The Classification And Recognition Algorithm Of Satellite Variation-status Based On Similarity Matching

Nowadays, it is a hot topic in the field of aerospace to achieve the classification and recognition of the satellites’ state based on the research of satellite telemetry data. The technology used to analyze the satellite telemetry data at present is mostly based on mathematical models of data or expert system. However, both the two methods have limitations. For example, it is unable to use a unified mathematical models to deal with various different types of satellite telemetry data. And expert system requires a complete knowledge base, but it is difficult to maintain a knowledge base. To solve these problems, this thesis focuses on the research of classification and recognition algorithm which is data-driven and is not sensitive to mathematical models.First of all, in order to get the characteristic parameters of multiple types of satellite telemetry data, and to design a universal algorithm, this thesis studies the algorithm of extract characteristic parameters based on the data(Including the characteristics of time and frequency domain, statistics and other transform domain). Then, this thesis applies the K-means algorithm and Fuzzy c-means algorithm to the classification of satellite states, and presents an improved algorithm based on Fuzzy c-means algorithm.Finally, this thesis designs an identification algorithm based on similarity matching. And conducts an experiment to verify the algorithm.The main achievements and contributions of this thesis are:1. This thesis uses Wigner-Ville distribution time-frequency analysis, and Wavelet transform theory to extract characteristic parameters of satellite telemetry signals, on the condition that lack of a priori conditions.2. This thesis puts the measured data as a time series, and uses the theory of fractal dimension to describe the fractal characteristics of the measured data quantitatively. This is an innovative point in this thesis.3. The thesis designs two kinds of classification algorithms, one is based on K-means algorithm, and the other is based on Fuzzy c-means algorithm. And presents an improved algorithm based on Fuzzy c-means algorithm.4. The thesis designs a recognition algorithm of variation-status. And uses the results of classification algorithms to implement the recognition of the satellites’ status.5. The thesis uses the algorithms which designed previous to analyze the measured data, and implements the classification and recognition of the satellites’ status. And analyzes the experimental results.6. This thesis develops a software which contains all the functionality in thesis, and it can be used to analyze satellite data.