The Analyzing And Processing Of Demodulated Signals Of FBG Sensors Based On Radial Basis Function Network

Fiber Bragg Grating (FBG) sensor, which belongs to the wavelength modulating sensor, has now become an international hotspot in the field of optical fiber sensor research. Compared with other types of optical fiber sensor, such as intensity and phase modulated optical fiber sensor, FBG sensor has been widely used for its superiorities of easy multiplexing, resistance to electromagnetic interference and so on. The main problem which blocks the utility of FBG is how to measure the wavelength shift accurately in Real-time.Firstly, along with the advantages and disadvantages of various demoudulation methods, the characteristics of the FBG sensing technology was introduced in this dissertation, and then the demodulation system based on the tunable fiber Fabry-Perot filter was analyzed in details.Secondly, due to the non-ideal performance of FBG package technics and optical devices in the demodulation system, and the effects of following processing related to photoelectric conversion, the waveform is no longer an ideal gaussion wave and has been distorted, so that there are greater difficulties about demodulation system appeared and accessed error if the conventional peak detection and gaussian fit methods are used. Based on these issues above, we proposed a method that can reconstruct FBG spectrum based on radial basis function network. Results obtained from MATLAB simulation based on the theory analysis showed that: compared with conventional peak detection and gaussion fitting algorithm, demodulation error can be reduced by using this method, and then accuracy of demodulation is improved.Finally, under laboratory conditions and specific requirements of algorithm, hardware, software and the function realization of the demodulation system were designed in this dissertation. In the hardware design part, DSP and the external circuit design were realized, the signal processing DSP chip of the demodulation system used TI's TMS320VC5416PGE; in the software design part, core algorithm and the software design associated with the hardware design were accomplished by mixed using of C and assemble program language. Procedures for the preparation of PC are compiled using C++Builder. Action of DSP which controlled by serial communication was carried out and, the visualization of the data output, data storage, and query capabilities were achieved.