Achieving high-accuracy, low-latency segmentation of infrastructure cracks on resource-constrained edge devices are a critical challenge in the operation and maintenance of intelligent transportation infrastructure. To address these issues, this study proposes a collaborative optimization framework that integrates architecture design, pruning, and quantization. First of all, a multi scale feature aggregation decoder is designed. It employs grouped large kernel convolutions, a gated attention mechanism, and an efficient upsampling module to enhance the perception of crack topology. Second, a structured pruning strategy based on Taylor expansion is proposed to remove redundant channels in a task aware manner. The model is further compressed by using hybrid linear symmetric 8-bit signed integer full quantization method. Experiments on the self-built dataset and two public datasets show that the proposed framework achieves significantly better performance than the mainstream model with only 2.4 million parameters and 4.6 giga floating-point operations per second (GFLOPs). After pruning and quantization, the parameter drops to 1.37 million, storage space is reduced by over 85%, and the mean intersection over union (mIoU) remains above 0.70. Visualization analysis and robustness tests further confirm the model's feature focusing ability and generalization performance in complex scenarios. This study provides a viable technical pathway for high precision infrastructure crack detection on resource constrained edge devices.

Precise tumor imaging and effective drug delivery remain significant challenges in cancer therapy. To address these challenges, a novel drug delivery system, Us-Fe NPs/PLA-PEG@PTX, was developed for combined MR imaging and anti-tumor therapy. Ultra-small Fe3O4 nanoparticles (Us-Fe NPs) were stabilized with citrate and modified with polylactic acid-polyethylene glycol (PLA-PEG) polymers to enhance biocompatibility and facilitate paclitaxel (PTX) incorporation. PTX was initially extracted from the bark of the Pacific yew tree. The system achieved an encapsulation efficiency of 72.3 % and a loading content of 6.74 %. PTX was released in a controlled manner at both neutral (pH 7.4) and mildly acidic (pH 5.5) conditions. MR imaging studies indicated potential for T1-weighted imaging. Us-Fe NPs/PLA-PEG@PTX demonstrated dose-dependent cytotoxicity (IC50 = 0.06 mu g/ mL) against MDA-MB-231 cells. PEG modification reduced cellular uptake. Flow cytometry showed G2/M phase arrest and apoptosis induction, and confocal microscopy revealed microtubule disruption. These results suggest that Us-Fe NPs/PLA-PEG@PTX is a promising nano-platform for MR imaging-guided anti-tumor therapy.

Federated learning (FL) trains a common model collaboratively without uploading datasets. However, there are three challenges in traditional FL: the uneven amount of user data results in training inefficiency, gradient transmission faces high communication overhead, and gradients leak the privacy information of training sets. To address these challenges, we investigate a fog-driven communication-efficient and privacy-preserving FL scheme based on compressed sensing (CS). It is worth noting that CS can address high communication overhead and privacy leakage problems to some extent. Serving as an active participant, each fog node collects IoT device data and trains a local model. Fortunately, such design effectively addresses the problems of the uneven amount of data and the substantial disparity in computational capabilities. More notably, the weight feature extraction technique can lower gradient communication overhead by reducing dimension, while differential privacy (DP) is employed to resist data attacks. Then, CS with ambiguity technique not only compresses gradients to reduce communication overhead, but also enables secure aggregation of gradients to resist model attacks. Additionally, the proposed scheme is resistant to collusion attacks. Theoretical analyses and simulation experiments demonstrate that our scheme can achieve communication efficiency and privacy preservation remarkably while maintaining model accuracy.

Compressive sensing (CS), a breakthrough technology in image processing, provides a privacy-preserving layer and image reconstruction while performing sensing and recovery processes, respectively. Unfortunately, it still faces high-complexity, low-security, and low-quality reconstruction challenges during image processing. Therefore, this article presents a secure low-complexity CS scheme with preconditioning prior regularization reconstruction. More specifically, the original image is compressed by a low-complexity LFSR-based sparse circulant matrix to obtain measurements. It is worth noting that measurements achieve preliminary distribution equalization through the Tanh sequence to acquire processed measurements. Furthermore, the privacy-preserving edge processing for processed measurements can achieve high security. Finally, preconditioning prior regularization CS reconstruction is designed to improve reconstruction performance. Simulation results and analyses demonstrate that the proposed scheme can achieve low-complexity sampling, high security, and superior reconstruction performance.

