Search

Although deep learning methods based on single sensors are widely applied in fault diagnosis, leveraging multi-sensor data to learn useful information remains a challenge. To fully utilize multi-sensor information, this paper proposes a lightweight improvement of the EfficientNetV2 architecture, combined with sensor fusion technology and transfer learning techniques, to develop an efficient and reliable new method specifically for motor fault diagnosis. First, the continuous wavelet transform is utilized to convert the signals from various sensors into time-frequency images, and the Mallat algorithm is employed to decompose each image into sub-band coefficients at different levels. Secondly, a fusion reconstruction method is constructed using coefficient absolute maximum and weighted average fusion rules to integrate the sub-band coefficients of multi-sensor time-frequency images at different levels. Subsequently, EfficientNetV2 is improved to enhance the model’s feature extraction capabilities, computational efficiency, and achieve lightweight effects. The EfficientNetV2-M0 network modifies the model’s depth and width multiplicity factors, reducing parameters and computational complexity. Furthermore, this network incorporates Diverse Branch Block (DBB) and Multidimensional Collaborative Attention (MCA) to enhance feature extraction under complex backgrounds, and the maximum cross-entropy loss function is improved by using label smoothing and focal loss to dynamically adjust the classification weights for improved accuracy. The network leverages pre-trained models obtained through transfer learning techniques for deployment, combining multi-sensor information fusion and the improved lightweight model for fault diagnosis applications. Finally, a fault diagnosis experiment is conducted using a motor state dataset. The experimental results demonstrate that the proposed method outperforms the control method in terms of diagnostic performance and robustness, with an accuracy of 100%, and it exhibits excellent performance even under conditions of small sample data, with an accuracy of 98.81%.

In the actual working process of the motor, there will be short-term overload conditions, which puts forward higher requirements for maintaining the overload capacity of the motor. Due to the limitation of the thermal limit on the overload capacity of the motor, taking a 250kW permanent magnet synchronous motor (PMSM) as an example, this paper proposes a thermal management scheme of companion-type direct cooling, and establishes a three-dimensional thermal network model considering the secondary circulation factor. At the same time, in order to explore the influence of the thermal management scheme on the overload capacity of the motor when there is an impact load, on the premise that there is a safety margin for the hottest spot temperature of the winding under the rated state of the motor, the impact load is simulated by setting the overload torque, the thermal network model is used to compare and analyze the temperature field of the motor based on the companion-type direct cooling and the indirect casing cooling, and combining with the insulation limit to quantitatively compare the overload capacity of the motor after impact load. Through the simulation test, it is verified that the motor based on the companion-type direct cooling has efficient heat dissipation and stronger impact-resistance overload capability. The proposal of the companion-type direct cooling scheme has important practical significance for excavating the load potential of the motor under actual operating conditions.

The advancement of computing power has facilitated the development of three-dimensional point cloud neural networks, and the issue of adversarial samples in point clouds has garnered increasing attention. However, Current adversarial training methods often result in reduced classification effectiveness of the model. Additionally, the scarcity of point cloud data poses challenges in acquiring sufficient training samples. This paper proposes a novel algorithm for classifying 3D point cloud adversarial samples, employing a combination of self-supervision and information gain. The algorithm first utilizes a self-supervised learning model to extract potential information from the data. Then, it employs information gain to screen the features of the adversarial samples and concentrate on the beneficial classification information. The selected features are fed back to the classification model through joint training, aiming to further enhance classification accuracy. In comparison to the standard training approaches of Pointnet, Pointnet++, and Dgcnn on the Modelnet40 and ScanObjectNN datasets, the proposed algorithm exhibits significant improvements in classification effectiveness. Specifically, on the ModelNet40 dataset, the algorithm achieves improvements of 0.70, 0.88, and 0.07 in classification effectiveness over PointNet, PointNet++, and DGCNN, respectively. Similarly, on the ScanObjectNN dataset, the algorithm demonstrates improvements of 0.06, 0.54, and 1.72 in classification effectiveness for the three networks, respectively. These results demonstrate that the algorithm is capable of overcoming the limitations of traditional adversarial training, which tends to reduce the generalization performance of the model, and further enhances the classification accuracy.

