Generally, there are two categories of methods that are applied for weld quality inspection: on-line methods and off-line methods. As welding is such a complex process, there are many different physical phenomena such as electromagnetic radiation, optical and acoustic emission and plasma generation that will occur. Along with these phenomena, different LB42708? types of signals are generated during the welding process and usually these signals are closely associated with the weld quality. Therefore, different on-line sensing methods have been developed to monitor the weld quality and detect the weld defects based on the signatures of the signals acquired during the welding process. Fortunko et al. [2] applied an electromagnetic transducer to evaluate the quality of butt welds in a non-destructive way.

Park et al. [3] used optical sensors such as infrared (IR) photodiodes and ultraviolet (UV) photodiodes to detect the spatters generated during CO2 laser welding. Gu et al. [4,5] developed Inhibitors,Modulators,Libraries a statistical approach to predict the weld quality based on the spectrum analysis of acoustic signals. Huang et al. [6,7] applied microphone as the sensor to monitor the depth of weld penetration based on the acquired and processed airborne sound signals and established a neural network model to predict the depth of weld penetration based on extracted acoustic signatures. A sensor to monitor the structure-borne acoustic emission is also employed Inhibitors,Modulators,Libraries to detect interior weld Inhibitors,Modulators,Libraries defects such as cracks, porosity, and weld penetration based on acoustic emission signals at a frequency from 50 to 900 kHz [8].

These on-line sensing methods developed Inhibitors,Modulators,Libraries to inspect the weld quality and detect the weld defects heavily rely on the efficient processing and accurate interpretation of the signals from the welding process. For the purpose of weld quality inspection, these on-line methods are sometimes not reliable if the signals are contaminated by noises from hostile industrial environment and misinterpreted. Besides on-line methods for weld quality inspection, a number of off-line methods are also available. For different Brefeldin_A external and internal weld defects, as shown in Table 1, there are different types of off-line inspection methods. For example, inspection methods based on ultrasonic, radiography (e.g., x-ray), and electrical eddy current and magnetic particle inspection are effective for detection of interior weld defects such as crack and porosity [9].

On the other hand, as non-destructive testing method, vision-based inspection systems are always employed to detect external defects such as reinforcement, root concavity, undercut, sharp corner, incomplete filled groove, root thoroughly dropout, misalignment of the welded metal sheets, and partial penetration. Jeng et al. [10] used a CCD camera to inspect the gap and alignment for the laser butt joint welding.