A Novel Reflective Optical Fiber Bundle Hydrogen Sensor Based on BP Network

The characteristics of a novel reflective optical fiber bundle hydrogen sensor were analyzed, a optical fiber bundle hydrogen sensor based on the BP artificial neutral network is presented in this paper, the neural network was used for processing the data collected from the optical fiber bundle hydrogen sensor, which could enhance the measuring accuracy, at the same time, the intrinsic and extrinsic influence are eliminated mainly, such as the effect of fluctuation in the light source, the loss of light and the shaking in the optical fiber bundle. In artificial neural network, using the sensor actual outputs as neural network´s inputs and using the calibrated concentration of hydrogen in experiment as the desired output, experimental results and numerical simulation by back-propagation (BP) artificial neural network shows the training method available, a linear precision of 0.1% for the optical hydrogen sensor was achieved.