| Popis: |
With the increasing number of space debris and extreme working environment, the space station faces the risk of cabin damage. Once the leakage occurs, it must be repaired in time. Identifying the shape of leaks can guide for astronauts to make plugging plans. The current research on leak identification methods is mostly aimed at circular leaks with different diameters, and there is little research on leak identification with different shapes. Therefore, it is of great significance to research leak shape identification methods. A method for identifying the shape categories of vacuum leaks based on improved variational mode decomposition and support vector machine is proposed in this paper: first, using the improved algorithm to determine the uniform number of modes K and adaptively optimize the quadratic penalty coefficient α. Then, apply the variational mode decomposition to the leakage signal and calculate each mode’s maximum frequency. Setting these frequencies as eigenvalues for training and testing the identification model is based on support vector machine. In the experiment, four shapes of the leaks were used: ellipse, rectangle, round, and square. The experiment proves that this proposed method has stable and high identification accuracy, and can realize the shapes categories identification for the leaks, which can provide an important basis for spacecraft in-orbit leakage maintenance. |
