Search

In order to enhance the suppleness of a lower limb rehabilitation medical robot during the re-habilitation process, this study proposes a multi-source information fusion lower limb motion intention recognition method based on surface electromyographic signals (sEMG) and lower limb joint angles. To solve the problem of data traffic surge during the collection process, a multi-source current limiting sliding time window algorithm (MLS) is proposed. The MLS algorithm controls the data flow through a flow limiting and sliding time window mechanism to ensure the efficiency and stability of the system in handling large data volumes. On this basis, the study combines the Back Propagation Generalized Algorithm Neural-network (BPGN) to construct a prediction model for lower limb joint angles. The experimental results show that under the same conditions of the algorithm, the fusion of multi-source information reduces the average error of knee joint angle prediction by 10.8° and the average error of ankle joint angle prediction by 7.2° compared with the method using a single lower limb joint angle signal. Under the same condition of input signal, the multivariate flow-limiting sliding time-window BPGN reduced the average knee joint error by 13.6° and the average ankle joint angle error by 8.5° compared to the BPGN intent recognition. The multivariate flow-limited sliding time window BPGN reduced the mean knee error by 11.2° and the mean ankle angle error by 7.4° compared to Radial Basis Function (RBF) Neural-network intent recognition. By integrating the sEMG signal and lower limb joint angle information, the system can more accurately capture the patient’s movement intention and realize more precise lower limb rehabilitation training.

The real-time simulation of large-scale agricultural operations will offer farmers data-driven and physically consistent decision support, facilitated by predictive digital twins. To construct a predictive digital twin, the initial step involves 3D reconstruction of plant geometry. In this paper, a high-resolution, accurate 3D reconstruction of tomato plants, Tomato-NeRF, is proposed, which is specially used for three-dimensional reconstruction of tomato plants. Our approach used a modular design to integrate ideas from their research paper into Tomato-NeRF. By using hash encoding to map coordinates to trainable feature vectors, we balance quality, memory usage, and performance in NeRF training. The proposal sampler targets key regions for rendering, and customized loss functions are designed to optimize specific tasks. The effectiveness of our approach is demonstrated by the ability to generate high-resolution geometric models from phone camera data. Comparative results show that Tomato-NeRF has significant advantages over Instant-NGP and MipNeRF in the tomato plant reconstruction task. The data acquisition method is simpler and more efficient than other reconstruction methods, providing a practical solution for real-time agricultural simulations.

With the development of permanent magnet synchronous motors (PMSMs), the accurate loss calculation is of vital significance for the performance evaluation of the PMSM system. As the loss of the PMSM system is determined by the control strategy, inverter, and the motor itself, which interact with each other mutually and complicatedly, in conventional loss calculation methods they are not considered synthetically. To calculate the PMSM system loss accurately, this paper proposes a strategy-circuit-field co-simulation methodology, which can take into account the mutual influence of control strategy, inverter, and motor. The proposed co-simulation platform synchronizes the control theory-based strategy simulation, Kirchhoff’s law-based circuit simulation, and the field-based motor simulation, which reveals the real physical process of PMSM systems. To further improve the loss calculation accuracy, a hysteresis-based iron loss calculation method is introduced and utilized. The co-simulation results show that the proposed method can reflect the behavior of the PMSM system more physically, and mutual influence between the three aspects can be presented. Finally, the corresponding experiments are conducted, and the results are compared with the simulation results, proving the proposed methodology’s feasibility and effectiveness.

Based on the fractional order sliding mode load observer, this paper presents an anti-disturbance speed control method of permanent magnet semi-direct drive transmission system for overhead manned equipment. The dynamic model of semi-direct drive transmission system is established using the Lagrange equation. Furthermore, the mathematical model of the permanent magnet synchronous motor (PMSM) is established based on the coordinate transformation theory. Subsequently, the fractional order sliding mode observer (FOSMO), which combines nonsingular terminal sliding mode control and fractional order theory is designed to observe the load disturbance changes in the permanent magnet semi-direct drive transmission system. Then, the concept of “active damping” is used to design the speed loop PI controller, and combined with the designed FOSMO to construct the composite speed controller. The designed composite speed controller is further improved by the anti-saturation design and parameter optimization. The results demonstrate that the designed FOSMO can accurately estimate the load disturbance on the PMSM output shaft. Additionally, the designed composite speed controller can realize stable speed control in the case of complex load disturbance, which meets the anti-disturbance speed control requirements of the permanent magnet semi-direct drive transmission system, and the control system’s robustness is further enhanced by adding the anti-saturation link.

Real-time simulation is important for ensuring the reliable operation of VSC-HVDC converters in power grids, particularly through the use of rapid control prototyping (RCP) and hardware-in-the-loop (HIL) based converter controllers. While real-time simulation is a common practice for modular multilevel converters (MMCs), it has been less frequently applied to the new class of hybrid cascaded multilevel converters (HCMCs). In this study, a universal equivalent model (UEM) is proposed for a range of HCMC topologies that combines accuracy and computational efficiency through the use of both CPUs and field-programmable gate arrays (FPGAs). The proposed UEM is derived using the hybrid five-level converter (H5LC), a compact, efficient, and fault-tolerant VSC within the HCMC family. The UEM relies on CPUs to simulate the main circuits and controls of the main converter, and utilizes FPGAs to calculate the instantaneous voltages of a large number of full-bridge submodules (FBSMs), flying capacitors, and DC-side pole capacitors. In addition, the FBSMs’ voltage-balancing and switching algorithms are implemented on the FPGAs. The proposed real-time CPU/FPGA-based H5LC-UEM is compared to an offline CPU-based detailed equivalent model to verify its accuracy.

This paper develops a joint strategy of condition-based maintenance and spare ordering for a multistate system that is subject to competing failures due to external shocks and self-degradation. Failures of this system are hidden and can be divided into two types: a one-stage hard failure and a two-stage soft failure. The states of the system are identified by periodic inspections and replacement is executed preventively or correctively in response to the defective or failed state. When the operating time of the system is reached the predetermined threshold $\tau $ , the spare is ordered. Furthermore, a threshold level $z$ is introduced to postpone the preventive replacement (PR) when the ordered spare is delivered before the defective state is first detected. Depending on the state of the ordered spare when replacement is required and the fault cause of the system when a corrective replacement (CR) is needed, all possible renewal events are analyzed to establish the joint optimization model. A modified artificial bee colony algorithm and discrete simulation algorithm are adopted to find the optimal solutions, and the correctness of the proposed model is verified by a numerical example. Moreover, the results from the numerical example indicate that the proposed strategy is superior to the comparative model, thus the applicability and effectiveness of the proposed model are illustrated.

