Multi-sensor Data Fusion Technology In The Interval Estimation

In the system identification and error analysis, Setmembership estimation interval analysis algorithm become effective methods of estimation. It become more and more popular in recently years, It is the most features: Assuming the system noise is a random process, It not need to know the interference of noise or noise power distribution case (unknown but bounded - Unknown But Bounded--UBB), according to the system input and output information provided by the to identify a parameter space with the observational data and a member of a collection of noise-compatible (or range). The range always contains the true value to be recognizable. The setmembership estimation interval algorithm research in a number of engineering has been applied.In some respects have made dramatic breakthroughs. Although the setmembership estimation theory, its application research has some results, these studies are not perfect. This subject has researched and analysisde of existing set membership is estimated based on the results on the existing range of set membership estimation algorithm is proposed to improve the range, a new analytical method - using multi-sensors data fusion technology for interval data pre-processing, to get a firmer estimate of range. This paper is divided into the following several sections:The first part: The main describes the setmembership estimation theory of the origin, principles, development and research status. On this basis, the introduction to an important set membership estimation algorithm—interval algorithm (box algorithm). Then a detailed description of the mathematical theory of interval-based algorithms. And focused on the analysis to a linear time-invariant system set membership interval analysis algorithms. And gives a detailed recursive process.The second part: the main introduces multi-sensor information fusion theory and research status. An emphasis on data fusion theory, composition, integration methods and integration of the means of a detailed introduction. A detailed analysis of the single-sensor and multi-sensor data fusion, and its integration of the simulation results comparedThe third part: the focus of this research project. On the basis of the above-mentioned two kinds of algorithms is proposed a new algorithm - data fusion in the range of set membership estimation algorithm applications. Studied in detail how to use multi-sensor fusion is a more effective and more comprehensive data. And detailed proof of the multi-sensor data collection can effectively reduce the total mean square error. And then use the data to be re-estimated under the set membership interval algorithm for system identification, so that identification of the range to be more compact, simulation experiments show that the new algorithm of interval optimization is effective. Finally, future research in this topic to put forward demands, of course, hope that the integration interval estimation algorithm has a certain degree of stability, that is, that the fusion center for the importation of a small "disturbance", its output has a strong robustness and other issues. There is a multi-sensor coupling between errors of data collected large amount of data in the form of diversity issues, these issues will affect the recognition accuracy of the estimates, and fast. Therefore, these have yet to be further research and explore the future.