Most of the compound fault diagnosis methods regard the compound fault as a new single fault type, ignoring the interaction of internal single faults, and the fault analysis is vague in granularity and poor in interpretation. An improved Transformer-based compound fault decoupling diagnosis method was proposed for industrial environments with very little compound fault data. The diagnosis process included pre-processing, feature extraction and fault decoupling. With introducing the decoder of the Transformer, the cross-attention mechanism enables each single fault label to adaptively in the extracted feature layer focus on the discriminative feature region corresponding to the fault feature and predicts the output probability to achieve compound fault decoupling. Compound fault tests were designed to verify the effectiveness of the method compared with the advanced algorithms. The results showed that the proposed method had high diagnostic accuracy with a small number of single fault training samples and a very small number of compound fault training samples. The compound fault diagnosis accuracy reached 88.29% when the training set contained only 5 compound fault samples. Thus the new method has a significant advantage over other methods.

Most of existing structural image super-resolution reconstruction algorithms can only solve a specific single type of structural image super-resolution problem. A structural image super-resolution network based on improved Transformer (TransSRNet) was proposed. The network used the self-attention mechanism of Transformer mine a wide range of global information in spatial sequences. A spatial attention unit was built by using the hourglass block structure. The mapping relationship between the low-resolution space and the high-resolution space in the local area was concerned. The structured information in the image mapping process was extracted. The channel attention module was used to fuse the features of the self-attention module and the spatial attention module. The TransSRNet was evaluated on highly-structured CelebA, Helen, TCGA-ESCA and TCGA-COAD datasets. Results of evaluation showed that the TransSRNet model had a better overall performance compared with the super-resolution algorithms. With a upscale factor of 8, the PSNR of the face dataset and the medical image dataset could reach 28.726 and 26.392 dB respectively, and the SSIM could reach 0.844 and 0.881 respectively.

An adaptive salp swarm algorithm was proposed by minimizing the makespan in order to solve the flexible job shop scheduling problem with transportation time. A three-layer coding scheme was designed based on random key in order to make the discrete solution space continuous. The inertia weight was introduced to evaluate the influence among followers in order to enhance the global exploration and local search performance of the algorithm. An adaptive leader-follower population update strategy was proposed, and the number of leaders and followers was adjusted by the population status. The tabu search strategy was combined with the neighborhood search in order to prevent the algorithm from falling into local optimum. The benchmark instances verified the effectiveness and superiority of the proposed algorithm. The influence of the number of AGVs on the makespan conforms to the law of diminishing marginal effect.

For the poor real-time detection capability of the current object detection model in the production environment of electronic components, GhostNet was used to replace the backbone network of YOLOv5. And for the existence of small objects and objects with large scale changes on the surface defects of electronic components, a coordinate attention module was added to the YOLOv5 backbone network, which enhanced the sensory field while avoiding the consumption of large computational resources. The coordinate information was embedded into the channel attention to improve the object localization of the model. The feature pyramid networks (FPN) structure in the YOLOv5 feature fusion module was replaced with a weighted bi-directional feature pyramid network structure, to enhance the fusion capability of multi-scale weighted features. Experimental results on the self-made defective electronic component dataset showed that the improved GCB-YOLOv5 model achieved an average accuracy of 93% and an average detection time of 33.2 ms, which improved the average accuracy by 15.0% and the average time by 7 ms compared with the original YOLOv5 model. And the improved model can meet the requirements of both accuracy and speed of electronic component surface defect detection.

A ship object detection algorithm was proposed based on a multi-head self-attention (MHSA) mechanism and YOLO network (MHSA-YOLO), aiming at the characteristics of complex backgrounds, large differences in scale between classes and many small objects in inland rivers and ports. In the feature extraction process, a parallel self-attention residual module (PARM) based on MHSA was designed to weaken the interference of complex background information and strengthen the feature information of the ship objects. In the feature fusion process, a simplified two-way feature pyramid was developed so as to strengthen the feature fusion and representation ability. Experimental results on the Seaships dataset showed that the MHSA-YOLO method had a better learning ability, achieved 97.59% mean average precision in the aspect of object detection and was more effective compared with the state-of-the-art object detection methods. Experimental results based on a self-made dataset showed that MHSA-YOLO had strong generalization.

