Analysis and design of a recurrent neural-network for linear-programming

Linear programming is an important tool for system optimization and modelling. This paper presents a recurrent neural network with a time-varying threshold vector for solving linear programming problems. The proposed recurrent neural network is proven to be asymptotically stable in the large and capable of generating optimal solutions to linear programming problems. An op-amp based analog circuit design for realizing the recurrent neural network is described. The asymptotic properties of the proposed recurrent neural network for linear programming are analysed. A detailed example is also presented to demonstrate the performance and operating characteristics of the recurrent neural network.