This paper proposes a two-stage pseudo anomaly-guided anomaly detection method (Two-stage Pseudo Anomaly-guided Anomaly Detection, TPA-AD) for axle-box bearing time-series anomaly detection (time series anomaly detection, TSAD) under the setting where only normal samples are available for training. The method first generates pseudo-anomalous windows near the normal boundary using a reconstruction model and per-feature target-error control. It then learns anomaly-sensitive representations through contrastive learning between normal and pseudo-anomalous windows, and finally produces window-level and point-level anomaly scores using k-nearest neighbors (KNN). Compared with existing methods that rely on known fault categories, real anomaly priors, or random anomaly injection, TPA-AD improves the separability of the normal boundary by constructing pseudo-anomalies in boundary neighborhoods and can jointly handle continuous and discrete features in mixed-variable scenarios. The main experiments are conducted on bearing fault detection datasets and degradation-process datasets, with an additional exploratory extension on 13 public TSAD datasets. The results show that the proposed method yields relatively stable anomaly responses, is sensitive to degradation evolution, and demonstrates a certain degree of broader applicability on public TSAD benchmarks and real high-speed-train-related bearing data.

The machining scheme selection (MSS) for features is to choose the optimal machining scheme for a feature before machining. To solve the issue of excessive human subjectivity in the traditional MSS, this paper proposes a simple and easy-to-use method based on process knowledge graph retrieval and through machining scheme similarity matching. First, process knowledge is extracted using natural language processing techniques, focusing on forming ternary groups such as part–feature, feature–attribute, and scheme–resource to construct a multi-level process knowledge graph. This graph is used to retrieve the available machining schemes for the features. Based on the part property, the feature basic information and manufacturing information are used to establish a feature information model and information coding dimensionality reduction. Then, considering the influence coefficient of the process parameter and the usage coefficient of the machining scheme, an improved cosine similarity formula is designed for MSS. According to the maximum similarity, the optimal machining scheme is matched to the feature. Finally, the effectiveness of this method is verified by selecting the machining schemes for six types of hole features on a typical shell part. The results demonstrate that the recommended schemes by the proposed method closely align with the existing mature schemes.

(AFSD) Additive friction stir deposition is a novel solid-state forming method that can form isotropic material structures without solidification defects and the forming characteristics of the forming area and the forming thermal process in AFSD can significantly change the performance of the workpiece, therefore the thermal analysis of the forming process in AFSD is particularly important. The simulation of AFSD thermal process of AA6061-T6 aluminum alloy is missing. In this study, the cell life and death technique in the finite element was used to observe the evolution of the thermal field during the deposition of a single-layer aluminum alloy, and the changes of the temperature field at different positions were analyzed.

Abstract Air conditioning is an integral part of a moving train, and when it fails, it will affect passenger satisfaction and electrical safety on the train. By analyzing the principle and failure modes of the air-conditioning system of a certain type of domestic rolling stock, a fuzzy fault tree is constructed in this article, which has the 33 minimum cut sets, for further quantitative analysis. Then, according to the experience of relevant technical personnel and the analysis of air conditioning failure data, the occurrence probabilities of the bottom events are evaluated using the popular triangular fuzzy number. This fault tree can be analyzed quantitatively to find out the probability of the top event and the fuzzy probability importance of the bottom events. Accordingly, we can know the parts of the air-conditioning system of rolling stock that are relatively prone to failure. Through the verification and analysis of real data, it is proved that this is an effective method to locate the weak parts in the air-conditioning system, which can help to identify the possible failure modes and causes, and guide failure diagnosis and prediction of the air-conditioning system, as well as optimize the maintenance program.

Machining scheme selection is presently time-consuming and inefficient due to singularity and uncertainty in decision-making. Typically, the selection is based on a singular feature, with eventual outcomes remaining uncertain. In response, this paper proposes a feature-group-oriented re-optimized bacterial foraging algorithm to solve this problem by selecting the optimal machining schemes for multiple similar features in one part at once. Our focus is on designing the re-optimized bacterial foraging algorithm to optimize machining schemes, which considers the processes of chemotaxis, fine-tuning, replication, and adaptive migration with re-optimization. Then, comparative studies with varying weights and algorithms are conducted as an illustration of machining scheme selection for the hole feature-group. The results indicate that the re-optimized bacterial foraging algorithm accurately produces three distinct machining schemes based on varying weights. Additionally, the optimal average value outperforms other algorithms in the comparative study, demonstrating the algorithm's validity and superiority.

Machining feature recognition is crucial for achieving automation and intelligence in the realm of intelligent manufacturing. It serves as a fundamental technology for the integration of Computer-Aided Design (CAD), Computer-Aided Process Planning (CAPP), and Computer-Aided Manufacturing (CAM). Traditional methods of machining feature recognition largely depend on manual processes, resulting in low efficiency and limited reusability. In this paper, we propose a novel approach utilizing a 3D convolutional neural network (3D CNN) that is particularly well-suited for recognizing machining features of mechanical components. The network takes triangular mesh data as input and employs a voxelization algorithm to convert the 3D model into a 3D Boolean matrix. This matrix is subsequently processed by the 3D CNN for feature recognition. Experimental results demonstrate that the proposed machining feature recognition method achieves an accuracy exceeding 99% on the 3D models from an open machining dataset. This work holds significant implications for the automatic recognition of machining features in mechanical parts.

