为了评估微型近红外光谱仪应用于现场水果糖度检测的可行性, 采用粒子群算法结合反向传播(BP)神经网络建立了苹果糖度的无损高精度快速检测方法, 研究了微型近红外光谱仪NIRscan以单波长和阿达玛变换两种测量模式获得的光谱数据, 应用多种不同的数据预处理方法和多元线性回归、偏最小二乘法、粒子群算法(PSO)、BP神经网络等算法建立分析模型。结果表明, 以阿达玛变换工作模式测得的光谱数据更好, 以1阶导数结合Savizky-Golay平滑算法作数据预处理, 应用PSO结合BP神经网络建立的苹果糖度预测模型具有更高的预测精度, 预测相关系数和均方根误差分别为0.9911和0.1502。该微型近红外光谱仪NIRscan用于苹果糖度的现场快速和高精度无损检测具有可行性。

The strong influences of temperature and vacuum on the optical properties of In_0.3Ga_0.7As surface quantum dots (SQDs) are systematically investigated by photoluminescence (PL) measurements. For comparison, optical properties of buried quantum dots (BQDs) are also measured. The line-width, peak wavelength, and lifetime of SQDs are significantly different from the BQDs with the temperature and vacuum varied. The differences in PL response when temperature varies are attributed to carrier transfer from the SQDs to the surface trap states. The obvious distinctions in PL response when vacuum varies are attributed to the SQDs intrinsic surface trap states inhibited by the water molecules. This research provides necessary information for device application of SQDs as surface-sensitivity sensors.

The liquid-core fiber with relatively high refractive index difference between the core and cladding is proven to be bending insensitive. The single mode condition and bending loss of the fiber with a mixture of toluene and chloroform as its core material are studied. The results show that the bending loss of this fiber is not only much smaller than the conventional silica single mode fiber but also can be tuned by the temperature and liquid mixture ratio. This kind of fiber may find its potential applications in all-optical network.