The development of information technology has promoted the expansion of the application field of facial recognition technology. Its mainstream recognition methods rely on deep learning algorithms for calculation, but the problem of large data computation brought by its system makes it difficult to apply to embedded platform devices. As a result, this study focuses on improving recognition systems built on lightweight backend networks and builds an embedded platform system environment using multi-scale feature fusion, anchor box size optimization, the addition of channel attention mechanism weighted features, affine face alignment, and file compilation. The experimental results showed that when the number of iterations was 300, the loss value (0.46) of the improved embedded algorithm was much smaller than that of other comparison algorithms (1.42, 1.73, 2.01), and its ACC value (0.924) was significantly better than other comparison algorithms (0.915, 0.909, 0.894). The minimum system testing time consumedwas7 ms. This deep learning embedded facial recognition algorithm has high recognition accuracy, a fast running speed, and is less limited by environmental conditions and data types. It is ideally suited for use in embedded hardware devices, broadening the scope of equipment matching and facial recognition algorithms’ applications. As a result, it is better suited to satisfy the demands of embedded devices and massive data processing jobs.

As the global pursuit of carbon neutrality has intensified, the influence of carbon emission limits on the supply chain has become increasingly evident. The purpose of this study is to investigate the optimal financing options of the supply chain under the carbon emission reduction percentage measure and to provide financing recommendations for carbon-emitting enterprises. This study focuses on a supply chain consisting of a capital-constrained manufacturer and a capital-rich retailer. Under the carbon emission reduction percentage measure, we use the Stackelberg game theory to determine the manufacturer’s and retailer’s optimal decisions for bank financing, zero-interest early payment financing, and in-house factoring financing. Finally, we find a U-shaped relationship between the price sensitivity of commodities and the retailer’s (manufacturer’s) optimal profit. The impact of the carbon emission reduction percentage on the supply chains’ financing decisions follows a quadratic pattern. Carbon emission trading prices exhibit a monotonic influence on the equilibrium financing options of the supply chain. When carbon emissions do not exceed the standards, internal financing becomes more favorable as the carbon emission trading price increases. Conversely, when carbon emissions exceed the standards, external financing becomes more prevalent as the carbon emission trading price increases. The findings assist supply chain managers in selecting the optimal financing method when confronted with carbon emission limitations. These conclusions contribute to the development of more targeted measures and the promotion of green supply chains.

Broad learning system has attracted increasing attention in the hyperspectral image (HSI) classification, due to its universal approximation capability and high efficiency. However, two main challenges remained. One is that the mapping from the original BLS input to the mapped feature (MF) is linear, which is difficult to fully represent the complex spatial-spectral features of HSI. The other is that BLS fails to explore the local geometric structure relationship between samples within HSI. To overcome the limitations mentioned above, we propose a local discriminative embedding broad learning system with graph convolutional (GDEBLS). To address the first challenge, GDEBLS utilizes the graph convolution operation to aggregate the node information in the adjacent graph to learn the context and obtain the rich nonlinear spatial-spectral features in HSI. To deal with the second challenge, our method utilizes a neighborhood selection approach based on manifold structure to calculate the true distances between samples in the manifold space, overcoming the limitations of Euclidean distance measurement. Next, We introduce local manifold structure and discriminative information into BLS. The experimental results show that the proposed method significantly surpasses other state-of-the-art methods.

The current detection methods for grassland fires mainly rely on manual means, which are costly, inefficient, and difficult to achieve real-time and full coverage detection. Therefore, the YOLOv5m-D model and its static and dynamic characteristics are proposed to detect and identify smoke and flame throughout the entire process of grassland fires, and its effectiveness is verified. The experimental results showed that in smoke recognition and detection, YOLOv5m-D model showed slow local convergence under the condition of low Learning rate, and the mAP value of YOLOv5m-D was 86.4% when the batch size was 16. In the comparison of mAP values under the optimal hyperparameters, the Faster RCNN value was 72.34%, SSD value was 75.90%, YOLOv5m value was 86.75%, and YOLOv5m-D value was 89.28%, which was higher than the comparison model. In flame recognition detection, in the Hu1 moment, the ordinary image sequence and infrared thermal imaging sequence, except for a few that were around 0.8, were mostly maintained at around 0.3-0.7, while the color tent was both below 0.1. After combining infrared images with flame static and dynamic feature recognition, the flames are basically recognized. In the comparison of single feature recognition time, the current frame recognition time of different images under the vast majority of features is lower than the reference frame. Overall, the YOLOv5m-D model proposed in the study and its static and dynamic characteristics are effective in detecting and identifying smoke and flame throughout the entire process of grassland fires, and have high practical effects for practical grassland fire detection.

With the development of computer technology and expanding environmental issues, machine learning has received more and more attention in the field of weather forecasting. Global Navigation Satellite System-Radio Occultation(GNSS-RO) technology is a kind of remote sensing technology. This investigation proposes an alternative to numerical weather forecasting model. The new method is based on machine learning utilizing GNSS-RO data to forecast the wind field in the Beijing-Tianjin-Hebei region of China. The dataset including temperature, humidity, pressure, wind speed and direction was obtained by numerical calculation in terms of historical monitoring data in Beijing-Tianjin-Hebei region. Then the models of wind fields forecasting based on machine learning were established with different neural network including Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN) and Deep Neural Networks (DNN). The prediction performance of different models was analyzed. The results demonstrate that LSTM and CNN have better performance on predicting the wind field than Deep Neural Networks. The wind speed error is about 1.4m/s, and the wind direction error is about 30°. Moreover, the time required for neural network to predict a new sample is about 1 second, which is only 0.2% of the prediction time compared with numerical model. Finally, the machine learning model can be used to predict the wind field effectively, with GNSS-RO data as the input in application. This paper pro-vides a new method in sight to use machine learning to forecast the regional wind field utilizing GNSS-RO data.

A decomposition-coordination optimization method termed as analytical target cascading-based exterior penalty function (ATC-EPF) is proposed to investigate the reliability allocation aiming at the low computational efficiency for large complex systems. Firstly, the optimization model is established via ATC-EPF, which is decomposed into the models of subsystems. Then the subsystems are calculated in parallel, which reduces the computational cost greatly in the condition of keeping the computational accuracy. The results are fine which are verified by two study cases including one numerical case and one engineering case. This research is of great significance to large-scale system planning and design.

