微波加热不均匀性一直以来都是从事微波加热控制方向研究人员心目中的热点问题。根据微波加热装置的物理结构建立了炉内各层表面的温度静态差分模型结合实验以求得微波加热的实际功率。再基于传热学的有限差分法建立三维空间中的温度分布模型，利用MATLAB以及COMSOL仿真对比验证了模型的有效性。假定微波均匀加热求得被加热介质的平衡温度与不均匀加热时的温度进行比对以找出微波加热过程中介质的部分温升平衡点，最后互相比对找出最优点为控制对象进行专家PID（proportion-integral-derivative）微波加热。实验结果表明，该方法能较为精确地测量出被加热介质任何时刻的平衡温度，使得微波加热在工业生产上有着更加广泛的应用。

针对现有10 kW高功率工业微波炉，采用继电器作为控制执行器，在使用传统控制方法加热时，温度存在较大超调和明显振荡，系统温度稳定性较低，为解决上述问题将反向传播神经网络PID（BPNNPID）控制引入到该装置微波加热温度控制中，并以自来水为加热对象进行仿真对比与实验验证。首先，利用现有输入输出实验数据，建立工业微波炉温度控制模型；其次，运用MATLAB/SIMULINK搭建高功率工业微波炉温度控制系统并进行仿真对比实验；最后，实验验证BPNNPID控制方法在加热5 kg自来水时工业微波炉的温度控制性能，实验结果表明，较常规PID、模糊PID控制，该方法在微波加热过程中对媒质温度控制超调更小且未发生明显温度振荡，有效改善了高功率工业微波炉工作时的系统温度稳定性，有助于提高产品质量和安全性能。

An optical field with sub-nm confinement is essential for exploring atomic- or molecular-level light-matter interaction. While such fields demonstrated so far have typically point-like cross-sections, an optical field having a higher-dimensional cross-section may offer higher flexibility and/or efficiency in applications. Here, we propose generating a nanoscale blade-like optical field in a coupled nanofiber pair (CNP) with a 1-nm-width central slit. Based on a strong mode coupling-enabled slit waveguide mode, a sub-nm-thickness blade-like optical field can be generated with a cross-section down to ∼0.28nm×38nm at 1550 nm wavelength (i.e., a thickness of ∼λ0/5000) and a peak-to-background intensity ratio (PBR) higher than 20 dB. The slit waveguide mode of the CNP can be launched from one of the two nanofibers that are connected to a standard optical fiber via an adiabatical fiber taper, in which a fundamental waveguide mode of the fiber can be converted into a high-purity slit mode with high efficiency (>98%) within a CNP length of less than 10 μm at 1550 nm wavelength. The wavelength-dependent behaviors and group velocity dispersion in mode converting processes are also investigated, showing that such a CNP-based design is also suitable for broadband and ultrafast pulsed operation. Our results may open up new opportunities for studying light-matter interaction down to the sub-nm scale, as well as for exploring ultra-high-resolution optical technology ranging from super-resolution nanoscopy to chemical bond manipulation.

Epilepsy is a common neurological disorder that occurs at all ages. Epilepsy not only brings physical pain to patients, but also brings a huge burden to the lives of patients and their families. At present, epilepsy detection is still achieved through the observation of electroencephalography (EEG) by medical staff. However, this process takes a long time and consumes energy, which will create a huge workload to medical staff. Therefore, it is particularly important to realize the automatic detection of epilepsy. This paper introduces, in detail, the overall framework of EEG-based automatic epilepsy identification and the typical methods involved in each step. Aiming at the core modules, that is, signal acquisition analog front end (AFE), feature extraction and classifier selection, method summary and theoretical explanation are carried out. Finally, the future research directions in the field of automatic detection of epilepsy are prospected.