The surface quality and accuracy of the workpiece after sub-aperture polishing are heavily dependent on the accuracy of the dwell time distribution and the size of the tool influence function (TIF). Therefore, there is a high requirement for the dynamic performance of the machine tool to achieve dwell time. This is an inherent problem of computer controlled optical surfacing (CCOS). The additional material removal layer introduced to avoid the negative value of the dwell time or the limitation of the dynamic performance of the machine tool will greatly reduce the processing efficiency and accuracy. In this study, an actively controllable time-varying tool influence function (CTV-TIF) processing method is proposed. The time constraint is liberated by actively controlling the spatial dimension of TIF. By establishing a global optimization algorithm for the dwell time distribution of the adaptive TIF, the deterministic machining process with high figuring ability and not constrained by the dynamic performance of the machine tool can be realized. Compared with the time-invariant tool influence function (TITIF) processing, the efficiency of the CTV-TIF method is increased by 65.9 %, and the residual surface shape error after processing is increased by 96.7 %, and the high gradient surface shape error distribution is better suppressed. This method can be perfectly combined with rotating water jet polishing to make the TIF time-varying. The experimental results show that the processing method of CTV-TIF can significantly improve the figuring efficiency of low frequency error, and has better figuring ability for the position where the gradient of surface shape error changes rapidly, so as to obtain higher precision surface shape. In addition, the proposed method can be applied to other sub-aperture polishing processes that are expected to achieve time-varying TIF and improve the existing sub-aperture polishing technology.

Laser additive manufacturing (AM) technology has become an important method for the manufacturing of high-performance aluminum alloy parts. However, the thermal effect of the molten pool and the defect formation mechanism are still the key issues restricting forming quality. To address this issue, this paper systematically investigates the effects of key parameters such as laser power and pulse frequency on the thermal conductivity, kinetic behavior, and defect control of the molten pool through multi-physics coupled numerical simulation to provide theoretical support for improving the quality of components. It is found that the laser power and pulse frequency play a key role in the molten pool morphology and defect generation, with too low a power leading to non-fusion and too high a power triggering overheating and cracking, and too low a frequency leading to unstable morphology and too high a frequency triggering grain coarsening and thermal stress cracking. The optimized process parameters (power 700–800 W, frequency 72–100 KHz) effectively improved the melt pool morphology and reduced the defects. This study reveals the intrinsic mechanism of melt pool dynamics and defect formation, which provides important instructions for optimizing the aluminum alloy additive manufacturing process.

During the product development process, traditional reliability qualification testing (RQT) is typically used to assess whether the product meets predefined reliability standards. However, this method often requires a large sample size and extended testing time, rendering it unsuitable for expensive high-reliability products with limited sample sizes. To address these challenges, this study leverages expert judgment to determine the prior distribution of failure rates, enabling the formulation of an RQT plan for Weibull cases based on posterior risks. Finally, the study illustrates the advantages and applications of this approach using data from certain aerospace products as an example.

In recent years, large-scale pre-trained models have developed rapidly in general fields, but there are few practically viable technical solutions for application scenarios in specialized fields, especially in sensitive areas such as healthcare, finance, and even military domains that require precise, professional, and credible knowledge for decision-making assistance. Concentrating data on cloud servers and then efficiently fine-tuning large-scale pre-trained models is impractical for industries with high data privacy protection requirements. This paper proposes a distributed training strategy for large-scale pre-trained models that integrates federated learning strategies. First, local data is fine-tuned offline on various edge devices. Then, the parameters of the efficiently fine-tuned small-scale models (e.g., LORA) are uploaded for aggregation on the server. Finally, intelligent services and solutions are provided for complex scenarios in specialized fields.

Psychiatric diseases are bringing heavy burdens for both individual health and social stability. The accurate and timely diagnosis of the diseases is essential for effective treatment and intervention. Thanks to the rapid development of brain imaging technology and machine learning algorithms, diagnostic classification of psychiatric diseases can be achieved based on brain images. However, due to divergences in scanning machines or parameters, the generalization capability of diagnostic classification models has always been an issue. We propose Meta-learning with Meta batch normalization and Distance Constraint (M 2 DC) for training diagnostic classification models. The framework can simulate the train-test domain shift situation and promote intra-class cohesion, as well as inter-class separation, which can lead to clearer classification margins and more generalizable models. To better encode dynamic brain graphs, we propose a concatenated spatiotemporal attention graph isomorphism network (CSTAGIN) as the backbone. The network is trained for the diagnostic classification of major depressive disorder (MDD) based on multi-site brain graphs. Extensive experiments on brain images from over 3261 subjects show that models trained by M 2 DC achieve the best performance on cross-site diagnostic classification tasks compared to various contemporary domain generalization methods and SOTA studies. The proposed M 2 DC is by far the first framework for multi-source closed-set domain generalizable training of diagnostic classification models for MDD and the trained models can be applied to reliable auxiliary diagnosis on novel data.