Based on the mechanical model of the point absorption Wave Energy Converter (WEC), a real-time complex conjugate (CC) maximum power point tracking (MPPT) control strategy consider power peak-to-average ratio (PAR) limiting under irregular waves is proposed. In the proposed MPPT strategy, the system can not only track the maximum power, but also limit the PAR of the output power to improve the controlled generator’s utilization rate. Two parameters, $R_{epot}$ and $X_{pto}$ , that determine the command current of the PMSG and the PAR, are introduced in the mathematical model of the system to control the system. And they are identified by the Grey Wolf Optimization (GWO) algorithm. During each half cycle of the sea wave, the parameters $R_{pto}$ and $X_{pto}$ will be real time updated to satisfy the control strategy. In addition, the system parameters $A$ , $B_{\mathrm {rad}}$ and $F(\omega)$ related to wave frequency are also calculated by the advanced quantitative wave analysis (AQWA) software. Finally, the proposed strategy is verified by the comparison on the simulation and the experimental results.

Virtual machine consolidation (VMC) is an effective way to solve the problems of high power consumption and low utilization in cloud data centers. However, large-scale virtual machine migrations (VMMs) can result in additional workloads, service-level agreement violations (SLAVs), and considerable energy consumption (EC). Existing studies have made great progress in this respect, but the following problems remain: first, the potential overload of the physical host is not considered in the load detection of the physical host; second, the resource-demand scaling of physical hosts is not considered during virtual machine (VM) placement, which results in the lack of accuracy in selecting suitable hosts. In view of the above problems, this study firstly constructs a virtual resource consolidation model based on green energy conservation (GEC-VRCM), which defines the specific process and related attributes of VMC, which is beneficial to improve the consolidation efficiency of virtual resources. Second, based on this model, we propose a dynamic virtual machine consolidation algorithm based on balancing energy consumption and quality of service (EQ-DVMCA) to achieve efficient consolidation of virtual resources. Finally, experiments show that, compared with the selected 12 benchmark algorithms and two advanced VMC algorithms, EQ-DVMCA not only reduces the number of VMMs and EC, but also maintains a high level of Quality of Service (QoS) and achieves a balance between EC and QoS.

The discussion context of this paper is big data processing of MapReduce by volunteer computing in dynamic and opportunistic environments. This paper conducts a series of simulations to explore the relationship between the overall performance of volunteer overlays responding to different workload of big data problems. The discovery from the simulations includes some optimization points in overlay size, going over which by adding more volunteers brings little benefit for the overall performance. Based on the discovery of optimization points, this paper proposes a bootstrapping protocol, which can adapt volunteers into variable-sizes overlays, enabling workflow of single-round MapReduce or multiple-round MapReduce, and a single or multiple overlays for each round. The variable overlays aim to create adaptive workflow during MapReduce processing, so that the optimization points can be caught. As another benefit, the unnecessary computing-capacities can be released during computing when the optimization points are reached. The case study shows a few optimization workflows that are formed by the proposed bootstrapping protocol to process the big data cases. The workflows lead to the optimization points and dynamically balance the workload at the same time. The experiment results have demonstrated that the optimization strategies have either achieved 36% or 71% higher performance than the plain MapReduce workflow and minimized the use of computing resources by releasing 12.5% to 75% volunteers during computing, where the original plain MapReduce must hold all the volunteers to the end of computing. The extensibility of the simulation parameterization to more diverse real-world applications have been clarified.

The insulation performance of ethylene propylene diene monomers (EPDMs) subjected to multi-stresses in the field is always obscure. To understand the effects of temperature and pressure on the performance of the EPDMs used as insulators for power cables, in this work, focuses were on the dielectric losses of EPDMs under the temperature range of 30– $120^{\circ }\text{C}$ and pressure range of 2–8 bar. The polarization and depolarization currents and the dielectric loss factor of EPDMs under the above conditions have been first measured. The relationship between the depolarization current and dielectric loss factor has been studied based on the correlation of the depolarization charge and the dielectric loss factor at low-frequency. The experimental results showed that the conduction currents and the dielectric loss factor increase with the increase in temperature, but the effect of pressure on the dielectric properties of EPDM is more complicated. There is a strong correlation between the depolarization charge and low-frequency dielectric loss factor, both of which can reflect the change law of the EPDMs’ performance. Moreover, the asymmetry coefficient of the dielectric loss factor at low-frequency can be a supplement with regard to describing the deterioration degree of EPDM insulation. The evaluation chart of the insulation performance of the EPDMs has been advanced according to the asymmetry coefficient and the total dielectric loss calculated by the polarization and depolarization currents.

In this work, we focus on the model-mismatch problem for model-based subspace channel tracking in the correlated underwater acoustic channel. A model based on the underwater acoustic channel’s correlation can be used as the state-space model in the Kalman filter to improve the underwater acoustic channel tracking compared that without a model. Even though the data support the assumption that the model is slow-varying and uncorrelated to some degree, to improve the tracking performance further, we cannot ignore the model-mismatch problem because most channel models encounter this problem in the underwater acoustic channel. Therefore, in this work, we provide a dynamic time-variant state-space model for underwater acoustic channel tracking. This model is tolerant to the slight correlation after decorrelation. Moreover, a forward-backward Kalman filter is combined to further improve the tracking performance. The performance of our proposed algorithm is demonstrated with the same at-sea data as that used for conventional channel tracking. Compared with the conventional algorithms, the proposed algorithm shows significant improvement, especially in rough sea conditions in which the channels are fast-varying.

In this paper, a microwave polarizer based on frequency selective surface (FSS) loaded with complementary split ring resonator (CSRR) for non-planar and conformal application is presented. The frequency-selective polarizer is fabricated on liquid crystal polymer substrates (LCPs) material with $50\mu \text{m}$ thickness, which shows very good mechanical flexibility for microwave band. The Frequency-selective polarizer consists of a metal FSS screen pair with double complementary split ring resonators aligned in the vertical direction without any displacement. The substrate integrated waveguide (SIW) cavity was introduced to enhance the Q-factor of the CSRR particles and the mechanical strength of the FSS screens at bending conditions. The measured results show that the CSSR-FSS polarizer with the minimum transmission loss for TE-polarization incidence wave is 0.75dB at the 10.24 GHz and 19.05 dB for TM-polarization incidence wave, with an extinction ratio of 18.3dB. Finally, the oblique incidence from normal to 60 degrees and bending tests with the bending radius of 9.8 cm and 8.8 cm of the CSRR-FSSs were also carried out, and the measured results indicate that the CSRR-FSS show a low susceptibility to bending behavior at the small bending radius beyond the 8.8 cm and different incidence angles from normal to 45 degrees. Both the simulated and experiment data indicated that the proposed ultra-thin CSRR-FSS shows excellent flexibility and polarization selectivity.

