Optimization of Neural Network Computation with use of Residual Number System for Tasks of Design of Neural Network Systems of Automatic Control

摘要

The article is devoted to the description of approaches to the solution of the actual problem of optimization of neural network calculations using the non-position number system (residual number system) for design problems of neural network automatic control systems. The problems arising in neural network control systems of industrial objects characterized by astatic properties are identified and analyzed. The approach proposed to solve these problems is confirmed by an example of solving a real problem.