Neuro-iterative algorithm for reconstructing distributed temperature patterns with fibre-optic measuring intelligent networks

A fiber-optic tomographic measuring intelligent network is used for carrying out a study of algebraic and neural network approaches for reconstructing distributed physical fields. The simulation was carried out in the Matlab software system. The researchers find out the advantages and disadvantages of neural network approach. They propose a neuro-iterative algorithm that combines the benefits of both neural network and algebraic approaches of reconstructive tomography. Training on model data, apart from the non-linear instrumentation a system, and the iterative part of solution in a proposed approach allows using more simple structures of artificial neural networks. The authors show a proposed algorithm performance experimentally.