Application of dynamic baseline adjustment based on swarm intelligence optimization in the signal processing of fiber SPR sensor

Abstract

In view of the difficulty in determining the appropriate parameters in traditional centroid method, which is used to detect the resonance point of SPR reflection spectrum, a dynamic baseline adjustment method based on the particle swarm optimization algorithm (PSO) is proposed. The method regards the parameter of spectral width (β), number of measurement points below the baseline (m), and the area ratio (λ) as the foundation for selecting the best dynamic baseline. Based on the condition of two-level fitness function, PSO algorithm is adopted to track the dynamic baseline with the best combination of parameters. To verify the effectiveness of the algorithm, on the one hand, design the simulation experiments which prove that the new algorithm is superior to other algorithms in tracking ability and anti-noise ability. On the other hand, design the application experiment under the condition of light source fluctuation, by measuring the spectral data of standard sucrose solutions with 8 different concentrations by the fiber SPR sensor, the calculation effects of resonance point by four centroid methods are compared. The experimental results show that the improved method has the best predictive ability in measuring the concentrations of measured sucrose solutions, its fitting degree between the calculated resonance wavelength and the concentration of measured liquid is 0.9963, the root mean squares error (RMSE) is 1.78. In addition, the dominant positions of three parameters are investigated, the results show that the fixed m method has the best effect which can provide a basis for the selection of appropriate centroid parameters. Finally, to prove the feasibility of PSO in determining the parameters of optimal baseline, PSO and other metaheuristic algorithms are implemented for comparison. The experimental results show PSO algorithm has great application potential in terms of measurement effect and optimization speed for the studied problem. Therefore, the improved method can effectively extract the centroid parameters and eliminate the influence of light source fluctuation, which verifies the feasibility of the proposed method in the signal processing of fiber SPR sensor.