Developing multifunctional nanoplatforms to comprehensively modulate the tumor microenvironment and enhance diagnostic and therapeutic outcomes still remains a great challenge. Here, we report the facile construction of a multivariate nanoplatform based on cancer cell membrane (CM)-encapsulated redox-responsive poly(N-vinylcaprolactam) (PVCL) nanogels (NGs) co-loaded with Cu(II) and chemotherapeutic drug toyocamycin (Toy) for magnetic resonance (MR) imaging-guided combination tumor chemodynamic therapy/chemoimmunotherapy. We show that redox-responsive PVCL NGs formed through precipitation polymerization can be aminated, conjugated with 3,4-dihydroxyhydrocinnamic acid for Cu(II) complexation, physically loaded with Toy, and finally camouflaged with CMs. The created ADCT@CM NGs with an average size of 113.0 nm are stable under physiological conditions and can efficiently release Cu(II) and Toy under tumor microenvironment with a high level of glutathione. Meanwhile, the developed NGs are able to enhance cancer cell oxidative stress and endoplasmic reticulum stress by synergizing the effects of chemodynamic therapy mediated by Cu-based Fenton-like reaction and Toy-mediated chemotherapy, thereby triggering significant immunogenic cell death (ICD). In a melanoma mouse model, the NGs show potent immune activation effects to reinforce tumor therapeutic efficacy through ICD induction and immune modulation including high levels of immune cytokine secretion, increased tumor infiltration of CD8+ cytotoxic T cells, and reduced tumor infiltration of regulatory T cells. With the CM coating and Cu(II) loading, the developed NG platform demonstrates homologous tumor targeting and T1-weighted MR imaging, hence providing a general biomimetic NG platform for ICD-facilitated tumor theranostic nanoplatform. Statement of Significance Developing multifunctional nanoplatforms to comprehensively modulate the tumor microenvironment (TME) and enhance theranostic outcomes remains a challenge. Here, a cancer cell membrane (CM)-camouflaged nanoplatform based on aminated poly(N-vinylcaprolactam) nanogels (NGs) co-loaded with Cu(II) and toyocamycin (Toy) was prepared for magnetic resonance (MR) imaging-guided combination tumor chemodynamic therapy/chemoimmunotherapy. The tumor targeting specificity and efficient TME-triggered release of Cu(II) and Toy could enhance tumor cell oxidative stress and endoplasmic reticulum stress by synergizing the effects of chemodynamic therapy mediated by Cu-based Fenton-like reaction and Toy-mediated chemotherapy, respectively, thereby leading to significant immunogenic cell death (ICD) and immune response. With the CM coating and Cu(II) loading, the developed NG platform also demonstrates good T1-weighted tumor MR imaging performance. Hence, this study provides a general biomimetic NG platform for ICD-facilitated tumor theranostics.

Compressed sensing (CS) is a popular signal processing technique. However, some of its performances still need to be improved for possible secure visual applications, including the optimization of the measurement matrix, privacy assurance, and the sparse recovery performance. To this end, we present prior-based measurement matrix design and sparse recovery algorithm for privacy-assured CS scheme in the cloud. More specifically, the measurement matrix is modeled by minimizing a Frobenius difference between the identity matrix and the Gram of the weighted sensing matrix. The gradient descent method is employed to derive the prior probability-weighted measurement matrix. Further, privacy-assured CS can be achieved by using within-row permutation and chaotic matrices. Finally, we also employ the prior information to enhance the accuracy of the sparse recovery algorithm by using prior probability-weighted orthogonal matching pursuit. Theoretical analyses and simulation results demonstrate that the proposed scheme can optimize measurement matrix, achieve privacy-assured CS, and improve sparse recovery performance.

The diversity and complexity of cracks pose significant challenges for the rapid and accurate detection of pavement defects. To address these challenges, this paper aims to enhance feature utilization and develop an end-to-end crack segmentation network (CSNet), with the goal of significantly improving detection accuracy. Firstly, the proposed model integrates dense parallel dilated convolutions, enabling it to capture local information across multiple scales effectively. Secondly, an innovative multiscale context fusion module, combined with an attention mechanism, is introduced to effectively aggregate deep features, enhancing the perception of cracks. Finally, a generalized dice loss function is employed to further improve the training efficiency. Extensive experiments were conducted on three public datasets, and a comprehensive comparison was made with mainstream segmentation models. The results demonstrate that the proposed CSNet performs outstandingly across multiple evaluation metrics, achieving the highest F1-score of 0.7968 and mIoU of 0.8094, significantly surpassing other advanced segmentation models.

