Students in the Department of Marine Engineering at Texas A and M in Galveston are being introduced to state-of-the-art neural networks and adaptive methods in their control courses and laboratories. Microprocessor-based controllers and models of ship equipment are controlled by feed-forward neural network algorithms in the newly funded laboratory. Novel features of this project are the adoption of optimal (Nlsp type) neural networks (i.e., networks which require the smallest number of arithmetic operations to propagate from input to output), and learning algorithms which guarantee convergence to an optimal solution. The equipment is used in ship design projects at all stages of the Integrated Learning System Program of the Marine Engineering Department, emphasizing intelligent control and automation.
