practices of many methods and theories on large data set also need deeper research.With the increasing amount of data of time series in hydrological databases,it is very important in f100d forecasting and f100d dispatching to study the methods of retrieving similarity and then find the rules and tendencies contained in the hydrological time series. Based on the research and comparison of different methods,this paper explored the similarity search method of time series which is adaptive to the characteristics of hydrological data. The work of the paper mainly includes:(1) Present a model for measuring the similarity between two hydrological time series. In this model, we adopt an intuitive dimensionality reduction technique for hydrological time series which is called Piecewise Average Approximation (PAA).(2) Combining the methods of sliding windows and MBR,a kind of query based on subsequence similarity matching has been implemented,then we use MBR storage to take the place of dot storage in feature space and make R+-tree the multi-dimensional index structure.(3) Discuss the data preprocess techniques for hydrological time series. Filling missing values,smoothing noise data and removing inconsistent data are all adopted to gain high quality data.(4) Based on the model,design and implement a hydrological time series similarity search system. After some experiments on artificial data set and practical hydrological data set,validation to the availability and correctness of the model is presented.