A news recommendation method based on Transformer and knowledge graph was proposed to increase the auxiliary information and improve the prediction accuracy. The self-attention mechanism was used to obtain the connection between news words and news entities in order to combine news semantic information and entity information. The additive attention mechanism was employed to capture the influence of words and entities on news representation. Transformer was introduced to pick up the correlation information between clicked news of user and capture the change of user interest over time by considering the time-series characteristics of user preference for news. High-order structural information in knowledge graphs was used to fuse adjacent entities of the candidate news and enhance the integrity of the information contained in the candidate news embedding vector. The comparison experiments with five typical recommendation methods on two versions of the MIND news dataset show that the introduction of attention mechanism, Transformer and knowledge graph can improve the performance of the algorithm on news recommendation.

The improved YOLOv5 object detection algorithm was used to detect the facial region of the driver and a multi-feature fusion fatigue state detection method was established aiming at the problem that existing fatigue state detection method cannot be applied to drivers under the epidemic prevention and control. The image tag data including the situation of wearing a mask and the situation without wearing a mask were established according to the characteristics of bus driving. The detection accuracy of eyes, mouth and face regions was improved by increasing the feature sampling times of YOLOv5 model. The BiFPN network structure was used to retain multi-scale feature information, which makes the prediction network more sensitive to targets of different sizes and improves the detection ability of the overall model. A parameter compensation mechanism was proposed combined with face keypoint algorithm in order to improve the accuracy of blink and yawn frame number. A variety of fatigue parameters were fused and normalized to conduct fatigue classification. The results of the public dataset NTHU and the self-made dataset show that the proposed method can recognize the blink and yawn of drivers both with and without masks, and can accurately judge the fatigue state of drivers.

An AT-YOLO algorithm based on improved YOLOv3 was proposed aiming at the problem that the existing deep learning-based defect detection algorithm for printed circuit boards (PCB) could not meet the accuracy and efficiency requirements at the same time. Feature extraction capabilities were improved and the number of parameters was reduced by replacing the backbone with ResNeSt50. SPP module was added to integrate the features of different receptive fields and enrich the ability of feature representation. The PANet structure was improved to replace FPN, and the SE module was inserted to enhance the expression capability of effective feature maps. A set of high-resolution feature maps were added to the input and output in order to improve the sensitivity to small target objects, and the detection scale was increased from three to four. K-means algorithm was re-used to generate sizes of anchors in order to improve the accuracy of object detection. The experimental results showed that the AT-YOLO algorithm had an AP_0.5 value of 98.42%, the number of parameters was 3.523×10⁷, and the average detection speed was 36 frame per second on the PCB defect detection dataset, which met the requirements of accuracy and efficiency.

There has been a trend of intelligent development using artificial intelligence technology in order to improve the efficiency of software development. It is important to understand program semantics to support intelligent development. A series of research work on program representation learning has emerged to solve the problem. Program representation learning can automatically learn useful features from programs and represent the features as low-dimensional dense vectors in order to efficiently extract program semantic and apply it to corresponding downstream tasks. A comprehensive review to categorize and analyze existing research work of program representation learning was provided. The mainstream models for program representation learning were introduced, including the frameworks based on graph structure and token sequence. Then the applications of program representation learning technology in defect detection, defect localization, code completion and other tasks were described. The common toolsets and benchmarks for program representation learning were summarized. The challenges for program representation learning in the future were analyzed.

A trajectory planning method based on improved particle swarm optimization (PSO) algorithm was proposed in order to improve the working efficiency of tunnel boring machine (TBM) disc cutter changing robot and reduce the motion jerk in the process of tool changing. The kinematics of the redundant joint robot was analyzed by using the position and pose separation method and the joint variable minimization strategy. The target trajectory was mapped from Cartesian space to joint space by using the inverse solution. The jerk continuous joint trajectory was constructed by using 5-degree NURBS curve for each joint. The objective function was constructed by the time jerk optimization, and the optimal time series was solved by using the improved PSO algorithm so as to complete the trajectory optimization. The trajectory planning of specific disc cutter change task was conducted, and the optimal trajectory of each joint was obtained. The optimization results show that the proposed trajectory planning method can provide an ideal trajectory for each joint of the tool changing robot and has strong trajectory tracking ability. The 5th NURBS interpolation method and the improved PSO optimization algorithm were used to ensure the shortest time and minimum impact of the trajectory, and improve the efficiency and stability of the operation.

A semantic segmentation network was designed for underwater images. A lightweight and efficient encoder-decoder architecture was used by considering the trade-off between speed and accuracy. Inverted bottleneck layer and pyramid pooling module were designed in the encoder part to efficiently extract features. Feature fusion module was constructed in the decoder part in order to fuse multi-level features, which improved the segmentation accuracy. Auxiliary edge loss function was used to train the network better aiming at the problem of fuzzy edges of underwater images, and the edges of segmentation were refined through the supervision of semantic boundaries. The experimental data on the underwater semantic segmentation dataset SUIM show that the network achieves 53.55% mean IoU with an inference speed of 258.94 frames per second on one NVIDIA GeForce GTX 1080 Ti card for the input image of pixel 320×256, which can achieve real-time processing speed while maintaining high accuracy.