Autonomous vehicle trajectory planning method based on the driving risk field is one of the current research hotspots. Currently, mainstream driving risk field models (DRFM) rarely consider basic factors such as vehicle geometry and heading angle, and ignore the impact of driving behavior characteristics on the driving risk field, the accuracy of safety assessment needs to be improved. Therefore, this paper proposes an extended driving risk field model (EDRFM) that considers geometric dimensions, heading angle, and driving behavior characteristics, and provides the basis for determining geometric dimensions and shape parameters, expanding the applicability of the DRFM. On this basis, EDRFM is applied to vehicle trajectory planning, and it is found that the target trajectory (denoted as P1) obtained using EDRFM is not achievable and the vehicle is unstable. This paper further proposes an arc correction method considering vehicle dynamics constraints to correct P1 to get P2. To ensure that the trajectory planning results meet the driving stability and motion of the vehicle. The results show that the EDRFM model considering arc correction can effectively complete vehicle trajectory planning tasks in complex scenes, and ensure the feasibility of trajectory planning and vehicle driving stability.

Road network control is challenging but critical in enhancing urban traffic. This paper proposes a management method for a road network with non-signalized intersections in the connected-vehicle environment, which coordinates connected and automated vehicles' movements to make the non-signalized intersections conflict-free and efficient. Firstly, the proposed method projects vehicles in road networks into virtual platoons and builds the road-network-wide conflict-free geometric topology considering the vehicles' conflicting relationships, which describes the geometry car-following relationship in virtual platoons. Then, a distributed linear controller is designed considering vehicle dynamics and communication topology to organize vehicles' movements with the desired geometric topology. Finally, simulations are conducted to verify the proposed method with different traffic demands. Simulation results show the proposed method can significantly improve traffic efficiency, as well as traffic safety.

In modern transportation, traffic congestion has become an urgent problem in large and medium-sized cities. In smart transportation systems, it is an effective solution to design load balancing path planning algorithms that can dynamically adapt to traffic conditions in order to avoid congestion. In this work, a traffic path planning algorithm based on data prediction (TPPDP) is proposed to find the path with the shortest travel time, which is built on a predictive model based on historical traffic data and current traffic information. Furthermore, a path planning algorithm based on data prediction with load balancing (TPPDP-LB) is also proposed, which combines the predicted information and the number of concurrent requests to achieve the path with shortest travel time while maintaining global load balancing. A specific distributed computing framework for TPPDP-LB algorithm is designed to reduce the runtime of the algorithm. The simulation results proved that both TPPDP and TPPDP-LB algorithms have the advantage of shortest travel time, and TPPDP-LB algorithm achieves load balancing of computing. It is also proved that the distributed computing framework designed for TPPDP-LP algorithm can effectively reduce the runtime of system as well as keep the accuracy of algorithm.

As a kind of indispensable equipment widely applied in construction sites, tower crane systems have complicated nonlinear dynamical characteristics, which make controller design challenges. Apart from the primary control objective of accurate slew/translation positioning and effective swing elimination, another important task is to ensure the satisfactory state transient performance, and simultaneously, the transportation time also needs to be as short as possible. To solve the above issues, in this paper, a new time-(sub)optimal trajectory planning method is proposed, which is the first solution to generate antiswing trajectories for four-degree of freedom (DOF) tower crane systems with state constraints. In particular, three auxiliary signals are constructed, and the reference trajectories for the jib slew and trolley translation can be obtained by the elaborately designed trajectories related to auxiliary signals. Hence, based on the reference trajectories for the jib/trolley, the payload can be driven to the desired location accurately with rapid payload swing elimination; moreover, the state variables, payload displacement, and their velocities are all restricted within the specific ranges. Finally, a series of hardware experiments are implemented to verify the effectiveness of the proposed method.

Current object detection techniques have difficulties in detecting small objects and a low level of accuracy in detecting occluded objects. To solve these problems, this paper proposes an object detection framework named FFAN which is based on Faster R-CNN that introduces a feature fusion network and an adversary occlusion network into the structure. The feature fusion network combines a feature map of low resolution and high semantic information with a feature map of high resolution and low semantic information using the deconvolution operation to increase the ability to extract low-level features in the network. FFAN then generates a single advanced feature map with high resolution and high semantic information which is used to predict the detection of small objects in the image more effectively. The adversary occlusion network creates occlusion on a deep feature map of the object, and generates an adversary training sample that is difficult for the detector to discriminate. At the same time, the detector classifies accurately the generated occluded adversary samples by self-learning. The two compete with and learn from each other to further improve the performance of the algorithm. We train FFAN on the PASCAL VOC 2007, PASCAL VOC 2012, MS COCO and KITTI datasets. A number of quantitative and qualitative experiments show that FFAN achieves a state-of-the-art detection accuracy.

