The assembly process of Printed Wire Board (PWB) components is becoming increasingly intricate and multifaceted. Due to the constraints imposed by electronic component specifications and manufacturing cost considerations, the coexistence of through-hole technology and surface-mount technology (SMT) is anticipated to persist for an extended period during the transition from traditional assembly methodologies to SMT. Most enterprises adopt a dispensing process, heavily relying on parameters delineated by equipment manufacturers or experiential insights derived from production practices. When modifications occur within an enterprise's production system, there is often a marked increase in defects associated with printing quality throughout the manufacturing process, significantly escalating repair labor requirements and man-hours expended. Furthermore, any occurrences of inspection failures can disrupt normal customer operations, resulting in substantial risks related to product quality assurance. To systematically enhance the issue of surface mount quality and production capacity, reduce investment in SMT dispensing equipment, and boost enterprise competitiveness, this study employs statistical data and applies 6σ and management statistics methods, utilize MINITAB statistical software as the analysis tool, redefine high customer satisfaction and low defect criteria, conduct process analysis, and derive optimal parameters for thick mold equipment. Production practice demonstrates that these new parameters significantly enhance surface mount quality while improving overall enterprise benefits.
The escalator system, which experiences data fluctuations during continuous daily operation, plays a vital role in subway stations. To accurately assess the health status of the escalator, it is crucial to predict its future condition. To mitigate the effects of significant fluctuations in data and leverage the correlations among cross-variables, the Multi-scale Average Pool Transformer (MA-Transformer) is proposed. The model pre-processes the data using average pooling layers with different kernel sizes, segregates the intricate escalator data into trend and residual components, and predicts them separately with two independent networks. This approach enhances the essential data representations and mitigates the adverse effects of frequent data fluctuations. Given the substantial number of variable types in escalator data, this paper introduces two modules that focus on the self-attention mechanism to model the inter-relationships between multiple variable types. Additionally, the trends and residuals output from different average pooling layers are fed into the self-attention layer, which provides a more comprehensive representation of features at the same time point. This enables the model to better capture the overall trend changes and the distribution of anomalies that deviate from the trend. This study collected data from 19 classes from 2 escalators at the engineering site for training and testing. Experimental results demonstrate that the MA-Transformer achieves superior detection accuracy, outperforming evaluated time-series prediction baseline models in terms of both Mean Absolute Error (MAE) and Root Mean Square Error (RMSE).
Traffic flow prediction is a key issue in urban traffic management, providing important support for urban planning and traffic diversion. In traffic flow prediction, LSTM (Long Short-Term Memory networks) and GRU (Gated Recurrent Units) capture temporal information with different focuses, but single-cell candidate states often lack full feature aggregation, limiting model performance. Challenges include long-term dependencies and large-scale data processing. This paper proposes an innovative dual-gating mechanism integrating LSTM and GRU candidate states to enhance accuracy. Experimental results show significant outperformance in Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE), improving traffic flow predictions and urban traffic management.
In present work, the failure analysis was conducted on a switching power supply using various methods including macroscopic morphology, circuit analysis, disassembly morphology, X-ray CT, microscopic morphology, and corrosion product analysis. The results indicated that the MOS tube was the location of the failure in the switching power supply. Additionally, the surface solder of the exposed leads was found to have corroded to some extent due to the influence of corrosion media. Combined with the working principle of enhanced NMOS pipe, the cause of the failure is the switching power supply under the high power board device temperature rise, plus the service environment humidity, the pin corrosion, the pin and the packaging interface sealing performance is reduced, the water molecules in the air into the packaging device, and there is a large electric field inside the MOS tube during the working process. The internal metal of the device was affected by comprehensive factors such as temperature, humidity, chlorine ions, and electric field distribution, leading to electrochemical migration and subsequent breakdown between G and D terminals.
Harvesting mechanical energy from low-frequency, multidirectional daily vibrations have been considered as an excellent candidate to replace the conventional batteries in many fields. In order to achieve effectively harvesting the energy in these vibrations, this paper designs a piezoelectric vibration energy harvester with a ball-bowl structure. Benefiting from the structural characteristics, the proposed scheme can capture vibrations in different directions, and the working frequencies show a good consistency and are close to the frequency of daily vibrations. The captured energy is transmitted to the piezoelectric beam by magnetic interaction and converted into electrical energy. The analysis and experimental results have demonstrated that the designed scheme has a good performance in harvesting multidirectional vibration energy.
