Annular parts are widely utilized in high-end manufacturing, which are often manufactured by ring forging to meet the requirement on mechanical properties. However, residual stress (RS) is inevitably introduced in the forging process, which affects the working performance and reduces the service life of the part. Therefore, to counter the determent of RS, it is first necessary to accurately obtain it. Different from sheet or structural parts, the RS distribution in an annular part is more complex. Existing methods based on RS measurement and finite element prediction did not incorporate the RS prior, so the accuracy is influenced by the simplified assumption on RS distribution. This paper proposes a new RS inferencing method for annular parts based on the use of deformation force and physical prior. Firstly, the RS distribution prior of the annular blank after heat treatment is obtained by finite element simulation and the clustering algorithm. Then, the prior is employed to calculate the coefficient matrix to reflect the relationship between the deformation force and the RS. Finally, the Tikhonov regularization method is used to solve the deformation force-RS inverse problem. The effectiveness and feasibility of the proposed method are verified in the simulation and experiment, respectively, which shows that the integration of RS prior distribution is able to reduce the ill-condition of the solving process to obtain an accurate solution of RS distribution in an annular part.
The society is now in the data-rich environment, and deep learning is widely used in bearing fault diagnostic technology due to the advancement of information technology. These methods typically need a large amount of data to support. However, in some practical cases, only few of samples are frequently available when a fault occurs rather than adequate data to be analyzed. This situation indicates that bearing fault diagnostic problems are frequently fewshot problems. In this work, a generic meta-transfer learning model with special neuronal processing parameters (MSNPP) is proposed. MSNPP avoids the issue of overfitting commonly encountered in traditional meta-learning approaches when solving few-shot problems and maintains excellent performance when extracting features with deep networks. Moreover, MSNPP discovers the connection between different tasks by analyzing a few samples and quickly adapts to new tasks. In MSNPP, a technique known as neuron transfer (NT) is used to manipulate neurons by scaling and shifting them. The scaling and shifting parameters are used as meta-learning hyperparameters to transfer within different tasks, which is the work of NT. Experimental result shows that MSNPP prevents the issue of overfitting in conventional meta-learning approaches and achieves satisfactory accuracy and robustness when solving few-shot problems in fault diagnosis.
Dimensional accuracy is one of the most important quality indicators of large shaft grinding, which directly affects the shaft service performance. Overcut/undercut caused by grinding wheel width and wear are the main factors affecting dimensional accuracy of large shaft grinding, which can be described by establishing physical models, and the physical models would be further used for compensation. However, the residual error always exists due to modeling uncertainty, and the residual error has nonlinear relation with compensation value, which is hard to completely eliminate. In order to solve the problem, this paper proposes a dimensional accuracy compensation method of large shaft grinding via residual error iteration with fuzzy approach. Two physical models are firstly established by considering the grinding wheel width and wear, respectively. The residual error after using these two models is further dealt with iteration, and the fuzzy approach is applied to dynamically calculate the compensation coefficients to improve dimensional accuracy while ensuring con-vergence. The experimental results show that the mean dimensional error is reduced 83% by using the pro-posed method, which is much better than other compensation methods.
Advanced virtual manufacturing systems refer to highly sophisticated computer-based systems which simulate and optimize manufacturing processes in a virtual environment. Recently, many researchers have presented research works in different areas of simulation and analysis of manufacturing process in virtual environments such as cloud manufacturing, virtual training systems, virtual inspection systems, virtual process planning, flexible manufacturing systems, virtual manufacturing networks, virtual monitoring systems, virtual manufacturing for optimized production process, virtual machining systems and virtual commissioning systems. The advantages of virtual manufacturing systems include improving the quality of the produced components, decreasing the quantity of waste materials and accelerate product and process design using virtual simulation and modification. As a result, accuracy as well as efficiency in process of part manufacturing can be increased by applying the virtual environments to manufacturing operations. Moreover, digital marketing by using virtual manufacturing systems can increase the added value in the process of part production. To analyze and modify the processes of part production, recent achievements in virtual manufacturing systems are reviewed and presented in the study. The applications of virtual manufacturing systems in creating manufacturing processes are discussed, and future research works are also proposed. It has been discovered that reviewing and evaluating recent achievements in the published papers can promote the process of manufacturing engineering using virtual simulation and modification.
The detection of foreign objects in nuclear power plant reactor is a key task in the operation and maintenance of nuclear power plants. Loose and falling foreign objects such as bolts can lead to fuel component damage and unplanned shutdown, posing serious hazards. Therefore, we propose a point cloud semantic segmentation method for foreign objects in nuclear power plant reactor based on the RandLA-Net model. Considering the correlation between point cloud collection error and curvature, the data augmentation method is improved to reduce the risk of model overfitting. By treating the boundary points of different classes as the hard examples, the hard example mining is designed to improve model generalization performance. Adding an improved test time augmentation method during model inference, the more reliable segmentation results are performed by multiple prediction on points. The experimental results indicate that the proposed method can achieve high-accuracy reactor point cloud semantic segmentation with mIoU of 0.992 and mAcc of 0.997.