Infrared feature extraction and prediction method based on dynamic multi-objective optimization for space debris impact damages inspection

Abstract

This paper develops a damage-risk assessing method with dynamic feature extraction optimization (DFEO) in the Thermal-wave image technique, to evaluate the hypervelocity impact (HVI) damages in dynamic meteoroid/orbital debris (M/OD) environment rapidly. The proposed DFEO not only investigates how to select effectively each representative temperature point (RTP) with consideration of similarities and differences of thermal characteristic data, but also studies how to utilize historical information to extract rapidly impact damage feature from thermal image sequence. A multi-directional prediction method (MPM) in DFEO calculates the degree of environment change to depict the configuration of previous Pareto set (PS) and predict the moving directions of the optimal solutions. Meanwhile, a dynamic multi-objective particle-swarm optimization strategy (DM-PSOS) in DFEO guarantees that the obtained initial population evolves successfully to Pareto Front (PF). Besides DFEO, the assessing method contains variable step-size search, clustering analysis and damage visualization, to enhance the detection efficiency. The experimental study on HVI damage detection demonstrates the capabilities of the proposed method.