Optimization of embedded sensor placement for small satellites

Abstract Dynamic environment testing is important in the manufacturing phase of a satellite before launch. However, redundancy is introduced into the testing by the massive placement of vibration sensors inside and outside the satellite, adversely affecting testing efficiency, especially in the development of satellite constellations. An intelligent algorithm-based sensor optimization method is proposed to maximize the relationship between measure points and reduce the number of internal sensors while ensuring the independence of external sensors. A frequency response matrix of internal and external sensors is constructed through a frequency-sweeping analysis of the satellite structure. Subsequently, external sensors are categorized into different groups by using principal component analysis (PCA) and K-means clustering analysis. A modified information entropy approach is applied to optimize the quantity of internal sensors. Through the practical application of sensor reduction in a specific satellite, the proposed method achieves a good reduction result.