Social manufacturing leverages the power of social networks and collaborative processes to enhance manufacturing capabilities and supports the sharing of ideas, resources, and information. However, traditional remote maintenance under a social manufacturing context lacks resilience against disruptions and cyber attacks. These issues often lead to interruptions in production. This article proposed blockchain-of-things-based edge learning contracts for federated predictive maintenance (FPM). First, given the diversity and heterogeneity of equipment, an open platform communication unified architecture (OPCUA)-based equipment meta-model is proposed to facilitate the interconnection and data sharing. Second, a Blockchain-of-Things-based secure access control approach is proposed, to directly collect data from controllers. This approach prevents tampering, unlike traditional local database collection methods. Third, to address the security and efficiency needs, an edge learning contract method is proposed for FPM. An integrated learning algorithm based on smart contracts is designed to achieve prediction performance that is comparable to local centralized training while reducing data transmission load and enhancing data security. Finally, a federated predictive maintenance platform is designed and implemented to enhance the system’s resilience, and its effectiveness is verified through case studies.

Porosity is a common defect in the laser powder bed fusion (LPBF) process, significantly limiting its potential in mass customization applications. Existing studies have shown that porosity can be effectively detected by analyzing the thermal history of the melt pool. However, the relationship between thermal features extracted from thermograms and porosity is highly nonlinear, making defect prediction challenging. This article proposes a convolutional neural network (CNN)-integrated Transformer (CiT) model for defect prediction and anomaly localization in LPBF. The convolutional-inception-transfer (CiT) model enhances predictive accuracy by combining global and local feature extraction while leveraging global average pooling instead of the classification (CLS) token after removing the Transformer decoder. Additionally, a lightweight multihead self-attention mechanism is designed to optimize the model structure, effectively reducing the number of parameters while maintaining accuracy. Furthermore, an anomaly localization method based on Score-class activation mapping (CAM) is introduced to identify potential causes of porosity formation, enabling defect traceability in LPBF. The proposed CiT model is evaluated against four state-of-the-art algorithms, demonstrating its superior performance in defect prediction and anomaly localization.

In the realm of industrial automation, corrosion represents one of the primary modes of failure, especially in the case of armored thermocouples exposed to temperatures ranging between 1073.15–1373.15 K. In this context, the selection of metal materials that can withstand high-temperature oxidation and corrosion is of paramount importance. Typically, the corrosion resistance of a given metal material is assessed by measuring the “annual corrosion rate” or “corrosion depth”, which can provide an estimated life expectancy value. However, such an approach fails to account for the individual characteristics of the material, and thus does not conform to objective laws. Rather, the corrosion life of a batch of metallic materials should follow the Weibull distribution, or possibly a normal distribution, as per recent studies that have examined the high-temperature oxidation corrosion mechanism of machine or core components. This investigation effectively combines the standard approach for evaluating metal corrosion resistance in the field of materials with the method of assessing component life in the domain of reliability. Furthermore, we consider the individual differences among materials and provide the life distribution function of metals in corrosive environments and thereby refine the evaluation of metal corrosion resistance. This study ultimately establishes a thermocouple accelerated life evaluation model that enhances the accuracy and efficiency of life evaluations for related products.

Propeller-crossing probability analysis is crucial for evaluating the impact resistance and foreign object exclusion capability of turboprop engines. However, due to the complex structure of the propeller and the uncertainties associated with the impact location as well as the flight attitude of the foreign object, developing a comprehensive model for analyzing the propeller-crossing process remains a significant challenge. This paper presents a novel simulation method that can obtain the probability of a foreign object successfully crossing the propeller using a high-fidelity structure model of the propeller and a comprehensive substituted model of the foreign object. To validate the performance of the proposed method, an analytical model is developed that takes into account the spatial structure constraints of the propeller and the foreign object. The proposed method is applied to calculating the probability of bird ingestion, and the results reveal that the increments in flight speed and aspect ratio of the bird have opposite effects on the propeller-crossing probability, and the values eventually converge to a constant value.