The performance of current PZTs array and Lamb wave based SHM methods is restricted by the temperature variation. As temperature has a significant influence on Lamb wave propagation which leading to false monitoring result even if the structure is under health status. A new temperature compensation method based on adaptive filter adaptive liner neural (ADALINE) network has been developed to compensate for the amplitude-change and phase-shift of Lamb wave due to temperature change. Network construction and procedure of compensation are discussed. The advantages of this proposed method are (1) Simple to be applied, (2) only a few baselines are required for a large temperature range. Experiments are conducted on a stiffened carbon fiber composite panel to verify the temperature compensation method under temperature range from -40ÁC to 80ÁC. Results show the presented method is effective. Damage image combined with this compensation method can detect the damage under temperature variations.