Implementation of neural network controller for unknown systems

With both neural network theory and custom VLSI technology becoming more advanced, it is now possible to implement adaptive-type control strategies using VLSI-based neural networks. The reported work addresses three issues: developing a general control system structure for control of unknown systems; developing the neural network paradigm for the controller, a multilayer feedforward network which is trained using a variant of the backpropagation algorithm; and the VLSI implementation of the neural network paradigm using basic analog VLSI building blocks. Simulations that support the proposed VLSI layout are presented.<
>