In the rapid development of urbanization, the sustained and healthy development of transportation infrastructure has become a widely discussed topic. The inspection and maintenance of asphalt pavements not only concern road safety and efficiency but also directly impact the rational allocation of resources and environmental sustainability. To address the challenges of modern transportation infrastructure management, this study innovatively proposes a hybrid learning model that integrates deep convolutional neural networks (DCNNs) and support vector machines (SVMs). Specifically, the model initially employs a ShuffleNet architecture to autonomously extract abstract features from various defect categories. Subsequently, the Maximum Relevance Minimum Redundancy (MRMR) method is utilized to select the top 25% of features with the highest relevance and minimal redundancy. After that, SVMs equipped with diverse kernel functions are deployed to perform training and prediction based on the selected features. The experimental results reveal that the model attains a high classification accuracy of 94.62% on a self-constructed asphalt pavement image dataset. This technology not only significantly improves the accuracy and efficiency of pavement inspection but also effectively reduces traffic congestion and incremental carbon emissions caused by pavement distress, thereby alleviating environmental burdens. It is of great significance for enhancing pavement maintenance efficiency, conserving resource consumption, mitigating environmental pollution, and promoting sustainable socio-economic development.

Because the fluctuation and uncertainty of wind power generation bring severe secure and economic challenges to power systems, wind power forecasting becomes the critical part of the management in power systems. In this paper, a hybrid approach based on unequal span segmentation-clustering is proposed to mine the variation trend information in the wind speed series for improving forecasting performance. Firstly, the wind speed series representation is proposed to accurately represent the major variation trend of the wind speed series. It reduces the influence of the non-stationary and irregular behaviors of the wind speed series on the unequal span segmentation-clustering. Secondly, the shape-position comprehensive evaluation is proposed to combine shape and position measures to evaluate the clustering of the trend segments with unequal length. It identifies the trend segments which have comprehensive similarity on both shape and position to better construct forecasting models. Thirdly, novel bayesian optimization mutation operators are proposed to optimally move or add cut points in the unequal span segmentation. It enhances the local search capability of the unequal span segmentation. By comparing different approaches using wind datasets from two wind farms, the effectiveness and advantages of the proposed approach are demonstrated.

A lightweight PCSNet-based segmentation model is developed to efficiently overcome insufficient performance in feature extraction and boundary loss information resulting from sampling operations. The proposed approach incorporates two key components: an enhanced shuffle unit and an improved inverted residual architecture to effectively reduce model parameters and enhance inference time. Additionally, introducing the generalized Dice loss (GDL) aims to address the prediction accuracy issue caused by category imbalance. Finally, this study employs gradient-based class activation mapping to visualize and interpret the learned features. To enhance model interpretability, experiments were conducted on three publicly available datasets and compared with existing popular segmentation models. The findings demonstrate that including the GDL in the lightweight PCSNet model leads to a significant improvement in prediction performance, with the highest mIoU reaching 78.93%. Additionally, the visualization of class activation mapping (CAM) further enhances PCSNet interpretability in terms of feature extraction.

A full parametric and linear specification may be insufficient to capture complicated patterns in studies exploring complex features, such as those investigating age-related changes in brain functional abilities. Alternatively, a partially linear model (PLM) consisting of both parametric and non-parametric elements may have a better fit. This model has been widely applied in economics, environmental science, and biomedical studies. In this paper, we introduce a novel statistical inference framework that equips PLM with high estimation efficiency by effectively synthesizing summary information from external data into the main analysis. Such an integrative scheme is versatile in assimilating various types of reduced models from the external study. The proposed method is shown to be theoretically valid and numerically convenient, and it ensures a high-efficiency gain compared to classic methods in PLM. Our method is further validated using two data applications by evaluating the risk factors of brain imaging measures and blood pressure.