An improved ORB-SLAM algorithm based on motion prediction was proposed by considering the influence of the camera’s own motion on the visual SLAM system aiming at the problem that the ORB-SLAM algorithm with fixed point feature extraction and matching strategy has large tracking and positioning error in different motion scenes. The point feature utilization rate of the previous frame and the uniform motion model were used to predict the mutually visual zone between two adjacent frames. The threshold of point feature extraction under different motion states was dynamically adjusted in real time. Then the accuracy of the system was improved while ensuring the stability of the system. A point feature matching optimization strategy based on motion prediction was proposed. The effective matching points within the mutually visual zone were quickly determined based on the uniform motion model. The matching search range was narrowed by combining the image pyramid in order to reduce many invalid matching processes. The comparison experiments were conducted on the TUM data set. Results show that the proposed algorithm not only has good real-time performance, but also improves the accuracy of the system.

A new 3D reconstruction network was proposed in order to resolve the difficulty of 2D detection method to detect defects with depth information. CasMVSNet with multiscale feature enhancement (MFE-CasMVSNet) was combined with the technology of point cloud processing for steel plate surface defect detection. In order to improve the accuracy of 3D reconstruction, a position-oriented feature enhancement module (PFEM) and a multiscale feature adaptive fusion module (MFAFM) were proposed to effectively extract features and reduce information loss. A density clustering method, curvature-sparse-guided density-based spatial clustering of applications with noise (CS-DBSCAN), was proposed for accurately extracting defects in different parts, and the 3D detection box was introduced to locate and visualize defects. Experimental results show that compared with the reconstruction method based on images, MFE-CasMVSNet can realize the 3D reconstruction of steel plate surface more accurately and quickly. Compared with 2D detection, 3D visual defect detection can accurately obtain the 3D shape information of defects and realize the multi-dimensional detection of steel plate surface defects.

To reliably and accurately predict the axial bearing capacity of concrete-filled steel tube (CFST) columns, a prediction model of CFST column axial bearing capacity with ensemble machine learning was developed and explained. The quality of the CFST column database was evaluated using the Mahalanobis distance, the prediction model of CFST column axial bearing capacity was established by the extreme gradient boosting (XGBoost) algorithm, and the optimal hyperparameter combination of the model was found using the K-Fold cross-validation (K-Fold CV) and the tree-structured Parzen estimator (TPE) algorithms. The predicted values of the optimized XGBoost model were compared with the calculated values of the existing methods and the unoptimized XGBoost model using different evaluation metrics. The Shapley additive explanations (SHAP) approach was used to produce both global and local explanations for the predictions of XGBoost model. Results show that, after hyperparameter tuning, the XGBoost model’s performance surpasses performance of relevant standards and empirical formulas, and the SHAP approach can effectively explain the XGBoost model’s output.

A point cloud instance segmentation model with a k-nearest neighbors (KNN) module featuring attention mechanism and an improved associatively segmenting instances and semantics (ASIS) module was proposed to address the problems of discrete segmentation and insufficient feature utilization in traditional 3D convolution-based algorithms. The model took voxels as input and extracted point features through sparse convolution of 3D submanifolds. The KNN algorithm with attention mechanism was used for reorganizing the features in the semantic and instance feature space to alleviate the problem caused by the quantization error of extracted features. The reorganized semantic and instance features were correlated through the improved ASIS module to enhance the discrimination between point features. For semantic features and instance embedding, the softmax module and the meanshift algorithm were applied to obtain semantic and instance segmentation results respectively. The public S3DIS dataset was employed to validate the proposed model. The experimental results showed that the instance segmentation results of the proposed model achieved 53.1%, 57.1%, 65.2% and 52.8% in terms of mean coverage (mCoV), mean weighted coverage (mWCov), mean precision (mPrec) and mean recall (mRec) for the instance segmentation. The semantic segmentation achieved 61.7% and 88.1% respectively in terms of mean intersection over union (mIoU) and Over-all accuracy (Oacc) for the semantic segmentation. The ablation experiment verified the effectiveness of the proposed modules.

