Historical Data Compression Algorithm. Real-time Database

Real-time database is the combination of real-time system and database technology. Real-time database has the characteristic of high speed, high data throughput and so on. It has been widely applied in automation field. Now, historical data storage is provided in most databases. Beacause of large number of acquisition points and short time period in control system, the amount of historical data is very large. Directly saving will cost vast space and make data transmission difficult.How to not only keep the characteristic of high speed, but also to increase the capacity of database? To supply good data structure, index structure and high-performance compression. There are three kinds of data compression technologies in real-time databae : loseless compression, lossy compression and two-stage compression. Lossless compression is generally on the base of generic compression theory, using classical algorithm such as Huffman compression algorithm. Lossy algorithm takes more consideration to the characteristics of the industrial real-time data, adopting some special points algorithm. Two-stage compression both use the two data compression technologies at the same time.This paper designed a historical data file system with B-tree index based on the features of the data structure in Intelligent Building Management System, and gives an efficient two-stage data compression scheme. Then deeply studied the various algorithms from generic loseless compression to numerical lossy compression methods, presents a mean square algorithm improved from the well-known swinging door algorithm developed by OSI softwrae Inc. which has a higher data compression ratio and can act as decreasing effect of measurement noise on data compression, it was applied to the first compression of the simulation data; furthermore, a new LZW data compression algorithm which has two aspects(the dictionary storage and storage length) of improvments from the classical LZW algorithm is applied to the secondary historical data compression algorithm with high quality. At the same time use a highly efficient application of the RLE algorithm in the state and time with the label features.In the end, the foreground of the model of historical database and its compression methods is discussed.