Benchmark of intelligent control techniques for surface mount technology (SMT) adhesive dispensing: Process overview and the application of a structured neural network controller

The application of a structured neural network controller to the dispensing of adhesive in the manufacture of mixed technology surface mount boards is discussed in this paper. Following an outline of the adhesive dispensing system and the various process faults that can occur in operation, the determination of a simple model of the system using standard system identification techniques is discussed. A controller is derived by mapping the control and fault diagnosis requirements onto a neural network architecture specifically designed to represent the adhesive dispensing process characteristics. The dispensing system response to the controller is given along with a comparison of the timing performance of the neural network with conventional proportional, integral and derivative (PID), rule based and fuzzy logic controllers.