The convergence of stochastic gradient iterative modal identification algorithm and application in modal analysis for space large manipulator

摘要

Nowadays the light constructors are widely used in the space large manipulators, and some properties of this kind of structure, such as low-frequency, large-range and durable vibration, have raise great attentions in aerospace area. In the paper we applied the wireless sensors networks (WSN) into the modal analysis of manipulators and proposed an interactive modal parameters identification method - SGIMID algorithm. This algorithm uses the stochastic gradient vector to improve interactive modal parameters identification (IMID) method, while the packet transmission mode between the modals is same to IMID algorithm. For the completeness of the proposed algorithm, we present the convergence proof of SGIMIDM. At last, we make applied SGIMID algorithm into the stiffness and frequencies coefficients identification of multi-DOF space manipulators. The experiment results show that, compared with IMID algorithm, the sensor nodes in SGIMID algorithm converge to the synchronous identified result with faster computing speed and less calculating burden.