Neural Network Processing of Optical Fiber Sensor Signals for Impact Location
- PDF / 460,131 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 89 Downloads / 181 Views
ABSTRACT A system to detect and locate impacts by foreign bodies on a surface was developed and tested. Fiber optic extrinsic Fabry-Perot interferometer (EFPI) strain sensors were attached to or embedded in the surface, so that stress waves emanating from an impact could be detected. By employing an artificial neural network to process the sensor outputs, the impact location could be inferred to centimeter range accuracy directly from the arrival time data. In particular, the network could be trained to determine impact location regardless of material anisotropy. Results demonstrate that a back-propagation network identifies impact location for an anisotropic graphite/bismaleimide plate with the same accuracy as that for an isotropic aluminum plate. OVERVIEW An investigation was initiated in the use of neural networks to determine impact location by processing the output of a network of fiber optic strain sensors distributed on a surface. This approach extends the results of Gunther, et all and Sirkis, et al4 , who showed that impact location by triangulation could be used by comparing the arrival times at several sensors of the acoustic signal generated by the impact. For this study, a commercially available neural network simulator running on a personal computer was used to train a network using a back-propagation algorithm'. The ability of the network to determine impact location by differential arrival times of acoustic signals was assessed by comparing network outputs with actual experimental results using impacts on a panel instrumented with optical fiber sensors. THE EFPI FIBER OPTIC SENSOR Optical fiber strain gages have been implemented using the conventional Extrinsic Fabry-Perot Interferometric (EFPI) sensor system as shown in Figure 12 . Here, a 1300 nm single mode silica fiber transmits light from a laser diode to the EFPI sensor element. At the opposite end of the input fiber, the laser light signal is partially reflected and partially transmitted, exiting the fiber (hence, an "extrinsic" sensor design) and traversing the air gap separating the ends of the input fiber and an output fiber used solely as a reflector. This signal and the signal reflected from the facing fiber endface interfere and propagate back through the input fiber. They travel through a fiber coupler to a photodiode detector. The observed intensity at the detector is given by' det:
1
2 ta a+ 2s tan[sin-'(NA)]
cos(-i+ X aa+ 2s tan[sin 4 (NA)]
'NeuralWorks Professional II plusTm v5.02, NeuralWare, Pittsburgh, PA 15276. 121 Mat. Res. Soc. Symp. Proc. Vol. 360 01995 Materials Research Society
where A is the amplitude of the first reflection, t is the transmission coefficient of the air-glass interface ('0.98), a is the fiber core radius, s is the length of the air gap, X is the wavelength of operation in free space, and NA is the numerical aperture of the single-mode fiber, given by NA=(n1 2 -n2 2 )1t2 , where ni and n2 are the refractive indices of the core and the cladding, respectively. For our application, the EFPI sensors wer
Data Loading...