In this paper, we propose to apply data augmentation approaches that provide more diverse training images, thus helping train more robust deep models for the Scene Text Recognition (STR) task. The data augmentation methods are Random Blur Region (RBR) and Random Blur Units (RBUs). Specifically, we first introduce RBR designed for the STR task. In training, RBR randomly selects a region and sets the pixels in this region with an average value. However, when RBR provides more various training samples for STR, it may make the samples ambiguous and reduce the recognition accuracy. To address the above problem, we also propose RBUs that divides the blur region into several units. Note that the pixels of one unit share the same value. In this way, RBUs can provide additional readable training samples and help train more robust deep models. Extensive experiments on several STR datasets show that RBUs achieve highly competitive performance. Besides, RBUs are complementary to commonly used data augmentation techniques.

In order to solve the problem of ship's curve trajectory-tracking control, the Norrbin nonlinear response model which can accurately describe the ship's motion state is selected in this paper. The hyperbolic tangent function is used to design the expected hemispheric angle equation, then the complex track control is transformed into a heading control problem. The fast terminal sliding mode (FTSM) is introduced together with the Backstepping control technique to reduce the system adjustment time, eliminate the chattering. Bying combined with extended states observer (ESO) and dynamic surface control (DSC) technique, the internal and external disturbances in real-time can be estimated and compensated, and the “explosion of complexity” caused by backstepping technique is solved. The state of the control system is bounded and stable, and the system error converges to zero. Matlab simulation proves that the controller can realize the trajectory-tracking control quickly and accurately, and has strong robustness to external disturbances.

The realistic logistics not only involves vehicle capacity and working hours, but also factors such as the second trip of vehicles and heterogeneous vehicles. Aiming at the constraints of these VRP variants, an algorithm is proposed to solve the heterogeneous vehicle routing problem with a second trip. In the process of calculating the initial solution, the concept of k-ring shaped Voronoi neighbors reduces the computational complexity and avoids the problems caused by outliers. When allocating customers, the idea of “Borrow-In and Borrow-Out” proposed in this article is adopted to rationalize the route planning results. In the calculation process of considering the second trip, the customers closer to the depot are regarded as the final order. For heterogeneous vehicles, this article adopts a greedy strategy. The experimental results of multiple instances show that if the depot provides a heterogeneous fixed vehicles fleet, compared to when only a single vehicle type is provided, the result is a reduced number of vehicles used by 4.46% on average, and the loading utilization rate of vehicles is increased by 4.02% on average. And after the calculation of considering the second trip, the number of vehicles used is reduced by 5.09% on average, and the loading utilization rate of vehicles is increased by 4.41% on average. It can be seen that the algorithm proposed in this article can obtain high-quality route allocation results for large-scale orders.

Accurate segmentation and analysis for each animal in surveillance video images will help poultry farmers to monitor and promote animal welfare. However, it is challenging to accurately segment each animal due to the similar appearance, different scales, rapid growth and adhesive areas of group animals. Meanwhile, lacking of useful training data also limits the effectiveness of animal segmentation algorithms. To address these problems, we first construct a chicken image segmentation dataset to study the behavior of chickens for intelligent monitoring and analysis. Then, we propose an effective end-to-end framework for chicken image segmentation, which can also be used for other animal image segmentation. An end-to-end multi-scale based encoder-decoder network is first utilized to extract multi-scale features. Then, an attention-based module is employed to extract and intensify effective features, thus better segmentation results can be obtained. Finally, a multi-output combined loss function is proposed to make effective supervision for better segmentation. Experimental results demonstrate the promising performance of the proposed framework for chicken image segmentation.

Communication overhead is a key task for the large-scale wireless sensor networks (WSNs) in the cooperative localization problems. This paper is aimed to propose a low-overhead algorithm without sacrificing the performance of localization, which is more practical for energy-saving WSNs. In this paper, the cooperative localization problem is formulated as a variational reference problem on a factor graph (FG). Combining the high performance of particle-based algorithms and the overhead advantage of parametric ones, a parametric variational sum-product algorithm (PVSPA) is proposed for the cooperative localization of WSNs. In prediction operation, to ensure high accuracy, messages referenced inside sensor nodes are particle-based. In correction operation, Gaussian parametric representation is adopted to the observation message and the communication overhead is obviously decreased. Simulation results show that the localization accuracy of PVSPA equally matches the classic particle-based algorithms, while the communication overhead of it is far lower.

This paper presents a $0.2\sim 3.8$ -GHz in-band (IB) full-duplex (FD) mixer-first receiver based on a vector self-interference (SI) canceller. To ensure the wideband orthogonality of the reconstructed SI signal while minimizing the noise figure (NF) degradation in FD mode, a low insertion loss two-stage polyphase filter (PPF) with a lacking resistance and capacitance (LRC) structure is used in the SI canceller. A 6-bits passive capacitor digital to analog converter (C-DAC) attenuator not only realizes the high-precision amplitude adjustment of the reconstructed SI signal but also avoids the linearity limitation of the back-end modules of the canceller and reduces the noise contribution of the front-end modules at the canceller output. A four-phase mixer-first receiver with enhanced out-of-band (OOB) interference suppression is designed to provide sufficient linearity to reduce the SI-induced nonlinear distortion. As a proof of concept, a prototype FD receiver is fabricated in CMOS 65 nm technology and measured in chip-on-board (COB) package. Measurement results show that it can operate in the frequency range of $0.2\sim 3.8$ -GHz with more than 25 dB SIC amount for 10 MHz bandwidth (BW) 64-QAM modulated signals in fixed (30 dB) channel attenuation test setup. The receiver achieves OOB input-referred third-order intercept point (OOB-IIP3) of +24 dBm and OOB input-referred second-order intercept point (OOB-IIP2) of +82 dBm at 1-GHz local oscillator (LO) frequency. In FD mode, the NF degradation is $\leqslant 1.5$ dB. The effective IB-IIP3 and effective IB-IIP2 are enhanced by 15 dB and 28 dB, respectively. The power consumption of entire FD receiver including the SI canceller is less than 44 mW. The active chip area is 0.55 mm2.