Automatic detection technology provides a reliable method for civil engineering distress detection. However, to overcome limitations of computational resources and the significant cost of image acquisition, this study proposes a simplified network parameter-based pavement crack classification network (PCCNet) to achieve efficient and robust crack classification. Firstly, a lightweight classification model is developed based on a shuffle unit and inverted residual architecture, designed to deliver high-performance pavement crack classification with limited computing resources. Secondly, a novel training method is proposed to accurately identify pavement defects on small-sample pavement images datasets. Additionally, the interpretability of neural network in pavement defect detection is enhanced by visualizing training process. The results demonstrate that the model achieved a classification accuracy of 97.89% on the augmented pavement image dataset and a classification accuracy of over 83% on multi-source asphalt pavement images. Furthermore, visualizing intermediate features further enhanced the high-precision recognition ability of the lightweight model.

With the increase of new sensing devices in Internet of things (IoT), data dimensions and types have risen dramatically. The traditional data collection structure cannot satisfy the requirements for multiclass data and access control. By integrating fog computing, we design a layer-aware fog computing model which supports a distributed privacy-preserving compressed sensing (CS) for multiclass data with iden-tity authentication in fog-assisted IoT. In our model, a novel distributed nested CS scheme samples and compresses the encrypted multiclass data in the sensing layer. The encrypted sampling data is trans-mitted to the fog node. Subsequently, the fog node embeds the identity watermark for later VIP user authentication in the cloud. Then, the fog uploads the watermarked data to the cloud for storage and reconstruction. When receiving the request, all results returned by the cloud are encrypted versions. In particular, the cloud performs the designed reconstruction transformation task on the sensing data for security. Specifically, the cloud only reconstructs sensing data for general users, but returns all multiclass data to the authenticated VIP users. In the experiment, we analyze the security of our scheme, discuss the impact of parameters on the reconstruction performance of different data, and summarize the rec-ommended parameter settings.(c) 2022 Elsevier B.V. All rights reserved.

Recently, compressed sensing (CS) based cryptosystem has received extensive attention in information security field. However, this cryptosystem cannot resist known-plaintext attack (KPA) under the secret key multi-time-using (MTU) scenario because of its linear sampling process. To address this concern, a privacy-preserving image CS scheme is proposed, which embeds a controllable noise-injected transfor-mation (CNT) into the linear sampling process of CS to resist KPA. First, a CNT operation is applied to pre-encrypt the image. Second, the pre-encrypted image is re-encrypted and compressed by using CS at the same time. Third, the final CS measurements are quantized into bits. Since the embedding of CNT operation destroys the linearity of CS, the proposed cryptosystem can resist the existing KPA method under the secret key MTU scenario. Lastly, an iterative total variation (TV) algorithm based on ADMM framework, called TV-ADMM, is proposed for image recovery. Simulation results demonstrate that the proposed cryptosystem significantly enhances the security of the CS-based cryptosystem with slightly sacrificing the compression performance in high compression rate cases.(c) 2023 Elsevier B.V. All rights reserved.

To seek a constitutive model that can both describe the viscoelastic response of asphalt mixtures well and be applied to finite-element simulation, a modified fractional-order Zener model (MFZM) was adopted from the mathematical expressions describing viscoelastic materials based on fractional calculus theory in this study. The MFZM was verified by the generalized fractional-Zener model, and the corresponding model configuration was provided. In this model, the dynamic response and relaxation-modulus function were derived. Based on the relaxation-modulus function, a numerical algorithm for the MFZM that can be implemented in a finite-element environment was proposed. The ability of the fractional-Zener model (FZM) and MFZM to characterize the dynamic response of asphalt mixtures was compared using dynamic modulus test. The dynamic modulus and semicircular bending (SCB) tests of the asphalt mixture were simulated using the proposed numerical algorithm in the software ABAQUS. The results demonstrate that, unlike FZM, MFZM can characterize the asymmetric characteristics of viscoelastic response of the asphalt mixture in the frequency domain. The simulation result of the dynamic modulus test agreed with the analytical solution of the MFZM in the frequency domain, which illustrated the availability of the proposed numerical method for the response of the asphalt mixture in the frequency domain. Moreover, the consistency between the actual experimental and virtual results of SCB proved the accuracy of the model and applicability of the proposed numerical method, allowing the model to be used in a finite-element environment.

