Microelectronics Inspection Using Fuzzy Logic Sensor Fusion

Current methods for inspection of microelectronics assemblies (i.e., printed circuits and multichip modules) are unreliable or extremely expensive. Available inspection systems uses a single-sensor technology such as vision, IR, X-ray or ultrasonics to locate defects. Each sensor type is capable of locating defects of a particular kind or class. No single sensor technology is capable of detecting all possible defect types with even a marginal reliability. This project proposes to develop a microelectronics inspection system that uses multiple low-cost sensors and an innovative method employing a neural-net-based fuzzy logic associative memory and Dempster-Schafer theory of combining data to produce an affordable and reliable microelectronics inspection system. During Phase I, we will develop and implement the algorithms, demonstrate the feasibility using test data, develop a system architecture and prepare a system concept document. During Phase II, we will prepare a detailed system design, build and test a laboratory prototype and demonstrate a working system to the government and industry.