Depth Of Indentation Prediction Model In Resistance Spot Welding

Post-welding test, as well as stabilizing welding parameters, was employed for welding quality guarantee in traditional production. However, the visual measurement, which was generally used to inspect the appearance quality of welding spot, meets numerous problems and consequently results in less production efficiency. Indentation of welding spot, which is closely related to welding spot quality, is one of the indicators for evaluating the welding quality. Result showed that large amounts of informtion about welding spot quality was contained in the electrode displacement signal and the surface image of welding spot. In this thesis, taking predicting depth of indentation as aim, by using the electrode displacement signal and surface image of welding spot as information sources, based on SVM, the following detailed research work was carried out:(1) Established a signal sampling system centering on laser displacement sensor, DA-5 displacement sensor, AC6115AD card and Rogowski coil to obtain the welding current, electrode voltage and electrode displacement signal in welding process. The electrode displacement signal from laser displacement sensor can be employed to characterize nugget formation process since data obtained has good repeatability and can reflect accurately how the welding current and electrode force acts on nugget formation;(2) Established a measurement platform centering on laser sensors, measurement devices and computer systems to obtain the actual depth of indentation. Result showed that welding current and electrode force are the main factors to influence depth of indentation. The parameter h, which reflects indentation depth, can be extracted from electrode displacement signal.(3)The digital image was segmented respectively by the iterative method and the OSTU method then. By contrast, satisfactory binary image can be obtained by the iterative method. The characteristic parameters-S₁, S₂, S, l₁, l2-which reflect the surface image of welding spot, and among which the S₂, S, l₁ can be used to predict solder joints because of their reflection of indentation depth.(4) Based on SVM, established a model to predict indentation depth. In this model, the electrode displacement characteristic parameter (h), welding current (I) and electrode force (F) are inputted parameters, the actual indentation depth (h_T) is outputted parameter. The correlation coefficient between model predictions and hT came up to 0.8955. In another model, the surface image characteristic parameters (S₂, S, l₁) , welding current (I) and electrode force (F) are inputted parameters, the actual indentation depth is outputted parameter. The correlation coefficient came up to 0.9958.