The electric shovel is a bucket-equipped mining excavator widely used in open-pit mining today. The prolonged direct impact of the bucket teeth with hard and abrasive materials such as ore during the process of the mining excavation can cause the bucket teeth to break and fall off prematurely, resulting in unplanned downtime and productivity losses. In response to this problem, we have developed a vision-based bucket teeth fault detection algorithm with deep learning. Using a dataset based on the images of both real shovel teeth and 3D-printed models, we trained a Faster Region Convolutional Neural Network (Faster R-CNN) to obtain the number of normal bucket teeth and the positions of the bucket teeth from the images, using the additional bucket dataset from 3D-printed models to pre-train the network for improving its detection accuracy on the real bucket data. We compared the resulting Faster R-CNN model with the ZFNet, the ResNet-50, and the VGG16 and found our Faster R-CNN model to perform best in terms of accuracy and speed.

Automatic modulation recognition with deep learning (DL) is challenging in distinguishing high-order modulation modes and balancing complexity against recognition accuracy. In this paper, we propose a novel dual-path modulation recognition framework named IRLNet, which consists of the improved residual stacks (IRS) and long short-term memory (LSTM). The IRS maintains the more initial residual information, learns the signal features in deep and shallow, and achieves various degrees of feature extraction. The model learns from the time domain I/Q, amplitude and phase information presented in the training data. The simulation results on RadioML2016.10B show that IRLNet performs stable in the training stage and has low spatial-temporal complexity. It also achieves a recognition accuracy of more than 93% at high signal-to-noise ratios (SNRs). Transfer learning is introduced to improve the efficiency of retraining, and robustness is proved by transferring the model to RadioML 2018.01A and HisarMod 2019.1. The simulation result shows that the training time is greatly shortened by about 25.6% in RadioML 2018.01A by introducing transfer learning. Moreover, IRLNet improves the confusion of high-order modulations and achieves a recognition accuracy of more than 90% in high SNRs.

When the large-scale wind power is sent out through the high voltage direct current (HVDC) transmission system and a DC commutation failure occurs, the voltage of AC bus at the sending end decreases first and then increases. Suppose the reactive power supported in the low voltage ride-through process by various reactive resources is not timely returned. In that case, it may aggravate the voltage rise caused by the commutation failure, and the off-grid risk of wind turbine under high-voltage will be aggravated. In order to reduce the off-grid risk of wind turbines caused by the DC commutation failure, a transient voltage control strategy of DC sending-end regulator based on the online sequential extreme learning machine (OS-ELM) voltage prediction model is proposed. Firstly, the influence factors of commutation failures are analyzed. Aiming at the key factors, the real-time voltage comprehensive prediction model based on OS-ELM is used to predict the voltage increase during the commutation failure process and uses the voltage prediction results to optimize the transient response of the synchronous condenser. A large-scale wind farm together with the HVDC system is established in PSCAD to verify the effectiveness of the proposed scheme. Simulation results show that the proposed scheme can reduce the risk of wind power off-grid risk under DC commutation failures and increase the speed of voltage recovery at the point of common coupling.

In recent years, the serious vibration and noise problem of high voltage filter capacitor has become the primary problem of noise reduction in high voltage direct current (HVDC) converter station. The current research mainly focuses on the experimental measurement of the vibration and noise of the filter capacitor, without considering the influence of the residual charge on the dielectric interface stress. In this paper, the experimental platform is built for measuring the residual charge of the dielectric film in filter capacitor component, and the measurement results show that there are non-uniform residual charges on the dielectric surface of filter capacitor. In order to study the influence of uneven distribution of charge on dielectric interface stress, 12 different residual charge distributions are designed, and the results show that the uneven distribution of surface residual charge has a great influence on dielectric interface stress, and the higher uneven distribution of residual charge can cause the larger the stress on the dielectric surface of filter capacitor. The negative residual charge can lead to larger dielectric interface stress, which has a more serious impact on the vibration of filter capacitor component.

The design of railway alignment and station locations involves two intertwined problems, which makes it a complex and time-consuming task. Especially in mountainous regions, the large 3-dimensional (3D) search spaces, complex terrain conditions, coupling constraints and infinite numbers of potential alternatives of this problem pose many challenges. However, most current optimization methods emphasize either alignment optimization or station locations optimization independently. Only a few methods consider coordinated optimization of alignment and stations, but optimize them sequentially. This paper proposes a concurrent optimization method based on a 3-dimensional distance transform algorithm (3D-DT) to solve this problem. It includes the following components: (1) To optimize the location of stations within specified spacing intervals, a novel perceptual search strategy is proposed and incorporated into the basic 3D-DT optimization process. (2) A combined-alignment-station 3D search neighboring mask is developed and employed to search for both the alignment and stations. In order to implement the perceptual process, two additional kinds of 3D reverse perceptual neighboring masks are also developed and employed in the algorithm. (3) Multiple coupling constraints between alignment and stations are also formulated and addressed during the search process. In this study, the effectiveness of the method is verified through a real-world case study in a complex mountainous region. The optimization results show that the proposed method can find high-quality alternatives satisfying multiple coupling constraints.

A novel adaptive tracking control strategy without singularity is proposed for quadrotor systems subject to uncertain dynamic parameters. Based on the design technique of bounded control, a new differentiable saturated function is proposed and subsequently adopted to ensure that the developed translational thrust remains within the preassigned range in the position control loop. A technical lemma is established to analyze the relationship between the convergence of the attitude tracking error and the position tracking error in the presence of thrust saturation. Then, by the compensation term in the form of an attitude error function, a singularity-free torque controller, which can avoid analytic calculations involving the desired attitude derivative, is presented in the attitude control loop. Finally, we prove that the proposed method can achieve asymptotic tracking without singularity. Simulation results illustrating the effectiveness of the proposed strategy are exhibited.

Mura is defined as visual unevenness on the display panel. It can cause unpleasant feelings, so it is necessary to perform Mura inspection during the display quality test. However, Mura is quite difficult to be detected because of its irregular shape and size as well as its low contrast. To solve this practical problem, we proposed a GAN-based model named UADD-GAN to detect Mura in this work. Consisting of a proposed UADD generator and a discriminator, the model is trained using only normal samples, after which the generator is able to simulate the distribution of normal samples. During training, the generator takes normal images as inputs and outputs their reconstructions, while the discriminator receives images and determines whether they are original or reconstructed ones, defiantly helping the generator to perform reconstructions better. The symmetric structure and the skip-adding operation make it easy for the UADD generator to reconstruct the normal samples well. In the detecting procedure, the generator performs worse in the reconstruction of samples with Mura so that we can distinguish them from the normal ones. To make full use of discriminator, we use multiple feature layers of the discriminator for supervision instead of using only the classification layer, helping the generator to reconstruct normal samples better. Meanwhile, a two-side detecting method was used to detect Mura since all the samples are not in a square shape and it greatly improved the detecting accuracy. We have conducted experiments of Mura data sets with different proportion and our research indicates that our proposed method surpasses other state of the art methods.