During the development of systems, traditional System Reliability Qualification Testing (SRQT) is typically utilized to assess whether they meet predefined reliability standards. However, this approach often demands substantial sample sizes and prolonged test durations, rendering it impractical for costly, highly reliable systems with limited sample sizes. Additionally, the extended test duration may not align with practical time-to-market pressures or budget constraints. To overcome these challenges, the study integrates subsystem data into the monopropellant engine system RQT plan's design. By leveraging Monte-Carlo simulation, subsystem data is modeled to create a system parameter distribution, enabling the formulation of SRQT plans based on posterior risks. An example of a monopropellant liquid rocket engine system is provided to demonstrate the advantages and applications of the proposed methodology.

In order to address the issues pertaining to the subjective nature and limited precision associated with selecting feature points in the point cloud of a large steel truss structure, this study proposes a batch automatic extraction approach for identifying key feature information, including boundaries, corner points, and bolt holes of large steel truss components. This method relies on the nested application of established processing algorithms such as Euclidean clusters, regional growth clusters, and random sampling consensus. In addition, a novel approach is suggested for validating the precision of feature information extraction through the utilization of standard theoretical models. The results of the experimental and large-scale lower chord tests demonstrate that our approach is not dependent on specialized software, exhibits excellent efficiency, and possesses an acceptable degree of automation. The findings of this study can provide accurate data support for reverse modeling, virtual trial assembly, and dimensional inspection of steel truss components.

Previous studies have explored the use of deep neural networks for electroencephalography (EEG)-based motor imagery (MI) recognition, but most of the models focus on the recognition performance achieved for a single subject and are challenging to transfer due to individual differences and low signal-to-noise-ratio of EEG signals. To date, few studies have paid attention to the balance between generalizability and personalization across subjects. To this end, we propose a co-teaching graph learning method for cross-subject EEG-based MI recognition. First, A novel graph learning approach is designed to improve feature extraction from a typical graph structure containing raw EEG signals. Second, two graph learning models are constructed to filter noisy data by using a co-teaching training strategy, preventing overfitting on noisy samples obtained from different subjects. The proposed model shows a 5.4% and 3.2% increase in accuracy of single- and multisubject four-class MI recognition tasks compared to the previous best method, respectively. Experimental results also demonstrate that it is easy to derive a model that can represent generic knowledge of multiple MI subjects and can be fine-tuned efficiently for new subjects.

With the development of internet technology, the number of illicit websites such as gambling and pornography has dramatically increased, posing serious threats to people’s physical and mental health, as well as their financial security. Currently, the governance of such illicit websites mainly focuses on limited-scale detection through manual annotation. However, the need for effective solutions to govern illicit websites is urgent, requiring the ability to rapidly acquire large volumes of existing website data from the internet. Web mapping engines can provide massive, near real-time web data, which plays a crucial role in batch detection of illicit websites. Therefore, in this paper, we propose a method that combines web mapping engine big data to perform unsupervised multimodal clustering (MDC) for illicit website discovery. By extracting features based on contrastive learning methods from webpage screenshots and OCR text, we conduct feature similarity clustering to identify illicit websites. Finally, our unsupervised clustering model achieved an overall accuracy of 84.1% on all confidence levels, and a 92.39% accuracy at a confidence level of 0.999 or higher. By applying the MDC model to 3.7 million real web mapping data, we obtained 397,275 illicit websites primarily focused on gambling and pornography, with 14 attributes. This dataset is made publicly.

A reliability demonstration test (RDT) has been used to determine whether a product meets pre-specified reliability requirements and to decide whether a batch of products should be accepted or rejected. For systems with high reliability, standard RDTs are no longer preferred because test plans often require long test durations and pose high risks to producers and consumers. Related reliability information, such as subsystem test data, which is often available prior to system testing, is often neglected. For the widely applied assumptions of exponentially distributed subsystems and systems, this paper proposes a reliability demonstration test plan derivation method that makes use of subsystem test data to derive the probability density function of system failure rate for small-sized sample data. In comparison with conventional RDT plans that make decisions solely based on system test data, a system test plan can be derived with much shorter test durations while keeping producer and consumer risks under control. The case study shows that our proposed method can reduce risk and shorten test duration when a test plan needs to be derived for a system with subsystem test data.

The Bayes plan of reliability qualification testing (RQT) for Weibull aero-space equipment is aiming to determine the qualification testing plan of equipment in the Weibull distribution. By prespecifying the shape parameter, the exponential distribution is a special case of the Weibull distribution. To handle these particularities, the RQT plan for aerospace equipment is determined by referring to the method under exponential. Under the prior in-formation of reliability index from similar equipment, the Bayes method and maximum risk criterion is adopted. It is well established the RQT plan of equipments with no failure and small simple size. Different methods to determine the hyperparameters are proposed and their application field also is studied. An illustration is provided by a case study on an aerospace equipment. It makes the proposed method especially suitable for Weibull aerospace with high reliability.