Summary Spotted mackerel were stored at 5 °C for 0–5 days. To reveal changes in freshness and provide the fundamental knowledge of sarcoplasmic reticulum (SR), structural and biochemical changes were investigated by optical microscope, transmission electron microscopy and biochemical detection, including pH, myosin Ca 2+ ‐ATPase activity, adenosine 5′‐triphosphate (ATP) content, SR recovery field and SR Ca 2+ ‐ATPase (SERCA) activity. The result exhibited that SR swelled on day 1 and ruptured after 2 days. Muscle structure showed significant changes from day 0 to day 5. SERCA activity and SR recovery field were 0.46 μm Pi/min/mg and 1.05 mg/g at day 0 and 0.106 μm Pi/min/mg and 0.109 mg/g at day 2, respectively, highly corresponding to the decline of pH, myosin Ca 2+ ‐ATPase activity and ATP content with a significant difference from day 0 to days 1 and 2. The biochemical properties of SERCA exhibited the maximum activity at pH 6.8–7.0 (0.408 μmol Pi/min/mg). Treated at pH 5.5 for 80 min or incubated at 35 °C for 40 min inactivated SERCA of 80%. Therefore, keeping at a freezing temperature of 5 °C and maintaining SR functionality was essential to delay freshness decrease.

The creep test is used as the main method of evaluating the rutting resistance of asphalt mixtures, but the variability has rarely been studied. Based on creep tests, this study investigated the stochastic nature of the mixture creep through mechanical modeling. First, creep tests were carried out on mixtures under different conditions. Consequently, the creep behavior was characterized using a fractional creep damage model (FCDM) and a variability analysis was performed. Second, a proposed creep stochastic damage model (CSDM), developed using a meso-mechanical approach, was utilized to characterize the variability behavior of the damage evolution, and the range of variation in damage evolution was predicted within a certain confidence interval. Subsequently, combined with FCDM, the variation of mixtures during creep was predicted. This study confirmed that the creep variability of mixtures is related to rutting resistance, and the better rutting resistance implies lower variability. The validation and prediction results show that the CSDM can well describe the stochasticity of damage, and that the variability of creep behavior for asphalt mixtures can be well predicted on the basis of the stochasticity of damage evolution. The physical mechanism of damage stochasticity in the mixture creep was clearly demonstrated on the mesoscopic scale.

本文阐述了谱矩阵的显式分层表达和多维随机振动频域疲劳分析方法,要研究了谱矩阵中的互谱对疲劳寿命预示的影响.基于一个受控的物理试验的实测数据,分别采用相平均统计、时平均统计和互谱置零方法来制定随机激励载荷谱矩阵条件,以此来计算同一机械组件的应力响应,按频域法估计其疲劳寿命,并与实测响应估计的疲劳寿命进行对比.研究结果表明,不同的方法给出的寿命预估值有明显差异,体现了谱矩阵条件制定时互谱保真的重要性.

针对某大型分段式固体火箭发动机试验模态分析,试验结果出现了直观"不平衡"的扭转振型,通过数据分析结合试验模型的可观测性状和固体发动机的结构形式,认为药柱的弹性特性影响是主要因素.由于固体火箭发动机是密封充压结构,用于保护药柱,因此不能对药柱粘贴测点进行验证测试.为此建立固体发动机对比仿真模型,并基于试验数据对仿真模型进行了修正,然后按照模态试验的激振方式进行谐响应分析.分析结果验证了直观"不平衡"的扭转振型,证明了药柱的弹性特性是导致上述现象的原因.当固体火箭发动机直径增加,装药量也将大幅增加,低刚度和粘弹性特性固体推进剂的弹性特性将对整体弹性特性的影响将不可忽视,因此针对大型固体火箭发动机提出了试验模态分析药柱的测量和激振要求.

Although the extensive application of the Internet of Things brings great convenience, it raises the concern of privacy leakage in the processes of data acquisition, analyzing, and sharing as well. In this article, multilevel privacy protection via compressed sensing (CS) is proposed, which has the advantages of compressed sampling, protection of data privacy, and controllability of data access. At the data acquisition end, the CS technique suitable for a resource-constrained environment is employed to sample and encrypt signals with the assistance of discriminant component analysis. Then, the encrypted data will be transmitted to the cloud in time. On the cloud service, signals protected by CS rarely disclose private information to malicious attackers, and they will be accessed by two-class authorized entities. One is the semiauthorized user with low privilege who can only get the features from encrypted data for the subsequent inference; the other is the full-authorized user who is capable of reconstructing the original data. We demonstrate the scheme through two case studies of a face recognition system and a human activity recognition system and analyze its performance.