Surgery is the most commonly used method of curing inverted papilloma (IP) or nasal polyp (NP). Although accurate preoperative recognition by computed tomography (CT) is a critical aspect of surgical planning, the minor CT imaging differences in such lesions may be a challenge. Therefore, we have devised a deep learning framework for automatic recognition of IP and NP in CT. The proposed framework involves two major steps: (a) use of a convolutional neural network (CNN) to preclassify lesions and (b) automatic IP/NP recognition. The preclassify CNN enables classification of CT slices according to anatomic structure. Separate networks are then implemented to differentiate IP and NP accordingly. Once the framework was trained using a CT dataset (5681 slices) from 136 patients, it outperformed other methods during evaluation, achieving 89.30% accuracy (area under the curve [AUC]=0.95) in classification. The proposed framework has clear potential as a clinical tool, enabling effective and highly accurate preoperative recognition of IP and NP.

Automatic image annotation is a key technology in image understanding and pattern recognition, and is becoming increasingly important in order to annotate large-scale images. In the past decade, the nearest neighbor model-based AIA (Automatic image annotation) methods have been proved to be the most successful in all classical models. This model has four major challenges including semantic gap, label-imbalance, wider range labels, and weak-labeling. In this paper, we propose a novel annotation model based on three-pass KNN (k-Nearest Neighbor) to address the aforementioned challenges. The key idea is to identify appropriate neighbors at each pass KNN. In the first pass KNN, we identify the several most relevant categories based on label feature rather than visual feature as traditional models. In the second pass KNN, we determine the relevant images based on multi-modal (visual and textual label) embedding features. As the test image has not been annotated with any label, we propose a pre-annotation strategory before image annotation to improve the semantic level. In the third pass KNN, we capture relevant labels from semantically and visually similar images and propagate them to the given unlabeled image. In contrast with traditional nearest neighbor based methods, our method can inherently alleviate the problems of semantic gap, label-imbalance, and wider range labels. In addition, to alleviate the issue of weak-labeling, we propose label refinement for training images. Extensive experiments on three classical benchmark datasets and MS-COCO demonstrate that the proposed method significantly outperforms the state-of-the-art in terms of per-label and per-image metrics.

Min-Max algorithm was widely used as a simple received signal strength (RSS-) based algorithm for indoor localization due to its easy implementation. However, the original Min-Max algorithm only achieves coarse estimation in which the target node (TN) is regarded as the geometric centroid of the area of interest determined by measured RSS values. Although extended Min-Max (E-Min-Max) methods using weighted centroid instead of geometric centroid were recently proposed to cope with this problem, the improvement in the localization accuracy is still limited. In this paper, an improved Min-Max algorithm with area partition strategy (Min-Max-APS) is proposed to achieve better localization performance. In the proposed algorithm, the area of interest is first partitioned into four subareas, each of which contains a vertex of the original area of interest. Moreover, a minimum range difference criterion is designed to determine the target affiliated subarea whose vertex is “closest” to the target node. Then the target node’s location is estimated as the weighted centroid of the target affiliated subarea. Since the target affiliated subarea is smaller than the original area of interest, the weighted centroid of the target affiliated subarea will be more accurate than that of the original area of interest. Simulation results show that the localization error (LE) of the proposed Min-Max-APS algorithm can drop below 0.16 meters, which is less than one-half of that of the E-Min-Max algorithm, and is also less than one-seventh of that of the original Min-Max algorithm. Moreover, for the proposed Min-Max-APS, 90% of the LE are smaller than 0.38 meters, while the same percentage of the LE are as high as 0.49 meters for the E-Min-Max and 1.12 meters for the original Min-Max, respectively.

The traffic scene understanding is the core technology in Intelligent Transportation Systems (ITS) and Advanced Driver Assistance System (ADAS), and it is becoming increasingly important for smart or autonomous vehicles. The recent methods for traffic scene understanding, such as Traffic Sign Recognition (TSR), Pedestrian Detection, and Vehicle Detection, have three major shortcomings. First, most models are customized for recognizing a specific category of traffic target instead of general traffic targets. Second, as for these recognition modules, the task of traffic scene understanding is to recognize objects rather than make driving suggestions or strategies. Third, numerous independent recognition modules disadvantage to fusing multi-modal information to make a comprehensive decision for driving operation in accordance with complicated traffic scenes. In this paper, we first introduce the image captioning model to alleviate the aforementioned shortcomings. Different from existing methods, our primary idea is to accurately identify all categories of traffic objects and understand traffic scenes by making full use of all information, and making the suggestions or strategy for driving operation in natural language by using Long Short Term Memory network (LSTM) rather than keywords. The proposed solution naturally solves the problems of feature fusion, general object recognition, and low-level semantic understanding. We tested the solution on our created traffic scene image dataset for evaluation of image captioning. Extensive experiments including quantitative and qualitative comparisons demonstrate that the proposed solution can identify more objects and produce higher-level semantic information than the state-of-the-arts.

Unlike the conventional coverage issue in camera sensor networks, the airport surface surveillance brings new challenges for achieving the desirable coverage. One of the biggest challenges comes from the deployment of camera sensors, i.e. due to the consideration of safety, the cameras are always installed on the land-side facilities in airports. In this paper, we propose a coverage enhancing method by deploying additional cameras for the boundary deployed heterogeneous camera networks. We convert the area coverage problem to maximize the barrier coverage in the area of interest by a boundary deployed camera network. The proposed method is divided into three phases: detecting the uncovered gaps in barriers; finding the optimal positions for deploying additional cameras; and optimizing the orientation of the additional cameras to cover the uncovered gaps. The simulation results confirm that our algorithm has better performance in coverage improvement and additional camera usage compared with other related methods.

Deep learning (DL) has emerged as a powerful image processing technique that learns the features of the data and produces state-of-the-art prediction results. The decade from 2010 to 2020 is a real revival of DL, which has come to a turning point in history. In image classification, many deep learning networks have been proposed by scholars, and each of them has its own strengthness. It is very important and efficient for the researchers and the developers to know the performance of these networks, especially for the beginners, so as to give them a transplant instruction by an objective evaluation index. In this paper, we constructed different data sets from three aspects, texture, shape, and measurement scale to test the performance of nine mainstream image classification networks. Cross-contrast experiments were performed to analyze the sensitivity of factors which influence the stability of image classification networks. Experimental results shown that in the 27 image datasets generated by the three image factors, the classification performance of AlexNet, GoogleNet, VggNet, and DenseNet is better. The perfomance comparison of these networks are showed and discussed in details. Code and pretrained models are available at https://github.com/liqiang12689/image-classification-finall.

