Improvement of measurement range of optical fiber displacement sensor based on neutral network

By studying the output characteristics of random type optical fiber displacement sensor and semicircular type optical fiber displacement sensor, a sensor of new structure was designed. Using the ratio of the two output signals as the output of the whole system, the measurement range was enlarged, the linearity was improved, and the errors of reflective and absorbent changing in target surface are automatically compensated. Meantime, an optical fiber sensor model of correcting static error and linearizing the output curve based on BP artificial neural network was set up. The intrinsic errors such as fluctuations in the light, circuit excursion, the intensity losses in the fiber lines and the additional losses in the receiving fiber caused by bends were eliminated. By discussing in theory and experiment, the error of nonlinear is 2.9%, the measurement range reaches to 3.0 mm and the relative accuracy is 2%. This kind of sensor offers such advantages as no electromagnetic interference, simple construction, high sensitivity, good accuracy and stability. (C) 2013 Elsevier GmbH. All rights reserved.