Nonuniform Strain Distribution Profile Reconstruction Of Bragg Grating Based On Chaos Genetic Algorithm

Since the prospect of application of fiber Bragg grating sensor is broad in the structure health monitoring system, and both of evolution computing and machine leaning are front science in the field of artificial intelligence and pattern recognition, the research of nonuniform strain profile reconstruction in Bragg grating based on chaos genetic algorithm and support vector regression is highly worth in theory and engineering practice.This dissertation firstly introduces coupled-mode theory to analyses the principle of strain sensor of Bragg grating, then focuses on the Runge–Kutta integrating method and transfer matrix method for reflected spectrum calculation. After discussing the advantage and disadvantage of genetic algorithm, an improved one is proposed with real code and fitness ranking scheme. For accelerating the convergence of genetic algorithm and avoiding getting into local best solution, the elements of chaos optimize algorithm and carrier mode are studied for creating chaos operator, which is embedded into the improved genetic algorithm to form chaos genetic algorithm. Whereafter, to validate the availability of local searching of chaos operator, four indexes are defined to weigh the reconstruction precision in both chaos genetic algorithm and improved genetic algorithm, when the strain profile is linear,polynomial,high-order and discontinuouse. Finally, Discusses detailedly the influence of generation,bandwidth of reflected spectrum and sample margin, especially the measuring noise, to the reconstruction precision of strain profile, following with the proposition ofε-support vector regression denoisng method, of which the result of simulation is satisfying.