Quality Estimation of the Resistance Spot Welding Based on Genetic K-Means Cluster Analysis

The electrode displacement signal of resistance spot welding process is monitored and mapped into a 15×25 element bipolarized matrix by means of some method of fuzzy theory. The electrode displacement pattern matrixes from different welding current are treated as gene to construct chromosome. The genetic K-means algorithm (GKA), which combines the simplicity of the K-means algorithm and the robust nature of the genetic algorithm, is utilized to realize clustering analysis and quality estimation of welded spots. The results of the clustering analysis indicate that the electrode displacement pattern matrix can provide adequate quality information of welded spots for machine learning and the method avoids complicated programming work. At the same time, compared with the conventional K-means algorithm, the GKA method has a strong ability to find the global optimal solutions and be able to quickly and effectively complete the clustering task under small sample circumstance. The results of clustering can realize quality estimation of the resistance spot welding and also can be used as a prior knowledge to provide the necessary support for the supervised machine learning to evaluate the weld quality.