Siamese network have been extensively applied in the tracking field because of its huge speed advantage and great precision performance in solving the tracking problems. In this paper, we propose an efficient framework for real-time object tracking which is end-to-end trained offline-Fully Conventional Anchor-Free Siamese network (FCAF). Specifically, as the backbone network in Siamese trackers is relatively shallow, resulting in insufficient feature information acquired by the trackers and lower accuracy, the deep network ResNet-50 is adopted to provide richer feature representation. Meanwhile, the introduction of multi-layer feature fusion module effectively combines low-level detail information with high-level semantic features, improving the localization performance. In addition, we propose the anchor-free proposal network (AFPN) to replace the region proposal network (RPN). AFPN network consists of correlation section, implemented by depth-wise cross correlation, and supervised section which has two branches, one for classification and the other for regression. In order to suppress the prediction of low quality bounding boxes, center-ness branch is added. We conduct extensive experiments on the OTB2015 and VOT2016 public datasets, demonstrating that our proposed tracker achieves state-of-the-art performance.

Multi-Terminal DC (MTDC) transmission systems typically operate under active power and DC voltage (P-UDC) droop control for active power dispatch, and its design is greatly influenced by DC network characteristics. Hence, this paper firstly analyzes P-UDC droop characteristics when implemented on Grid-Side Voltage-Source Converters (GSVSCs) of a MTDC network, in which variations of DC cables' resistances with temperature are taken into account. The analysis establishes that the actual droop characteristics of GSVSCs are nonlinear, different from the widely assumption of linear relationships. This finding is the fundamental for the design of the accurate power dispatch for GSVSCs. Therefore, an improved Fuzzy Search based Droop Control (FSDC) scheme for MTDC systems is proposed, which performs real-time search to capture the accurate droop curves of GSVSCs over full range of permissible DC voltage and active power. The FSDC is implemented without the assumption of the prior knowledge of DC network parameters which are subject to variations and uncertainty, and instead the accurate droop curves are periodically obtained and updated. The technical viability of the FSDC and its superiority over the conventional linear droop scheme under the variations of DC cable resistances and power flow variations is verified by a four-terminal HVDC model established in MATLAB/Simulink.

Multi-target tracking (MTT) is among the fundamental problems in the field of video analysis and monitoring. In the tracking-by-detection framework, data association is one of the most important and difficult problems. In this paper, we propose a framework to obtain the appearance features of a target in an end-to-end fashion, which fuses high-level and low-level semantic information. The high-order feature map is abstracted using the high-order apparent relationship for each target between the current frame and the previous frames, whereas the similarity matrix is used to describe the high-order features of the target. The best matching relationships between targets are obtained using hierarchical data association and the Hungarian algorithm. This proposed method is called Multi-target tracking Based on High-order Appearance Feature Fusion (MTT-HAFF), which can handle a large number of input sequences, local association failures, and identity exchanges that result from unreliable detections. The results show that the proposed algorithm has a good robustness for long-term occlusion tracking.

When the suspension rope of the crane system is relatively long and its mass cannot be simply ignored compared with that of the payload, or when the crane rope is underwater, it will exhibit flexible characteristics, that is, the bending deformation will occur during swing. Undoubtedly, unreasonable bending deformation of the crane rope will further excite larger-amplitude payload swing, which will seriously influence the efficiency and safety. To this end, for flexible rope crane systems, this paper carries out in-depth analysis and puts forward an effective control method. Specifically, by introducing the concept of lumped mass method and virtual spring, a nonlinear dynamic model is established. Furthermore, in order to solve the problem that velocity signals are not measurable, a nonlinear control strategy, which does not need velocity signals for feedback, is proposed in this paper. Finally, rigorous theoretical analysis and a series of experimental results verify the effectiveness of the proposed control strategy.

Feature selection has been a powerful tool to handle high-dimensional data. Most of these methods are biased toward the highest rank features which may be highly correlated with each other. In this paper, we address this problem proposing stratified feature ranking (SFR) method for supervised feature ranking of high-dimensional data. Given a dataset with class labels, we first propose a subspace feature clustering (SFC) to simultaneously identify feature clusters and the importance of each feature for each class. In the SFR method, the features in different feature clusters are separately ranked according to the subspace weight produced by SFC. After that, we propose a stratified feature weighting method for ranking the features such that the high rank features are both informative and diverse. We have conducted a series of experiments to verify the effectiveness and scalability of SFC for feature clustering. The proposed SFR method was compared with six feature selection methods on a set of high-dimensional datasets and the results show that SFR was superior to most of these feature selection methods.

