The subsea manifold is a core component of subsea production systems, and its reliability directly affects production efficiency. To address leakage localization under abnormal conditions such as sensor failure, a physics-consistent intelligent inversion and fault-tolerant localization method is proposed. First, a nonlinear attenuation model for leakage-induced dynamic pressure signals is established, and multi-stage noise separation and hybrid parameter identification are performed to enhance leakage feature extraction under interference. Subsequently, multimodal features are extracted using a sliding-window strategy to construct a high-dimensional feature space for characterizing the dynamic responses of leakage events. Finally, a feature-enhanced physics-informed neural network is developed, in which a customized physics-constrained loss function and a two-stage training strategy are introduced to coordinate data-driven learning with physical consistency. Guided by the physics-constrained loss function, the spatial mapping of leakage locations is learned from the integrated features of the remaining valid sensor signals, rather than from a single sensor or a fixed sensor set. Experimental results show that the average localization error is only 6.63 cm under the complete-sensor condition and is limited to 15.48 cm even when four sensor signals are missing, indicating that the proposed method maintains high localization accuracy under incomplete sensor information.

While reliability-based design optimization (RBDO) significantly improves engineering safety, computational expenses and accuracy concerns restrict its use in complex, high-dimensional, nonlinear, and black-box scenarios. To address this challenge, a Kriging-AMV-MCS reliability optimization method (KAMRO) is proposed in this paper. The algorithm integrates the Kriging surrogate model, an improved advanced mean value (AMV) method, and a Monte-Carlo simulation (MCS) validation mechanism, aiming to achieve efficient and robust design optimization. First, through an adaptive sampling strategy based on the expected feasibility function and space filling criteria, actively constrained regions are intelligently identified and sample distribution is optimized. Second, a two-stage optimization framework is adopted. In the first stage, the penalty function is used to quickly approximate the feasible domain, and in the second stage, explicit constraints are combined for precise optimization. Finally, an MCS-guided post-optimization process is proposed to further optimize and improve the design scheme when the reliability is insufficient. Through validation with engineering cases, including a two-dimensional analytical example, a seven-dimensional gear reducer structural optimization example, and a three-dimensional pressure control head structure design, this RBDO framework significantly reduces computational costs while ensuring design reliability, offering both optimization accuracy and engineering practicality.

Electrical control systems are widely used in industrial applications, and faults may seriously affect their safe and stable operation. Therefore, fault diagnosis is essential for system condition monitoring and health assessment. Interactive faults involve two or more faults that occur together and influence each other. They are a common and complex fault type in electrical control systems. Existing methods usually rely on raw signals or single-feature modeling. They cannot fully describe the correlation and time-delay features in interactive faults. This limits the effective separation of different fault patterns. To address this issue, a feature-interaction image construction method is proposed for interactive fault diagnosis in electrical control systems. It jointly encodes inter-variable correlations, time-delay relations, and amplitude variation information to transform raw time-series signals into more discriminative images. On this basis, a fault diagnosis model is developed by combining Res2Net and ESCA. This improves the extraction and identification of complex interactive fault features. A subsea blowout preventer (BOP) electrical control system is used to evaluate the performance of the method. The results show that the separability of interactive fault features is improved. Besides, the local confusion between classes is reduced.

Subsea blowout preventer （BOP） is a critical safety barrier for preventing blowout accidents. The hydraulic control system is a key component that determines whether the BOP can operate properly. Once the hydraulic control system fails， the wellhead cannot be closed in a timely manner during a blowout event， thereby causing the consequences to escalate exponentially and posing threats to the environment and personnel. It is necessary to explore an accurate and rapid fault detection and diagnosis method for the hydraulic control system. Considering that the hydraulic control system involves multiple types of feature variables， high complexity， complex coupling relationships， and serious shortage in fault samples under actual operating conditions， a fault detection and diagnosis method combining data preprocessing algorithms and deep learning neural networks is proposed. First， the collected key operational data such as pressure and flow rate were preprocessed， and kernel principal component analysis （KPCA） was introduced to compress high-dimensional and strongly nonlinear original variables into more representative feature information， thereby reducing the learning difficulty of the model and improving its discriminative capability. Then， to address the problems of low proportion and imbalanced class distribution of fault data， a conditional tabular generative adversarial network （CTGAN） was adopted to augment a small number of fault samples， making the data structure closer to the real operating condition distribution. Finally， a deep belief network （DBN） classification model was trained using the augmented dataset to construct a complete fault detection and diagnosis framework. The test results show that the model maintains high identification accuracy under different data imbalance conditions， with the diagnostic accuracy stably exceeding 93% and even up to 98%. It is concluded that deep learning models integrating data augmentation can effectively overcome the constraints imposed by insufficient fault samples on diagnostic performance and provide more valuable technical support for the intelligent operation and maintenance of subsea BOP hydraulic control systems.