IP protocol is the core of TCP/IP network layer. However, since IP address and its Domain Name are allocated and managed by a single agency, there are risks of centralization. The semantic overload of IP address also reduces its scalability and mobility, which further hinders the security. This paper proposes a co-governing Multi-Identifier Network (MIN) architecture that constructs a network layer with parallel coexistence of multiple identifiers, including identity, content, geographic information, and IP address. On the management plane, we develop an efficient management system using consortium blockchain with voting consensus, so the network can simultaneously manage and support by hundreds or thousands of nodes with high throughput. On the data plane, we propose an algorithm merging hash table and prefix tree (HTP) for FIB, which avoids the false-negative error and can inter-translate different identifiers with tens of billions of entries. Further, we propose a scheme to transport IP packets using CCN as a tunnel for supporting progressive deployment. We deployed the prototype of MIN to the largest operators' network in Mainland China, Hongkong and Macao, and demonstrated that the network can register identifier under co-governing consensus algorithm, support VoD service very well.

Spectral efficiency and transmission security are both essential metrics for wireless communication. Rate compatible modulation (RCM) is an adaptive transmission scheme to improve the spectral efficiency. However, the mapping matrix of RCM is fixed. Once the mapping matrix is known to eavesdroppers, information will be leaked. Physical layer encryption (PLE) is a secure approach to enhance the security using encryption at physical layer. In this paper, we propose a secure transmission scheme for RCM, namely, secure RCM (S-RCM). The construction method of RCM mapping matrix is of great random. Combined with PLE, S-RCM mainly uses the secret key extracted from the channel information to generate mapping matrix. Due to the randomness of channel, the mapping matrix will also be random. Security analysis shows that the demapping of S-RCM must be based on the right mapping matrix. It also shows that S-RCM owns extremely large cipher mapping matrix space, which makes it impossible for eavesdropper to guess the mapping matrix. Furthermore, known-plaintext attack (KPA) is hardly impossible to get the mapping matrix due to the high calculations and multiple possibilities. Simulation result shows that the S-RCM have the same throughput performance and bit error rate (BER) performance as RCM. In addition, the simulation results also show that eavesdropper faces extremely high complexity, which is thought safe.

Given that humans and animals can learn new behaviors in a short time by observing others, the question we need to consider is how to make robots behave like humans or animals, that is, through effective demonstration, robots can quickly understand and learn a new ability. One possible solution is imitation based meta-learning, but most of the related approaches are limited in a particular network structure or a specific task. Particularly, meta-learning methods based on gradient-update are prone to overfit. In this article, we propose a generic meta-learning algorithm that divides the learning process into two independent stages (skill cloning and skill transfer) with a noise mechanism which is compatible with any model. The skill cloning stage enables a good understanding of the demonstration, which helps the skill transfer stage when the robot applies the learned experience into new tasks. The experimental results show that our algorithm can alleviate the phenomenon of overfitting by introducing a noise mechanism. Our method not only performs well on the regression task but is significantly better than the existing state-of-the-art one-shot imitation learning methods in the same simulation environments (i.e., simulated pushing and simulated reaching).

In recent years, speech recognition technology based on deep learning model has made great progress, and the accuracy of speech recognition has reached more than 90%. In foreign language learning, speech evaluation is an important application. Billions of foreign language learners need to practice effective pronunciation. However, due to the different goals between speech recognition and speech evaluation, a single speech recognition model cannot be directly applied to pronunciation evaluation. This paper proposes a DDNN (double-layer deep neural network) model, which includes the speech text alignment model and speech recognition model. In the first layer of the speech alignment model, a new Viterbi algorithm method is proposed to find the best path for the alignment of speech and text. In the second layer of speech evaluation and scoring, we are the first to use the CNN (Convolutional Neural Network) and RNN (Recurrent Neural Network) on the encoding part of Attention. The accuracy of CTC model reaches 76.7%, and that of attention model is 81.2%. The experimental results show that the speech and text alignment method is effective, and the speech evaluation results based on the Attention model are better. The FRR (false rejection rate), FAR (false acceptance rate), and DER (diagnostic rate) in the Attention model were 4.5%, 5.1%, and 17.9%, respectively. At the same time, the evaluation of each sentence of the DDNN model in the online experiment is within 1 second, so the model can also be applied to the online real-time evaluation of speech pronunciation.

Recently, unsupervised monocular training methods based on convolutional neural networks have already shown surprisingly progress in improving the accuracy of depth estimation. However, the performance of these methods suffers deeply from problematic pixels such as occluded pixels, low-texture pixels, and so on. In this paper, we introduce a method to a mask by the statistic of error maps for segmenting the problematic pixels. Different from the conventional methods which use additional segmentation networks to classify problematic pixels, we use a multi-task learning architecture to generate identical mask, mean mask, and variance mask for filtering the problematic pixels. Experimental results show that our proposed method has satisfactory performance compared with other relative methods on the KITTI dataset. Moreover, we also apply our method to the UAV dataset VisDrone, and the results also indicate the effectiveness of the method in detecting moving objects.

Mobile crowdsourcing (MCS) has shown great potential in received signal strength (RSS) fingerprint collection, in which an incentive mechanism plays a critical role to motivate users' participation. However, how to quantify the quality of the gathered fingerprint data is still not addressed well in the design of incentive mechanism for MCS-based fingerprint collection. In this paper, a sensing quality metric is proposed to characterize the joint impact of users' privacy protection and the spatial coverage of the submitted data. Given a limited budget, a basic incentive mechanism is devised to recruit appropriate users to maximize sensing quality. Considering that the cost of each user is regarded as private information and users may be attempted to misreport their costs to increase the revenue. Hence, an auction-based incentive mechanism is proposed to achieve the truthfulness of users' costs, which is truthful, individually rational, computationally efficient and budget feasible. Simulation results show that our proposed schemes outperform the baseline schemes and the experiment with real-world data is carried out to evaluate the performance of our proposed basic incentive mechanism.