This paper designs a robust discriminative tracking method based on correlation filter and key point matching, the highlight of which is the ability to adapt appearance changes. Toward the complicated and changeable real-world scene, a multi-channel of optimal selection correlation filter (MOSCF) is proposed, and it can model the holistic appearances by learning short-term dynamic features. Meantime, a key point feature set is maintained for representing the long-term stable appearances. In order to combine the similarity evaluation scores of two appearance features, a confidence fusion framework is established to obtain the final output confidence map. Particularly, partially occluded object is located by a kind of local key point voting mechanism. To alleviate error accumulation caused by inappropriate learning, an adaptive learning ratio is set to independently update the filter template corresponding to each channel of MOSCF. The results of the qualitative and quantitative experiments on the OTB-100 benchmark suggest that the proposed tracker outperforms several state-of-the-art methods.

Wireless local area network fingerprint-based indoor localization schemes have been widely studied because of the increasing requirements of location-based services (LBSs). The features of fingerprint based localization are known to have higher precision in indoor environments than traditional methods, such as triangulation. However, the precision depends on the amount of pre-created received signal strength (RSS) fingerprints, which is associated with the number of reference points (RPs) of the RSS measurements and the available signal sources in the environment. In this paper, we consider the resource limitations of todays' wireless environment and propose an improved fingerprint-based localization approach that adapts a path loss model for fingerprint creation and localization. Based on the proposed approach, we present two related localization schemes. The first is a path-loss-based fingerprint localization (PFL) scheme and the second is a dual-scanned fingerprint localization (DFL) scheme. The PFL attempts to improve positioning precision, and the DFL attempts to guarantee positioning reliability. Several simulations are performed, and they show that the proposed schemes improve the positioning precision and reliability in resource-limited environments, which would improve the practicability of fingerprint-based localizations in indoor LBSs.

As the smart cities emerged for more comfortable urban spaces, services, such as health, transportation, and so on, need to be promoted. In addition, the cloud computing provides flexible allocation, migration of services, and better security isolation; therefore, it is the infrastructure for the smart cities. Single instruction-set architecture (ISA) heterogeneous multi-core processors have higher performance per watt than their symmetric counterparts and are popular in current processors. In current cloud computing, which integrates a few fast out-of-order cores, coupled with a large number of simpler, slow cores, all cores expose the same ISA. The best way to leverage the effectiveness of these systems is to accelerate sequential CPU-bound threads using fast cores, and to improve the throughput of parallel memory-bound threads using slow cores. However, shared hardware resources, such as memory, respond to requests from all cores, which interfere with each other, leading to both low speed for fast cores and low throughput for slow cores. In this paper, we propose a dynamic resource partitioning (DRP) method for single-ISA heterogeneous multi-cores, which partitions the shared resources according to both threads' requirements for the shared resources and the performance of their running cores. The key principle is to profile both threads' resource characteristics at run-time and the performance of the cores that the threads are running on to estimate demands for resources. Then, we use the estimation to direct our resource partitioning. Moreover, we integrate our DRP with current memory scheduling policies to improve the system performance further for the two methods being orthogonal.

The traditional surgical navigation system combining preoperative CT and intraoperative ultrasound is widely used in open hepatectomy, but the problem of this system is that there are some errors in terms of the precision of the time and the space during the surgery. In order to solve the above problems, this paper introduces augmented reality technology into the surgical navigation system. In order to accurately describe the biomechanical characteristics of the liver and let the navigation system perform more accurately, this paper uses Tetgen to perform tetrahedral partition on the triangle mesh data obtained after the three-dimensional CT was reconstructed, and then the surface mesh data and the internal tetrahedral data were obtained and they were used to describe a whole liver model. The surface triangle mesh data is used to render graphics and describe the surface topology change, and the internal tetrahedral data combined with mass-spring theory is used to simulate deformation. Subsequently, the ex vivo pig liver was used to experimentally verify the accuracy of gravity deformation of the liver model, and the results show that the error was mainly distributed between −2mm and −2.5mm. At the same time, this paper uses the NDI Polaris infrared tracking system to carry out precision experiments on the augmented reality module, and the measured error is 1.55± 0.29mm. Finally, various modules of the system are integrated to finish the experiment in which the ring-shaped lesions are cut from the ex vivo pig liver with the aid of augmented reality. The experimental error is 0± 1.26mm, and with the assistance of the general purpose graphics processing unit (GPGPU), the refresh rate is above 200FPS. The results prove that the liver surgery navigation system proposed in this paper is excellent in terms of real-time performance and accuracy, which can help doctors accurately locate the tumor during surgery and perform ideal resection.