Applying image processing to an intrinsic multi-mode fibre optic strain sensor

Light intensity patterns produce at the output of a multi-mode optical fibre are complex in nature due to the interference of the many modes which propagate down its length. Small strains exerted on such a fibre change its physical properties, such as length and refractive index, and hence alter the propagation characteristics of the light as it travels along the fibre. These effects result in an alteration of the complex modal interference pattern at the fibre output. It is possible to produce unique patterns at the fibre output over a wide range of strains, and use an artificial neural network as a means of associating a specific strain with a given pattern. This has allowed the construction of a simple, yet effective strain sensor. The system described consists of three basic blocks: a multi-mode fibre acting as a sensor; digital signal processing of the output pattern after acquisition; and finally, a neural network as the interface between the pattern and its corresponding strain value