Continuous ultrasonic welding (CUW) is a manufacturing process used to join thermoplastic composite materials using high-frequency ultrasonic vibrations. Under pressure and vibration, the welding material heats up, melts and the welding partners are joined while they are cooling down. The development of CUW is getting raised attention over the last decade since the use of thermoplastic composites (TCs) in the production industry and especially the aerospace sector demands for economic and fast joining techniques. In order to establish this welding technique in the industry it is necessary to develop methods which enable to do quality assurance on the produced welding seams. In the experiments described in this paper the potential of thermal measurements for the quality assurance of the ultrasonic welding of TCs is investigated. Different thermal systems are evaluated and it is described how the data of the most promising technique is further processed and fed to an artificial intelligence (AI) algorithm. The here trained networks are created to predict lab-shear-strength values and reach mean absolute error scores of below 1 MPa. The described investigations hint that thermography seem to be a valuable resource to make more reliable quality assurance in the process of CUW of TCs and therefore might help to establish the use of this welding technology further in the industry.