A reliability qualification test (RQT) is used to judge whether a batch of products meets pre-specified reliability requirements. For high reliable systems, traditional RQT plans often require long test time or pose high risks to both producer and consumer. To cope with this problem, this paper proposes a new method to derive RQT plans by making use of subsystem data, when both system and subsystems are following exponential distributions. Compared with conventional RQT plans that construct system test based on which decisions are made, the proposed method enables deriving system test plans with much shorter test time and keeping producer and consumer risks under control at the same time. A case study is presented to prove the validity of the proposed method.

Grid remapping is an important mechanism for data transfer between different grids in the Earth modeling system, and searching for relevant points is one of the most complex and time-consuming operations. At present, there are few efficient and robust searching algorithms for unstructured grids. However, with the rise of unstructured grid based model and the improvement of model resolution, providing an efficient search algorithm for unstructured grid is in urgent need in engineering practice. In order to improve the performance of searching process between unstructured grid and other grids when remapping, this paper implements two search algorithms based on KD (K-dimension) tree (the nearest neighbor search and range search). Compared with the performance of brute force search, it is found that the efficiency of search algorithms based on KD tree are much higher than that of brute force search, including the nearest neighbor search and range search. The experimental results provide practical experience for the development of remapping software in the future.

Lithium-ion battery is the main energy source widely used in many fields. Therefore, it is particularly essential for estimating the health of lithium-ion battery accurately, especially in important fields such as aerospace, rail transit and satellite. For lithium-ion battery, the battery capacity is a health index (HI) that best reflects its performance degradation. By estimating the battery capacity, the health status of the lithium-ion battery can be clearly identified. However, there are technical barriers to the direct measurement of battery capacity in engineering, and many characteristics and capacities of lithium-ion batteries have abrupt changes, so that it is difficult to calculate the battery capacity accurately by formula calculation. In this paper, a new method of genetic programming combined model is proposed, which can calculate the capacity of lithium-ion battery by formulating multiple monitored features with a certain precision. Therefore, the functional relationship between multiple features and HI is well measured, which lays a good foundation for the subsequent life prediction of battery.

To deal with the sensor scheduling in continual object tracking of low earth orbit constellation,a method based on optimal search of information decision tree is proposed.The myopic information increment is extended to non-myopic one.The information decision tree is established and the pruning technology is utilized to search the best branch.The simulation results indicate that the excessive scheduling and lower tracking precision of myopic scheduling have been reduced effectively;The introduction of pruning technology has reduced the computation burden greatly.

There are a large number of indirect schema mappings between peers in the network. To improve the efficiency of data exchange and queries, indirect mappings are needed to be composed. Defined the combination operations of schema elements in indirect mappings, and gave the expression of indirect mappings. Proposed a strategy, named schema element back, to solve the problem of indirect mapping composition, and gave the indirect mapping composition generation algorithm based on such strategy. Experiments showed that indirect mapping composition can improve the efficiency of data exchange, and compared with other non-full mapping composition generation algorithms, and indirect mapping composition generated by our algorithm based on schema element back strategy can completely eliminate the infection of media schema with no reduction of the composition efficiency.

To improve the efficiency of data exchange between data sources,and reduce the redundancy and errors,a data model named extension hyper graph data model as the common data model for life cycle information integration of equipments was proposed.Mechanism of schema mapping optimization based on nested mapping was studied.Method to judge nested mapping was provided.A new algorithm for automatic generation of nested mappings using the basic mappings was presented.Enquiry performance of mapping relationship before and after optimization was compared.Experiments showed that it could not only simplify mapping representation,reduce mapping storage scale,and improve efficiency of enquiry and data exchange.

There is a large scale of schema mappings between heterogeneous data sources in the network,which is used for information sharing.As the schema of the data sources change,many direct schema mappings evolved to indirect ones.In this paper the definition and expression of indirect schema mappings are given.To realize the management of indirect schema mappings,the composition of indirect schema mappings is studied.An algorithm for indirect schema mapping composition is proposed,and the effective of indirect mapping composition is analysed by comparative experiments.

To deal with the problem of emitter platform identification caused by the feature measurement uncertainty of the platform from multi sensors, this paper proposes a new identification algorithm based on interval Dempster-Shafer theory (EDST), which models the identification output of each sensor as interval values and combines the interval outputs through interval evidence combination rules. A number of simulations are presented to demonstrate the identification capability based on the IDST algorithm. Simulation results show that the proposed algorithm can not only process the interval input data, but also can deal with scalar input data.

自治异构数据源信息共享的主要问题是如何在P2P环境下对自治数据节点的信息进行统一访问.采用分层结构组织数据源节点能够提高查询效率,减小计算开销,但需要节点根据彼此相似度实现局部的聚类.给出了数据源节点信息发布的形式化描述,提出了基于模式元素匹配的自治异构数据源多重聚集模型以及聚类组织构建过程,采用TA算法解决top-K聚类节点搜索问题,并在此基础上提出TAL算法.实验结果表明,TA和TAL算法能够高效地解决节点聚类排序的问题,特别是TAL算法在聚类节点范围较大时计算性能优于TA.