Extreme learning machine based on local receptive fields (ELM-LRFs) is a very fast method that can be used for feature extraction and classification. Bidirectional long-short time memory network (BLSTM), a widely used type of recurrent neural network (RNN) architecture, has showed excellent performance in time series processing fields. In this paper, we combine the superiority of above algorithms and propose a fast and accurate hybrid deep learning model which is named DELM-LRF-BLSTM for ECG signal recognition. This model uses the segmented heartbeats as input and employs a deep ELM-LRF to gain significant local spatial features. Then we fed the features to a three-layer BLSTM and it can extract temporal features for ECG signal recognition. The combination of ELM-LRF and BLSTM can not only consider the local information in a heartbeat, but also consider the long-distance dependence between heartbeats. Experimental results on MIT-BIH Arrhythmia dataset show that the proposed DELM-LRF-BLSTM algorithm has high accuracy and sensitivity, up to 99.32% and 97.15% respectively, which verifies the effectiveness and feasibility of the model. Moreover, only 6.1 millisecond is needed for once heartbeat recognition operation. Due to it's high performance and low computational complexity, the proposed algorithm is feasible for practical use.

Pipes are important components for well drilling, production, and transportation in petroleum industry. However, due to factors such as vibration, corrosion, fatigue damage, etc., pipes are prone to cracks. This paper presents a novel ACFM probe with flexible sensor array for the inspection of inner and outer cracks in pipes with different diameters. The simulation model of the ACFM probe with flexible sensor array is established by the finite element method. The influences of the crack length and the lift-off on the distance between the Bz peaks and troughs are analyzed respectively. The relationship between the offset distance of the crack from the pickup coil center and the detection sensitivity is studied. The probe made by flexible PCB and the testing system are developed. The experiments are carried out. The results of experiments and simulations show that the probe with flexible sensor array can detect inner and outer surface cracks for pipes with different diameters, and both circumferential and axial cracks can be identified. The distance between the Bz peaks and troughs can measure the crack length. When the crack is offset from the center of the pickup coil, the adjacent pickup coil can compensate for the detection sensitivity.

Modeling modular multilevel converter (MMC) for electromagnetic transients is challenging due to the massive number of switching devices. The fast and simultaneous switching events lead to high computational burden as the number of submodules in the MMC increases. Average value models (AVMs) can increase simulation speed by representing converter dynamics using decoupled AC-DC equivalent circuits. However, it is challenging for the state-of-the-art AVMs to represent the blocking operations of the MMC with the decoupled AC-DC equivalent circuits. Blocking modules with additional ideal switches are introduced to the AVMs to emulate the blocking behaviors of the MMC. However, the use of ideal switches bring additional modeling challenges, such as numerical oscillation due to current discontinuity and complicated circuit structures. This paper proposes a new AVM to deliver accurate simulation results with fast simulation speed. Moreover, based on a unified arm module representation, the proposed model can represent the MMCs with different submodule types with the identical circuit structure. The proposed model is validated using test systems based on a two-terminal HVDC system and 11-terminal CIGRE DC grid. It is shown that the proposed AVM provides faster simulation speed compared to the detailed equivalent model and better simulation accuracy than the state-of-the-art AVMs.

In this paper, a new method called the Virtual Cutting Method is proposed to evaluate the angularity index (AI) values of 3D point cloud coarse aggregate images with the aim of characterizing the angularity of aggregates on conveyor belts. The 3D point cloud images of coarse aggregates were first captured, preprocessed, and segmented into single 3D aggregate objects. Based on the processed 3D aggregate images, intersection contours were extracted using a series of intersection planes with an equivalent angle between two adjacent planes. The AI was evaluated by averaging the angularity of the contours using the gradient method, which was used in the AIMS2 system. Statistical analysis was then performed to select the optimum angle between two adjacent planes. It was found that an angle of five degrees was the ideal angle, as it can balance the execution time and effectiveness of the method. Finally, the AI results of the Virtual Cutting Method were compared with those of 2D and 3D Projection Methods. It was found that the AI rankings of the three methods for different aggregate textures are generally consistent. The findings of this study conclude that the Virtual Cutting Method can be employed to quantify the angularity of a single aggregate or aggregates in piles on conveyor belts based on 3D point cloud images.

Under the non-uniform transmission period, the satisfactory integrated design problem combining active fault-tolerant control and network communication saving was studied for nonlinear network control systems (NCSs) with actuator saturation constraints based on discrete event-triggered communication scheme (DETCS). First, the model of a non-uniform transmission closed-loop fault system was established, which includes the network attribute, actuator saturation constraints, estimation value of fault and state, event-triggered condition, disturbance information and so on. Second, the method of fault estimation with H∞-performance was proposed by a continuous time state observer. Furthermore, a satisfactory integrated design method integrating event-triggered fault accommodation and network communication saving was given for non-uniform transmission nonlinear NCSs with α-stability and H∞/H2 performance. Finally, a MATLAB simulation was conducted to verify the effectiveness and feasibility of the proposed method based on a nonlinear NCS example. In order to reflect the real network transmission circumstances fully, a semi entity experimental platform was also built to verify the effectiveness and feasibility of the proposed method.

As a new Non-destructive Evaluation (NDE) technique, capacitive imaging (CI) has been used to detect defects in oil pipelines, composite sucker rods and storage tanks in recent years. When a CI sensor is used to detect corrosion defect on a conducting surface with or without an insulation layer, the depth information of corrosion defect is difficult to acquire due to the non-linearity of the probing field. This paper proposed a corrosion depth inversion method based on the lift-off effect of the CI technique. The proposed depth inversion method, which includes three steps, namely establishing the lift-off curve, obtaining the scan curve and fitting inversion, are introduced and evaluated. In the FE simulations, the error rates of the inversion depths are less than 1.20%. CI experiments were carried out to acquire the depth information of a step shape specimen using the CI experimental system and the proposed corrosion depth inversion method. The results show that the depth inversion method can meet the need of obtaining the actual depth information with the error rates of the inversion depths being less than 10.00%. A CI experiment on a machined and corroded specimen was carried out to further demonstrate the feasibility of the proposed inversion method for corrosion defects with different depths. Both results from FE models and experiments indicate that the proposed method is promising to achieve the quantitative NDE of the Corrosion Under Insulation (CUI).

Currently, most consensus control approaches need each agent to access all its neighbors' states directly. The distributed consensus issue of second-order multi-agent systems subject to position constraints is studied without such requirements. The only condition for the communication topology is to include a directed spanning tree. An innovative reference position is provided to deal with the position constraints while eliminating the need for neighbors' velocity variables. An adaptive control method is designed by constructing a sliding-mode-esque variable so that each agent's transformed position can converge towards the reference position. This new method can guarantee uniform boundedness of each closed-loop signal as well as asymptotic consensus, and the requirements to meet the position constraints are satisfied at all times. Numerical simulation verifies the correctness of the theoretical results.