In this article, we will review our advances in intelligence modeling the mind and explore various efforts to mimic the processes that would allow humans to perform cognition files and line widgets of processing. It presents ideas and stresses how understanding cognitive functions is vital for AI progress. Three principal modeling strategies are examined: cognitive mind modeling, assignation mind modeling, and intelligent body mind modeling. In the article, he explores a few theories, such as symbol systems and the ACT-R framework. It further argues that learning seems limited in encoding models of the mind, citing its successes in language-oriented tasks and its limitations in translating basic mechanisms like reward to model-level language and data from neuroscience. Additionally, it investigates ways of assessing and enhancing learning theory of mind abilities based on tasks that utilize end-to-end function. The review reiterates the need for an improved understanding of human-level thought processing to develop intelligent systems that can better discern intentions and feelings. It provides some ideas on evaluating this based on deep learning approaches. The review argues that viewing learning through the lens of systems yields new insights into human thought and social behavior that could enable the development of novel and more adaptive AI technologies.
There are many factors affecting the damping properties of magnesium alloys. The damping capacities and microstructure of Mg-6Zn-1.5Cd-0.5Zr(abbr. MC) alloy were investigated on three different treatment states such as cast alloy, solid solution alloy and extruded alloy. The result of research shows that the average grain size of MC alloy as cast, as solution and as extrusion is 80μm, 40μm and 4-20μm, respectively. Throughout the test strain range, the strain damping capacities of MC alloy as cast is lower than that of Mg-6Zn-0.5Zr (abbr. M)alloy as cast. The length between weak pinning points on the dislocation line is reduced by Cd elements solution in the a-Mg matrix. The strain independent damping capacities (Ql-1) of MC alloy as cast equal to about 0.00108, as well as solid solution alloy and extruded alloy almost. When the strain amplitude (ε) exceeds the first critical de-pinning strain amplitude (ε1cr), the strain-dependent damping value (QH-1) of the solid solution and extruded alloys are greater than that of the cast alloy. When strains ε > 4.18×10−4, QH-1of solid solution alloy and extruded alloy reach above 0.01025 and 0.01057, respectively. In the test range, the damping capacity QH-1of MC alloy in casting state, solid solution state and extrusion state reaches the maximum value, which are 0.01201, 0.01406 and 0.01491, respectively. The damping mechanism of MC alloy accords with Granato and Lücke ( G-L) theory.
Existing methods include traditional methods and deep learning-based methods, but there are problems such as inefficiency and poor generalization ability. In this paper, a new method named IARUNet is proposed, which integrates Improved Atrous Spatial Pyramid Pooling (IASPP) module and ResUNet. The IASPP module replaces the convolution module in ResUNet network, which improves the ability of building extraction. Notably, experimental results from the Massachusetts dataset demonstrate that IARUNet achieved an average intersection over union (MIoU) of 84.31%, a recall rate of 89.42%, a precision of 90.04%, and an F1 score of 89.94%, which is significantly improved compared with other existing models. In addition, the IASPP module enhances the retention of context information, further improving the extraction effect.
The bi-directional bending deformation of the lens-type space deployable boom (LSDB), with high specific stiffness and high storage ratio, is prone to result in structural damage during folding and unfolding process. Mechanical analysis models of the LSDB are established based on theoretical equivalence and finite element method, and comparative verification is conducted. The influence of design parameters such as lay-up angle, arc radius, and material thickness on the stress and folded strain energy of LSDBs is analyzed. The results show that the simulation analysis is close to the theoretical calculation results, which can basically validate the rationality of models. The stress of LSDB in the folded state decreases with the increase of arc radius, increases with the increase of material thickness, and the influence of lay-up angle on stress has symmetry and is affected by the proportion of arc radius to coller radius. The folded strain energy increases with the increase of material thickness, decreases with the increase of roller radius and arc radius, the influence of lay-up angle on strain energy also has symmetry, and there exists an optimal angle that minimizes the strain energy. The analysis conclusion can provide design references for improving the strength and deployment performance of the LSDB.