Offshore jacket platforms operate for extended periods in harsh marine environments, during which a series of aging-related defects may occur in their structures, leading to reduced load-bearing capacity. Efficient computation of environmental loads and residual ultimate strength, enabling real-time monitoring of safety levels, is critical to ensuring their safe operation. However, existing computational methods suffer from systemic weaknesses and low efficiency, and there is also a lack of effective approaches for analyzing defects in aging structures. To address these issues, A calculation method for the digital twin of aging jacket platforms within the Digital Healthcare Engineering (DHE) framework is proposed, aimed at enabling efficient analysis of environmental and defect parameters obtained from on-site measurements. Specifically, an environmental load computation framework is established based on surrogate models, Morison equation, and wave theories, while a residual ultimate strength computation framework is developed using multi-scale finite element models, nonlinear finite element analysis, and structural defect simulation methods. This approach forms a standardized analytical workflow that balances computational accuracy and efficiency, reduces computation time, and enables effective analysis of monitoring data. The effectiveness of the proposed method is validated using a case study of a specific jacket platform.

Deepwater riser systems are critical infrastructures connecting subsea production systems to offshore floating platforms. During the long-term service, riser systems undergo fatigue accumulation and continuous degradation under complex ocean loads and coupled platform motions, while also being exposed to sudden hazardous events. Existing studies have focused on structural dynamics and fatigue of risers, with limited attention to performance evolution under abrupt hazards. Resilience can effectively characterize the degradation, retention, and recovery of system performance or functionality during such disturbances. Accordingly, a probabilistic resilience evaluation framework is established for deepwater riser systems under multi-source disturbances. Progressive structural degradation is characterized using coupled dynamic analysis and fatigue modeling, while sudden disturbances are described by hazard probability models. Both are integrated into a dynamic Bayesian network to capture the time-varying evolution of multi-state system performance. A production response model is developed to relate system health states to production capacity, and a curve area-based dual-metric method is proposed for resilience evaluation in terms of remaining useful life and production response. A case study in the South China Sea demonstrates that the proposed method captures the time-varying evolution of system performance and functionality under multi-source disturbances and enables quantitative evaluation of life and production resilience.

Buried natural gas pipeline leaks can cause significant harm to life, property, and environment. Thus, the risk analysis of buried natural gas pipelines is essential. Failure mode and effects analysis (FMEA) has been widely utilized for engineering risk assessment. The current study provides a new integrated risk analysis method based on cloud model and FMEA with some merits over other conventional approaches. First, this method absorbs the virtue of cloud model in reflecting the randomness and fuzziness of experts’ linguistic assessment in case of insufficient information. Second, a hybrid dynamic algorithm enhanced with dimensionally reduction technique is proposed to determine expert weight, considering the individual information and the agreement degree of assessments. Third, a new integrated weighting method combined with the simplified best-worst method and cloud-entropy method is applied to compute the risk factors’ weights, which can fully reflect the risk factors’ relative importance. Fourth, the risk ranking based on compromise strategy is obtained by using the decision logic of VIKOR (Multi-criteria Optimization and Compromise Solution) to seek a compromise between group utility and individual regret. The detailed processes of the methodology are described by implementing risk analysis of the buried natural gas pipeline system of a gas field in China. The proposed model successfully prioritized 27 failure modes, revealing that failures related to mechanical and thermal stress within the corrosion effect category constitute the highest risk. These findings provide clear guidance for pipeline operators to allocate inspection and maintenance resources more effectively, enhancing overall pipeline safety and integrity.

Identifying early faults can ensure the safety of hydraulic systems. This is needed for industrial production. Obtaining sensor data and conducting a single diagnosis is the main fault diagnosis method. This method is prone to overlooking early faults. To address that, a digital twin (DT)-driven fault diagnosis method is proposed for detection of early faults in hydraulic systems. A fault diagnosis model capable of identifying faults at varying degrees of severity is established. An iterative updating mechanism is employed to construct a DT model of the hydraulic system. A feedback-based verification process is integrated into the DT framework. It is used to correct incorrect diagnoses. Misclassified faults are fed back into the system for re-diagnosis, and a fault degree matching mechanism is implemented to confirm fault severity. The method is validated using a laboratory hydraulic system. Experimental results demonstrate that the method improves the accuracy of early fault detection.

To address the problems of insufficient endurance of autonomous underwater vehicles (AUVs), difficulties in the on-site utilization of marine green energy, and the lack of underwater wireless charging solutions for multiple loads, this paper proposes a modular underwater wireless power transfer (UMWPT) system based on multi-node resonant circuits (MNRC), which can achieve simultaneous multi-load output with only one high-frequency inverter and effectively support energy supply for multi-AUV cooperative operations. Firstly, models of impedance, mutual inductance and eddy current loss for a single-node coupler in seawater are established, and through finite element simulation, the coil current, magnetic field strength at characteristic points and variation of magnetic induction around the coil in seawater are analyzed; on this basis, a circuit model is built for a dualoutput system, and the influence of eddy current loss on multi-node resonant circuits is studied. Finally, a dualoutput prototype is constructed and tested at 300 kHz, which achieves a maximum total output power of approximately 45 W and a peak transmission efficiency of 77.0% at a standard salinity level of 40 parts per thousand, and the output performance under different transmission distances is tested to provide a theoretical basis for system power distribution.