Additive Friction Stir Deposition (AFSD) is a key technology in additive manufacturing, where process parameters greatly impact deposition layer properties. Currently, there is no established method for systematically improving these properties through parameter investigation. This study addresses this by applying six machine learning (ML) classification algorithms to a dataset of 130 samples, classifying the ultimate tensile strength (UTS) based on feed rate, rotational speed, and downforce. The Support Vector Machine (SVM) algorithm achieved the highest accuracy at 95.3%. This research provides a precise method for classifying AFSD process parameters and demonstrates the potential of ML techniques for optimizing manufacturing processes, presenting a novel approach to enhancing AFSD deposition layer performance.

Additive friction stir deposition (AFSD), in which molten metal materials are formed into free-form stacked structural parts according to the path design, may have a wide range of applications in high-efficiency mass production. In this study, experiments were conducted for the rotational speed in the AFSD parameters of 6061 aluminium alloy bars to investigate the effects of different rotational shear conditions and heat inputs on the properties of the deposited layer for diameter bars based on the analysis of the micro-morphology, micro-tissue composition, and mechanical properties. The width and thickness of each layer were constant, approximately 40 mm wide and 2.5 mm thick. The particle undulations on the surface of the deposited layer were positively correlated with the AFSD rotational speed. Continuous dynamic recrystallisation in the AFSD process can achieve more than 90% grain refinement. When the rotational speed increases, it causes localised significant orientation and secondary deformation within the recrystallised grains. The ultimate tensile strength of the deposited layer was positively correlated with the rotational speed, reaching a maximum of 211 MPa, and the elongation was negatively correlated with the rotational speed, with a maximum material elongation of 37%. The cross-section hardness of the deposited layer was negatively correlated with the number of thermal cycles, with the lowest hardness being about 45% of the base material and the highest hardness being about 80% of the base material.

Fault samples of marine engine are extremely scarce, and there are unavoidably some hard samples with small inter-class differences, which pose a serious challenge to fault diagnosis of marine engines. This paper proposes a deep metric learning method, namely deep concentric Siamese network (DCSN), to apply strong forces to hard samples towards their corresponding correct distribution areas under small-sample conditions. First, DCSN is committed to learn discriminative information from limited fault samples through a carefully designed metric learning strategy. Then, DCSN distinguishes hard samples using inner and outer boundaries, and applies strong forces to them, making the deep model more focusing on the correct classification of hard samples. Third, DCSN shrinks the distribution area of intra-class samples, which improves intra-class compactness and inter-class separability. Finally, the experimental results on the marine engine fault dataset show that the proposed DCSN yields higher diagnostic performance compared to the considered competitive methods.

The common faults encountered during the operation and maintenance of the Fuxing train sets are summarized and classified into refrigeration faults, ventilation faults, control and power supply faults, and heating faults. Through the analysis of the composition and working principle of the air conditioning system of the train set, the causes of these failures are summarized, and corresponding treatment methods are proposed for different types of failures to provide important reference for the daily operation and maintenance of the air conditioning system of the Fuxing train set and emergency disposal.

Aiming at the problems of high manual participation and long time-consuming sequencing in the process planning of typical spacecraft parts, a process route optimization method based on improved hybrid genetic algorithm is proposed. By analyzing the relationship between parts' machining features, the machining method of machining features is disassembled into steps and a step forward graph is established, and the initial process route is generated based on the constraints of state change rules. The optimization objective function considering the use cost and replacement cost of processing resources is constructed. Aiming at the problem of process route decision-making, the coding method of adding step sequence information and processing resource information is proposed. In order to ensure that the process route sequence meets the process rule constraints, the OBX crossover operator is used to construct the adaptive mutation operator, and the improved hybrid genetic algorithm is constructed by using the artificial bee colony algorithm for process route optimization. The feasibility of the method is verified by taking a spacecraft cabin part as an example.

In this paper, a kinematic separation calibration method of 6R series manipulator is proposed, and its absolute accuracy is improved by a binocular camera and standard sphere. First, a geometric error mapping model for the robotic arm was established, and the error parameters were divided into position parameters and attitude parameters for calibration purposes. Second, in the process of solving error parameters using numerical algorithms, it is easy to encounter matrix ill-conditioned problems. The spectral correction iteration method is introduced to improve the calculation accuracy. Third, three standard balls are installed at the end of the robotic arm as markers, and the center coordinates are measured using a binocular camera to obtain the actual end pose parameters. To verify the effectiveness of the proposed method, a simulation model verification was designed, and the results showed that the separation calibration method was the best. Finally, the IRB-1200 robot was successfully calibrated using the proposed method; the average robot position and angle error after calibration was significantly decreased. The position accuracy was improved by 66.9%, and the attitude accuracy was improved by 86.2%.

