We developed a method of output signal processing for distributed fiber-optical measuring systems. This method is based on neural-like principles of data processing. A mathematical model of the three-layered perceptron was used to reconstruct the physical field distribution measured by a distributed interferometric system. We proposed an algorithm that is based on the committee method of recognition of the dynamic object detected by a fiber-optic measuring system.
Keywords: Fiber optic, sensor,
measurement, neural network, perceptron, tomography, pattern
recognition, committee method
Full text: DjVu file (Read more about
DjVu format),
text sent to the editor PDF