红外光谱分析是基于分子振动与跃迁理论的鉴别物质化学组成的技术。 得到的光谱数据常常具有较高的维数和重叠度, 这给后续的数据处理带来困难。 为此提出一种GK可能C均值聚类算法(GKIPCM), 引入了GK聚类算法的马氏距离测度与改进的可能C均值聚类算法(IPCM)的模糊隶属度与聚类中心更新方程, 使样本的距离测度具有自适应性且避免了聚类中心的一致性。 GKIPCM算法具有分类精度更高, 分类准确率对参数敏感性低的优点。 将四组洗净白菜作为光谱分析对象, 分别施加三种农药(高效氯氟氰菊酯)配比, 采用安捷伦Cary 630 FTIR光谱仪采集白菜的傅里叶中红外光谱(FT-MIR)。 首先对样本进行预处理, 使用多元散射矫正(MSC)对光谱数据降噪, 消除数据偏移量; 其次, 由于采集到的数据波数范围为4 300~590 cm^-1, 数据维数达到了971维, 故使用主成分分析(PCA)对数据实现降维, 降维后的数据维度减小到了23, 且23个主成分的累积贡献率高达99.60%; 但各类光谱的特征信息依然混杂在一起, 故使用线性判别分析(LDA)提取特征鉴别信息, 进一步将数据降至3维; 最终, 运行模糊C-均值聚类算法(FCM)得到较优初始聚类中心, 使用GKIPCM算法对四类降维后的光谱数据进行聚类分析, 并与GK聚类算法与IPCM聚类算法的运行结果作对比。 GKIPCM算法的总迭代时长为0.218 8 s, 分类准确率达到了97.22%。 相较之下, GK算法与IPCM算法的准确率分别为63.89%和91.67%, 运行的总时长为0.093 8与0.062 5 s。 从实验结果可看出, GKIPCM算法可以通过分析光谱数据从而完成对不同程度农药残留进行定性分析的任务。

田间土壤属性复杂且随时间变化, 快速精准地获得多种土壤理化指标数据对指导精细农业操作具有重要意义。 为避免土壤水分带来的干扰, 基于光谱技术的土壤成分含量预测需在土壤样本干燥的情况下进行光谱测量, 然而土壤水分同样是指导农业生产的重要指标。 为同时预测黑土区土壤有机质(SOM)、 水分(SMC)、 总铁(Fe)和pH值, 提出测量湿土土壤样本的可见-近红外光谱, 并采用标准正态变量变换(SNV)-连续小波变换(CWT)法分解光谱反射率, 逐样本进行SNV后, 以Mexh为小波基函数进行10个尺度(2¹, 2², …, 2¹⁰)的分解, 并与常用光谱处理方法进行对比, 包括高斯滤波(GS)、 一阶导数(FD)、 连续统去除(CR), 数学变换等7种方法。 将74个样本数据划分为两组, 其中50个作为建模集, 24个作为验证集。 经SNV-CWT变换后, 每个尺度的小波系数与每个目标变量间置信度小于0.05的波段作为随机森林(RF)预测模型的输入变量, 以各尺度验证模型精度为标准确定每个预测目标的最佳分解尺度; 通过计算最佳尺度小波系数与土壤成分间的皮尔森相关系数(PCC), 基于模型的相关系数(MBC)和灰色关联度(GRD), 判断各属性的特征波段, 且分别以三种相关系数作为指标, 以过滤式筛选法建立不同属性的RF估测模型。 结果表明: 与7种常用的处理方法相比, SNV-CWT分解后四种土壤成分的预测精度均有提高, SOM, SMC, Fe和pH对应的最佳分解尺度分别为7, 8, 1和10。 在以多维特征作为输入变量的情况下, SOM与SMC的验证模型决定系数(R²)即可达到0.90和0.93。 三种分析方法中以MBC计算的相关系数为波段筛选指标建立的模型精度最佳, 其中SOM与SMC的R²均为0.94, 且Fe(R²=0.67, Mse=0.01%, RPD=1.76)与pH(R²=0.80, Mse=0.1, RPD=2.24)的模型精度具有大幅度提高, 可应用于多种土壤理化指标数据的提取与监测。

We fabricated a three-dimensional microstructured optical waveguide (MOW) in a single-crystal using the femtosecond-laser writing and phosphoric acid etching techniques, and observed excellent midinfrared waveguiding performance with low loss of ～0.5dB/cm. Tracks with a periodic arrangement were written inside the yttrium aluminum garnet (YAG) crystal via femtosecond laser inscription, and then etched by the phosphoric acid (H3PO4) to form the hollow structures. The evolution of the microstructure of tracks was investigated in detail. The function of the MOW was analyzed by different numerical methods, indicating the proposed MOW can effectively operate in quasi-single-mode pattern in the midinfrared wavelength range, which agrees well with our experiment results.

This work demonstrates the Nd:YAG waveguide laser as an efficient platform for the bio-sensing of dextrose solutions and tumor cells. The waveguide was fabricated in an Nd:YAG crystal with the cooperation of ultrafast laser writing and ion irradiation. The laser oscillation in the Nd:YAG waveguide is ultrasensitive to the external environment of the waveguide. Even a weak disturbance leads to a large variation of the output power of the laser. According to this feature, an Nd:YAG waveguide coated with graphene and WSe2 layers is used as substrate for the microfluidic channel. When the microflow crosses the Nd:YAG waveguide, the laser oscillation in the waveguide is disturbed and induces fluctuation of the output laser. According to the fluctuation, the microflow is detected with a sensitivity of 10 mW/RIU.

We report on the operation of passively Q-switched waveguide lasers at 1 μm wavelength based on a graphene/WS2 heterostructure as a saturable absorber (SA). The gain medium is a crystalline Nd:YVO4 cladding waveguide produced by femtosecond laser writing. The nanosecond waveguide laser operation at 1064 nm has been realized with the maximum average output power of 275 mW and slope efficiency of 37%. In comparison with the systems based on single WS2 or graphene SA, the lasing Q-switched by a graphene/WS2 heterostructure SA possesses advantages of a higher pulse energy and enhanced slope efficiency, indicating the promising applications of van der Waals heterostructures for ultrafast photonic devices.

The bonded distributed feedback (DFB) fiber laser (FL) acoustic emission sensor and the intensity response of the DFB-FL to external acoustic emissions are investigated. The dynamic sensitivity of the DFB-FL is calibrated by a referenced piezoelectric receiver. In the DFB-FL we used here, the minimum detectable signal is 2×10 6 m/s at 5 kHz. Using wavelet packet technology, the collected signals are analyzed, which confirms that an intensity-modulated DFB-FL sensor can be used to detect acoustic emission signals.