近日,在安阳市安阳县邓庄村的农房前,一口看似并不起眼的水井旁边,河南省自然资源监测和国土整治院的技术人员正在忙碌地进行水位校测、水质现场测试、水样采集等工作.

In order to solve the problems of low success rate and low assembly efficiency of the coordinator gyro components in the actual assembly, a prediction model of the combination of gyro components based on the grey wolf optimization algorithm and the short-term memory neural network (GWO-LSTM) was proposed. The model first uses sparse learning feature selection method to extract key assembly parameters of gyroscope components; Then, a violin chart discrete coefficient analysis method was proposed to statistically classify drift performance data. Based on the statistical classification results, a GWO-LSTM neural network classification and prediction model was established. Finally, the effectiveness of the GWO-LSTM neural network classification and prediction model was verified using the assembly of gyro rotor and coil components as an example.

Structural experiment course is a extremely significant course in civil engineering, and under the influence of new engineering background, structural experiment course has higher requirements for perfection, comprehensiveness, innovation and universality.Through the understanding of domestic and foreign issues, this paper describes the process of measuring the sensitivity coefficient of resistance strain gauges and measuring the compressive strength of concrete by rebound method, through the application of ABAQUS software analysis of the building test course in the future development, summarizes the application of ABAQUS in the modeling process of building structure test course, and proposes and constructs a set of reasonable reform and adjustment methods suitable for cultivating applied talents, and made a more systematic demonstration in terms of innovation.

The processing of large-scale cabin support parts for manned spaceflight requires repeated installation, processing and measurement, resulting in low processing efficiency and poor processing accuracy. This paper proposes a method of using a mobile processing robot to complete the processing of bracket parts, and studies the difficult problem of station planning in this method. This paper establishes the coordinate system and coordinate transformation system of each object in the processing process, rasterizes the working space of the processing cabin, and uses the inverse solution of the robot to filter out the feasible station area in the working space, and through the type of support that can be processed The optimal Z-direction feasible site hierarchy is selected with the number of site points, and finally the smallest site set is selected in the optimal hierarchy to complete the site planning of the mobile robot. In this paper, we select a mobile robot with a KUKA manipulator to randomly process a space capsule for experimental verification. The results show that this method can effectively improve the processing efficiency of the capsule bracket.

Compound faults and their involved single faults often have severe overlap in traditional feature spaces, and the strong background noise unavoidably exacerbates the degree of overlap. Aiming at the problem, this article constructs a multi-level discriminative feature learning method, namely deep progressive shrinkage learning, to progressively suppress intra-class dispersions using a few feature-level shrinkage modules and a decision-level shrinkage module for separating compound faults from single faults. First, soft thresholding is embedded as a key part of feature-level shrinkage modules to gradually eliminate noise-related information in the multi-layer feature learning process, in which thresholds are adaptively set using attention mechanism. Second, in the decision-level shrinkage module, high penalties are imposed on the samples that are far from their class centers. Finally, the efficacy of the method in compound fault diagnosis along with single faults has been verified through a variety of experiments.

Interest in the construction of prefabricated ice rink for international competition has increased in recent years, where the ice sheet is directly supported by soft thermal insulation materials. However, bending failure in the ice sheets for these rinks is highly possible because of different compression and tension behaviors. Moreover, the mechanical behaviors of the artificial ice produced layer-by-layer in rinks remain unclear. Therefore, microstructure observations, hardness tests, and three-point bending tests were conducted in this study to better understand artificial ice. First, the crystal structures were obtained through observations in both the vertical and horizontal directions. Then, the hardness of the ice surface at different temperatures, water qualities, ice-making methods, and surrounding environmental conditions was measured using the Shore hardness apparatus. Finally, systematic three-point bending tests on 80 effective ice specimens under a wide range of loading and ice-making parameters were performed. The results show that artificial ice is a typical kind of columnar ice with smaller grain sizes at lower surfaces. The ice surface hardness, roughly normally distributed, was mainly affected by temperature and ice-making mode. Moreover, it was found that all the test ice beam exhibited brittle fracture, and the flexural strength ranged from 0.84 to 2.47 MPa, with the maximum average at a strain rate of 1 × 10–4 s–1. Based on these test results, empirical functions for the effects of the investigated parameters on the flexural strength and effective modulus were developed. Also, the relationships between flexural and tensile strength for artificial ice were established using Weibull law and the coupled criterion. In addition, the linear regression model was established and verified using different prediction methods to predict the ice flexural behavior in practical rinks based on the measured hardness in a simple, reliable, and nondestructive way. The current experiment and analysis are beneficial for the design, operation and maintenance of prefabricated ice rinks.

Aiming at the problem that the sequence of operation in the processing of complex parts directly affects the processing cost, an improved harmony search algorithm with the lowest processing cost as the optimization goal is proposed. The constraint relationship between the operation is represented by a process priority diagram. Then use the topological sorting method on the graph to generate the initial sequence as the initial harmony of the harmony memory. The crossover operator is introduced to ensure the feasibility of the new generation of harmony, and the local search method is used to change the machine and tool resources available for the operation to avoid falling into the local optimum. Experimental results show that the improved harmony search method can effectively solve the cost optimization problem of step sequencing.