The production of volatile compounds in model casein systems with varying fat levels as affected by low-frequency ultrasound

The effects of low-frequency ultrasound on the production of volatile compounds in model casein protein systems containing various fat concentrations of 2%, 4% and 6% (w/w) were investigated. Ultrasound application was performed at 20 kHz for up to 10 min which corresponded to energy densities ranging from 9.54 to 190.8 J mL⁻¹. Similar volatile compounds were detected both in pure fat and mixtures of casein and fat (CF) systems. These volatiles belonged to the groups of aldehydes, ketones, esters, alcohols and hydrocarbons, which were the products of oxidation of lipids or protein degradation due to acoustic cavitation. The amount of fat in the casein systems had minor effects on the production of volatiles, whereas the production of volatile compounds was significantly affected by the ultrasound treatment. Short sonication times <5 min generated similar volatile profiles to the untreated samples. In contrast, prolonged sonication for 5 and 10 min considerably increased the production of volatile compounds and the amounts of fatty acids. Thus, the application of low–frequency ultrasound for short periods should be considered to minimise the production of volatile compounds which can ultimately affect the taste.     

Synergistic effect of high-intensity ultrasound and β-cyclodextrin treatments on browning control in apple juice

This work evaluated the synergistic effects of combined high-intensity ultrasound (HIU) with β-cyclodextrin (β-CD) treatments on inhibiting browning of apple juice and explored the mechanism through simulation system. The combined treatment of 300 W HIU with 0.006 g mL⁻¹ β-CD had a synergistic impact on maintaining juice colour, resulting in a 39.06% reduction in browning degree, only a 36.64% decrease in total phenolic content, and a 17.82% reduction in PPO activity. The inhibition of enzymatic browning in simulated system revealed that HIU suppressed the enzyme (Polyphenol oxidase, PPO) and β-CD inhibited enzyme (PPO) and embedded substrate (polyphenol). The results of spectroscopic analysis showed that the particle-size distribution of PPO narrowed, the content of α-helix in the secondary structure increased, the fluorescence intensity increased, and the maximum wavelength was red-shifted after HIU and β-CD treatment. Changes in structure could further result in PPO activity loss. Hence, the combined treatment could synthetically alleviate the browning of apple juice.     

Ultrasonic-assisted phosphate curing: a novel approach to improve curing rate and chicken meat quality

The effects of ultrasound-assisted phosphate curing (UPC), phosphate curing (PC), and deionised water curing (DC) on the quality of chicken breast meat were investigated. The analysis of tissue sections, low-field nuclear magnetic resonance (LF-NMR), and headspace solid-phase microextraction gas-phase mass spectrometry (HS-SPME-GC-MS) of water distribution, myofibrillary structure, and volatile components of the meat samples were performed to reveal the mechanism. Also, the curing rate, water holding capacity (WHC), and shear force of meat were assessed. The results revealed that ultrasonic-assisted salting improved the meat salting rate, WHC, and tenderness by mechanically destroying the muscle fibre structure, increasing the space between the muscle fibre bundles, promoting water migration and uniform distribution, and reducing water freedom in the treated samples. In addition, HS-SPME-GC-MS analysis revealed that PC significantly reduced the content of volatile flavour components in the chicken breast meat compared with DC (P < 0.05); however, the combination treatment with PC and ultrasonic curing could reconstruct the flavour loss caused by phosphate. Thus, our cumulative results suggest that the ultrasonic technology may effectively assist the processing technology in improving the quality of meat and reconstructing the flavour loss in meat products.     

Effects of ultrafiltration combined with high-pressure processing,ultrasound and heat treatments on the quality of a blueberry–grape–pineapple–cantaloupe juice blend

This study investigated the effect of ultrafiltration (UF) combined with high-pressure processing (HPP) at 550 MPa, 25 °C for 5 min, ultrasound (US) at 520 W, 40 °C for 10 min and heat treatment (HT) at 90 °C for 3 min on the microbial, physicochemical and sensory properties of a blueberry–grape–pineapple–cantaloupe juice blend during 104 days of storage at 4 °C. After UF, the shelf life of the HPP- and US-treated clear juice blends were 104 and 72 days during the storage at 4 °C respectively. HPP, US and HT treatment minimally affected the anthocyanin and total phenol contents, while HPP better maintained the ascorbic acid levels and sensory properties in the clear juice blend during the storage. Therefore, HPP combined with UF was identified as a prospective processing technique in the fruit juice industry.     

