Damage detection technology based on guided wave relies on comparing response signals with baseline records. However, the temperature variation can also produce large changes in guided wave signals, thereby causing a false alarm of damage detection. For eliminating the influence of temperature variation in the process of damage detection, independent component analysis(ICA) method is developed to process guided wave response signal. As a blind source separation algorithm, ICA can extract independent unknown source components from mixed signals. Therefore, the ICA method can separate the source signals of the guided waves from the mixed signals disturbed by the temperature, and achieve the purpose of eliminating the interference of temperature changes. In this paper, the experiment of bolt looseness damage detection is made in the environment of temperature variation. The collected signals are utilized in damage localization estimation after processing by ICA. It shows a good result for bolt looseness localization, which indicates that this method is robust to the temperature variation exhibited in damage detection based on guided wave.