In two-phase flow, accurately capturing sharp interfaces is essential. Physics-Informed Neural Networks (PINNs) provide a new way to capture interfaces. This paper introduces a new framework that uses Discrete-time Physics-Informed Neural Networks (DtPINNs) to solve the volume of fluid (VOF) advection equation, offering a new approach to interface capturing in two-phase flows. With this framework, we propose an adaptive collocation-point refinement algorithm, which improves the precision of capturing sharp interfaces, reducing errors and diffusion. Experiments, including cases with translation and deformation, show that the DtPINNs method maintains sharp interfaces, outperforming traditional numerical methods. In a 2D single-vortex deformation case, DtPINNs achieved a relative \({L}_2\) error of 0.0760, much lower than the 5th-order WENO-JS scheme (0.5534) and the 1st-order Upwind scheme (0.8879). This shows that DtPINNs are better at capturing complex interface shapes while keeping accuracy and minimizing numerical diffusion.
Recently, intelligent computation algorithms have been extended into every field, and the trend of digitization and intelligence in the decision-making system of supply chain management is becoming more and more obvious. In this article, the pricing intelligent optimization management for the duopoly supply chains is considered and the optimal pricing model based on the Stackelberg-Nash equilibrium model is constructed. The corresponding autonomous two-level gradient-based equilibrium algorithm for the pricing management systems is designed to analyze the modeling characteristics of the pricing optimal model with the simulation results. Firstly, the profit models of the retailers and manufacturers are given by analyzing the demand and occurred costs for the duopoly supply chain systems. Based on the above models, the optimal pricing strategy based on the Stackelberg-Nash game are constructed and the corresponding Nash equilibrium (NE) points for the optimal pricing are obtained according to the profit models. Furthermore, the novel two-level gradient-based algorithm for the pricing management systems is designed to obtain the optimal equilibrium points of the unit retail price, unit wholesale price, and length of the replenishment cycle. Finally, the designed gradient-based algorithm is implemented through the simulation, and the optimization characteristics of the game-based optimal model are analyzed. The NE-based optimal prices and the convergence of the proposed algorithm for retailers and manufacturers are illustrated by the simulation results.
This paper focuses on studying and analyzing the kinematics characteristics of the Naval Ship Survival Training Simulation System (NSSTSS) in the case of longitudinal shake of the anti-sinking chamber on the ship. Based on the analysis of the motion state and the designed system structures, a mathematical model of the system is established. Geometric and analytical analyses methods based on mechanical theory allow determining the relationship between the kinematics characteristic parameters of the system. With the piston-cylinder active drive system, the kinematics motion characteristics of each part are determined including their position, velocity and acceleration. Accordingly, the forward kinematics (FK) and inverse kinematics (INV) problems are considered and analyzed based on actual system parameters. With the piston stroke limited to oscillation within 80mm in both directions, the longitudinal shake angle of the anti-sinking chamber does not exceed 3.9 (degree). The angular velocity and angular acceleration values of the corresponding joints are determined according to the input kinematics values of the piston. With the requirement that the anti-sinking chamber has an actual longitudinal shake value not exceeding 2 (degree) in both directions according to real-time constraints, the required linear motion value, velocity and acceleration of the piston in the cylinder are specifically determined through the INV problem. The research results serve as the basis for analyzing the system's movement capability, accurately controlling the positions of interest and contributing to completing the problem of describing the overall motion state of the entire system.
This paper mainly introduces the wind speed and wind direction detection device based on the annular wind tunnel. The device consists of two parts, which are the wind speed detection part with the standard pitot tube as the standard device and the wind direction detection part with the standard dial as the standard device. Through the analysis of the principle, detection method and uncertainty of the wind speed and wind direction detection device, it shows that the device has high accuracy and good reliability in the measurement of wind speed and wind direction.
Digital transformation is the process of transforming the traditional practice-based fishery management system into that based on science and calculation. Among them, the use of data and algorithms to analyze and predict the data is the most influential technology. This paper first studies the necessity of the digital transformation of the high sea fishery governance, and analyzes the application of computational science in the high sea fishery governance by analyzing the challenges faced by computational science in the high sea fishery governance. On this basis, this paper proposes that the international community should accelerate the deep integration of computing science and fishery governance in the high seas, and take data as an important engine to promote the collection and analysis of global Marine environment information, so as to realize the sustainable utilization of fishery resources in the high seas.
Isothermal compression tests of ingot-extruded GH4710 alloy were conducted within the temperature range of 1050 ℃ to 1150 ℃ and at strain rates ranging from 0.01 s-1 to 5 s-1. The analysis focused on the flow behavior and the critical condition for dynamic recrystallization (DRX). The stress-strain curves exhibited four stages: strain hardening, abnormal softening, rheological softening, and typical DRX (TDRX). Precise determination of the critical strain (\({\varepsilon }_c\)) associated with the initiation of abnormal softening and TDRX was based on the inflection points in the work hardening rate curves (\(\theta\) - \(\sigma\)). Furthermore, the constitutive equations for abnormal softening and TDRX initiation were established.