文物建筑外墙风化检测与分析

通过北京大学红楼外墙风化检测实例，提出了文物建筑外墙风化检测的重要性，并介绍了砖体动态弹性模量超声检测，粘结拉拔检测和砖体含盐量检测方法。     

近场MVDR聚焦波束形成被动定位方法

该文系统地研究了最小方差信号无畸变响应（MVDR）聚焦波束形成方法的定位性能,特别是对该方法定位性能受某个参数的影响情况进行了较详细的研究。首先分析了近场情况下线列阵接收信号的模型,然后分别推导了常规聚焦波束形成和MVDR聚焦波束形成的定位原理。仿真比较了这两种定位方法的定位性能,以及MVDR聚焦波束形成被动定位方法在不同的阵元间距和不同的目标距离情况下的定位性能。仿真结果表明,与常规聚焦波束形成相比,MVDR近场聚焦波束形成在提高分辨率的同时,减小了＂混叠＂影响,具有更为平滑的背景和更低的旁瓣级,同时认为,为了得到比较好的定位效果,目标应处于相对于＂近场＂的位置,从而为目标精确定位提供依据。     

面向乳腺超声计算机辅助诊断的两阶段深度迁移学习

目的 为了提升基于单模态B型超声（B超）的乳腺癌计算机辅助诊断（computer-aided diagnosis,CAD）模型性能,提出一种基于两阶段深度迁移学习（two-stage deep transfer learning,TSDTL）的乳腺超声CAD算法,将超声弹性图像中的有效信息迁移至基于B超的乳腺癌CAD模型之中,进一步提升该CAD模型的性能。方法 在第1阶段的深度迁移学习中,提出将双模态超声图像重建任务作为一种自监督学习任务,训练一个关联多模态深度卷积神经网络模型,实现B超图像和超声弹性图像之间的信息交互迁移;在第2阶段的深度迁移学习中,基于隐式的特权信息学习（learning using privilaged information,LUPI）范式,进行基于双模态超声图像的乳腺肿瘤分类任务,通过标签信息引导下的分类进一步加强两个模态之间的特征融合与信息交互;采用单模态B超数据对所对应通道的分类网络进行微调,实现最终的乳腺癌B超图像分类模型。结果 实验在一个乳腺肿瘤双模超声数据集上进行算法性能验证。实验结果表明,通过迁移超声弹性图像的信息,TSDTL在基于B超的乳腺癌诊断任务中取得的平均分类准确率为87.84±2.08%、平均敏感度为88.89±3.70%、平均特异度为86.71±2.21%、平均约登指数为75.60±4.07%,优于直接基于单模态B超训练的分类模型以及多种典型迁移学习算法。结论 提出的TSDTL算法通过两阶段的深度迁移学习,将超声弹性图像的信息有效迁移至基于B超的乳腺癌CAD模型,提升了模型的诊断性能,具备潜在的应用可行性。     

基于压缩感知的 HIFU 回波信号降噪研究

由于传统的降噪处理方法很难干净地去除高强度聚焦超声(HIFU)信号中的噪声,提出利用压缩感知(CS)对 HIFU 回 波信号进行降噪。 在观测矩阵的设计中将传统的高斯随机观测矩阵改进为稀疏循环结构化矩阵,减少了构造观测矩阵和重构 信号的时间。 仿真实验表明,与带通滤波器、小波降噪方法和经验模态分解(EMD)降噪方法相比,该方法得到的信号的重构信 噪比(RSNR)更高,重构均方差(RMSE)和最大误差(ME)更小。 用不同方法对不同温度下获得的 HIFU 回波信号进行去噪并提 取二次谐波激发效率,发现采用该方法得到的二次谐波激发效率曲线方差和波动更小,验证了该降噪方法在实测信号中的优 越性。     

管道缺陷检测全聚焦数据处理及成像方法研究

管道缺陷检测是保证管道系统安全运行的必要手段,本文采用全聚焦成像的方法对管道缺陷进行成像与识别。利用有限元软件ABAQUS对基于L(0,1)模态导波的管道缺陷检测方法进行数值模拟研究,利用中值滤波、希尔伯特变换及信号包络锐化方法对信号进行预处理,然后用处理后的信号构建全矩阵数据,最后利用全聚焦算法实现缺陷成像,最终显示成像结果。实验结果表明,将信号进行预处理后再成像,可有效提高分辨率,实现缺陷的高精度可视化。     

基于红外同步的室内超声定位系统的设计

提出了一种室内定位方案,利用红外作为发射超声信号的同步计时信号、FPGA内部多路计数模块同步对超声计数,并将采样得到的多路数据通过SPI总线分别上传至主控芯片STM32,并对数据进一步处理,最后由定位算法得到目标点的位置。主要实现的功能是FPGA多路信号采集、FPGA与SPI通信的设计以及定位算法的设计,精度可以达到cm级,在实际应用中具有较高的参考价值。