Path Optimization Problem of Multi-Probe Flying Probe Tester

The Flying Probe Test (FPT) is a primary method for PCB testing, boasting advantages such as high precision, extensive coverage, and flexibility. The efficiency of FPT crucially hinges on the probing path. This study models the problem akin to the Traveling Salesman Problem in the context of two-probe testing. It considers the relationship between time and distance, introducing additional constraints to ensure probe synchronization during testing. The proposed solution employs heuristic algorithms, delivering high-quality solutions quickly. We introduce an encoder-decoder architecture based on deep learning to augment the solution’s generalizability. Experimental results affirm its effectiveness, showcasing superior performance in both training and generalizing to larger-scale problems.