A hierarchical federated learning framework based on wireless device-to-device (D2D) networks was proposed to solve the problem of large communication resource consumption and limited device computing resources faced by deploying federated learning in wireless networks. Different from the traditional architectures, the hierarchical aggregation was adopted for model training. The architecture performed the intra-cluster aggregation through D2D networks, and each cluster performed the decentralized training at the same time. A cluster head was selected from each cluster to upload the model to the server for global aggregation. The network traffic of the central node was reduced by combining the hierarchical federated learning and decentralized learning. The degree of the vertices in the D2D networks was used to measure the model convergence performance. The head selection and bandwidth allocation were jointly optimized by maximizing the total degree of selected cluster heads. An optimization algorithm based on dynamic programming was designed to obtain the optimal solutions. The simulation results show that compared with the baseline algorithm,the framework can not only effectively reduce the frequency of global aggregation and training time, but also improve the performance of the final model.

The online monitoring dynamic inner projection to latent structures (OM-DiPLS) model was proposed aiming at the problems of false alarm and missing quality data of some periods in actual industrial process fault detection. The model can be updated when quality data of some periods were missing by introducing quality data of time delay. An online monitoring dynamic internal concurrent latent structure projection (OM-DiCPLS) model was proposed based on the OM-DiPLS model in order to better monitor the unpredictable information in the quality variables. The process data and quality data were projected into the covariant subspace with relevant inputs and outputs, the input principal subspace with unrelated outputs but relevant processes, the input residual subspace, the unpredictable output principal subspace and the output residual subspace. Process monitoring was realized by constructing corresponding statistics for each subspace. The Tennessee-Eastman process simulation experiment shows that the proposed algorithm effectively improves the effective detection rate of quality-related faults and reduces the false alarm rate of quality-unrelated faults.

A multi branch Siamese network target tracking algorithm based on dual attention mechanism was proposed in order to solve the problem of inaccurate positioning in the SiamRPN++ single target tracking algorithm when the target was briefly occluded and the appearance drastically changed. SiamRPN++ with lightweight backbone network was adopted as the basic algorithm. The algorithm was combined with lightweight channel and spatial attention mechanism in order to improve the anti-interference ability when dealing with occlusion challenges during the tracking process. A template branch was added from the previous frame, and the appearance changes of the target were dynamically updated. The ability to distinguish between foreground and background was enhanced during the tracking process using triplet loss. Local expansion search was conducted based on the speed of the target’s movement in order to enable timely and accurate tracking of the target even after short-term occlusion. The experimental results showed that the improved algorithm improved the success rate and precision of the OTB100 dataset by 2.4% and 1.6%, respectively, compared to the original algorithm. The average center position error decreased by 28.97 pixels, and the average overlap rate increased by 14.5%.

In reality, the structure of most graphs could be noisy, i.e., including some noisy edges or ignoring some edges that exist between nodes in practice. To solve these challenges, a novel differentiable similarity module (DSM), which boosted node representations by digging implict association between nodes to improve the accuracy of node classification, was presented. Basic representation of each target node was learnt by DSM using an ordinary graph neural network (GNN), similar node sets were selected in terms of node representation similarity and the basic representation of the similar nodes was integrated to boost the target node’s representation. Mathematically, DSM is differentiable, so it is possible to combine DSM as plug-in with arbitrary GNNs and train them in an end-to-end fashion. DSM enables to exploit the implicit edges between nodes and make the learned representations more robust and discriminative. Experiments were conducted on several public node classification datasets. Results demonstrated that with GNNs equipped with DSM, the classification accuracy can be significantly improved, for example, GAT-DSM outperformed GAT by significant margins of 2.9% on Cora and 3.5% on Citeseer.

The loading state of integrated cutter holder system was analyzed by combining numerical simulation with experimental research, in order to improve the bearing performance of integrated cutter holder system. Combined with the automatic assembly process of cutters, the linkage relationship between each structural parameter in integrated cutter holder system and the influence of different structural parameters on its bearing performance were studied. The structural parameters significantly affecting the bearing performance of integrated cutter holder system were optimized based on the weight matrix method of orthogonal test. Results showed that the influence degree of each test factor on the bearing performance of integrated cutter holder system in descending order was as follows: the neck fillet radius of rotating block, the width of rotating block, the vertical distance between cutter shaft and rotating block shaft. The optimal scheme for comprehensive performance of integrated cutter holder system was obtained as follows: the width of rotating block was 107.5 mm, the neck fillet radius of rotating block was 60.0 mm, and the vertical distance between cutter shaft and rotating block shaft was 97.5 mm. Compared with the original scheme of integrated cutter holder system, the overall maximum deformation was reduced by 11.31%, the maximum stress of end cap was reduced by 34.07%, and the maximum stress of rotating block was reduced by 41.01%.