In this paper, we examine a method for automatic chord estimation (ACE) using foundational models. We present a series of configurations for a system that estimates chord progressions by fine-tuning a foundation model with self-supervised learning, which addresses the problem of insufficient labeled data in music informatics, and examine how hyperparameters and minor configuration changes in the fine-tuning process affect estimation accuracy. The purpose of this study is to show how an ACE system using a foundation model pre-trained from unlabeled data should be specifically designed and implemented, and to indicate the factors during training that affect its accuracy. In the study, the conditions at learning time were compared using MusicFM, which was published in a previous study that used the foundation model to acquire musical information. The comparison revealed that while the application of techniques in deep learning, such as the use of the Nadam optimizer and the implementation of stepwise fine tuning, improved accuracy, factors that increase the complexity of learning per epoch, such as the weighting of the loss function and the complexity of the output layer, could hinder the improvement in accuracy. Overall, the model performed as well or better than existing models in estimating chord progressions for triads. These findings contribute to the promotion of the use of pre-trained foundational models in ACE, helping to understand the knowledge from the first steps of implementation to further extensions.
Sintering process is one of the most important parts during the whole iron and steel production, the quality of the produced sinter is directly affected by the stability of the sintering process, and high-quality sinter is beneficial to the subsequent iron-making process. As a crucial parameter characterizing the state of the sintering process, burn-through point (BTP) is typically utilized to determine whether current sinter production is stable, therefore the stable control of the BTP is the key to controlling the stability of the entire sintering process. In this paper, an intelligent sintering process control method consisting of two modules is proposed. One module based on the deep neural network (DNN) is designed to achieve accurate prediction of BTP, the other module is designed to stabilize the BTP by adjusting the sintering machine speed, according to the deviation between the predicted BTP from the first module and the target BTP set by the expert. Moreover, the speed adjustment model of sintering machine is established by a convolution neural network (CNN), whose input design is similar to the observation logic of manual adjustment by operators. The experimental results show that the sintering machine speed adjustment recommended by the method is highly consistent with that in the actual sintering production process. The proposed intelligent sintering process control method provides a favorable guidance for the speed adjustment of sintering machine, and implies the foundation for the closed-loop control of the sintering machine speed.
In this study, we conduct a network traffic analysis of Finnish voting advice applications. We analyze the potential third-party data leaks in 8 voting advice applications first during the 2023 parliamentary election and then during the 2024 presidential election. The goal is to study whether sensitive information, such as individuals’ political opinions, leak to third parties. Furthermore, the paper also describes the change in privacy of voting advice applications after the leaks found in 2023 were covered in the media. The findings indicate that media attention and public discourse has had a clear positive impact on the privacy of Finnish voting advice applications, but a few data leaks and privacy concerns still remain. We conclude by discussing how to protect political opinions and democratic processes in increasingly digital society.
The electronic safety and arming device test system is developed to ensure the reliability of weapon systems. In view of the low test efficiency and high requirements for testers in manual testing during the current testing process, this paper proposes a Labview producer/consumer model that uses PXI bus technology to control multiple instruction parameters to achieve automated electrical performance function testing of the product to be tested. The basic functions of the automated test system, the overall system architecture, circuit principle design, software design and test process are described. The system aims to improve the test efficiency and accuracy of electronic safety and arming devices, reduce human errors through automated test processes, and provide detailed data analysis and reporting functions. The test verification results show that the test system hardware design can well meet the test requirements, the software mode is easy to use, the functions are complete, and it has good results in practical applications.
Music Information Processing has gained popularity in computer science, with the Generative Theory of Tonal Music (GTTM) used in various applications, such as music analysis and melody generation. However, the ambiguity of GTTM and the inconsistency of its rules make its computerized implementation difficult. This paper addresses a specific ambiguity related to the height of the time-span tree, a key component of GTTM. We propose three methods for defining this height and evaluate their effectiveness in generating prolongational trees, another output of GTTM. Our experiments demonstrate that one of the proposed methods is somewhat useful. We conclude that overall stability rather than local stability conditions is crucial for generating a prolongational tree. The proposed method advances the automatic application of GTTM, especially in generating prolongational trees, and ultimately contributes to the further development of research based on GTTM.