Subsea control systems operating under high hydrostatic pressure and low temperatures require high reliability. Redundancy is a well-established method for improving reliability and availability and for reducing the frequency of replacements. Nevertheless, under extreme subsea conditions with multi-source uncertainties, redundancy design may induce a redundancy paradox: nonlinear growth in resource consumption can ultimately reduce overall system reliability. Therefore, an optimization framework that unifies stress condition characterization, component-importance quantification, and reliability-cost coupling is essential. To address this challenge, a multi-model fusion-based redundancy allocation method for subsea control systems under Bayesian stress-conditional importance is proposed. This methodology establishes an end-to-end redundancy allocation framework, integrating multiple models while unifying the modelling process and coupling mechanisms. A Bayesian stress-conditional importance metric is introduced as the basis for redundancy decisions, and uncertainties in the impact of component failures on system reliability are addressed. Systematically evaluates component importance, feasibility, and redundancy allocation, transforming the redundancy assignment problem into a multi-objective optimization problem. Subsequently, a multi-objective particle swarm optimization algorithm incorporating the Bayesian stress-conditional importance metric is employed to determine the system's optimal redundancy configuration, aiming to maximize system reliability while minimizing costs. The method is validated against benchmark results from a typical series-parallel system and a complex bridge system. A case study on a subsea light intervention equipment control system further demonstrates high reliability and accurate optimization performance. Overall, the proposed approach supports high-reliability design and contributes to safe and dependable operation of subsea control systems.

Significant failure dependence exists among components of engineering systems, leading to rapid failure propagation and state changes of components. Such compounding dynamics render conventional maintenance strategies prone to inefficiency like over-maintenance, under-maintenance, or maintenance delays. To address the issue of dynamics, this paper proposes a dynamic importance ranking-driven sequential multi-objective maintenance optimization (SMOMO) model. Using a rolling time window approach, the integrated importance of components is dynamically evaluated and ranked based on their real-time states within each window to guide maintenance decision-makings. The window then rolls forward, updating the calculation of long-term expected system cost and availability, based on which a Multi-objective Particle Swarm Optimization (MOPSO) algorithm optimizes maintenance scheduling for cost minimization and availability maximization across the rolling horizon. The proposed methodology is demonstrated and validated using a case study of subsea transportation system. Results illustrate the necessity of dynamic importance evaluation for capturing evolving system status and the effectiveness of the proposed model in optimizing maintenance schedules.

Engineering systems are characterized by their large number of components and high degree of integration. Maintenance activities are crucial to sustaining system availability during the long service period. The traditional individual maintenance strategy based on component failure, resulting in frequent and unscheduled maintenance activities, which may lead to high maintenance costs and compromise overall system reliability. A multi-level predictive group maintenance optimization method is proposed, considering the stochastic degradation dependency among multiple components. By integrating the natural degradation process and the influences of random disturbances, a more accurate prediction model for the remaining useful life is established, which serves as the resilience evaluation metric. Based on the system structural hierarchy, the group maintenance strategy is respectively optimized at the component level, subsystem level, and system level. The group maintenance strategy is optimized from both technical and economic perspectives. From the technical perspective, the real-time resilience for the system is optimized. The economic objective is to realize the optimal maintenance cost during the whole service life cycle. A case study on the subsea Christmas tree is carried out to illustrate the advantages of the proposed group maintenance optimization method.

The study of flutter characteristics and mechanism of ultra-long wind turbine blades is contingent upon aeroelastic tests of scaled-down models. The construction of accurate aeroelastic scaled-down blades is a recognized challenge in the field. This study theoretically constructs aerodynamic-structural similarity criteria for wind turbine blades. The similarity relationship between the natural frequency and length ratios of 3D isometric similarity (3DIS) and shell similarity (SS) is expressed as Am = 1/Al and Am = 1/Al2, respectively. Based on the 3DIS and SS, the stiffness control strategy for scaled-down blades is proposed and its feasibility verified. To explore the applicability of the aerodynamic-structural similarity criteria, the NREL 15 MW blade CSD model with composite material layup and CFD model are constructed. The error functions to evaluate aerodynamic similarity and SS are derived and defined. The aerodynamic similarity and SS errors of the scaled-down blades with different length ratios are calculated. On this basis, a scaled blade model construction method that allows researchers to make open selections based on manufacturing and test conditions is proposed. The similarity criterion, stiffness control strategy and error evaluation method are broadly applicable to composite shell structures represented by blades, providing theoretical basis for the scaled-down models of aeroelastic tests.

Due to the significant hazard posed by gas leaks in enclosed spaces and the current lack of mature research in this area, the present paper proposes a methane-leak dispersion prediction method based on a semi-empirical closedform model. It is tailored to small-orifice, high-speed leak scenarios in semi-enclosed spaces such as commercial kitchens, and optimizes both gas concentration forecasting and sensor-placement design. A laboratory platform was used to simulate the methane leakage process, and a two-stage parameter calibration strategy combining particle swarm optimization (PSO) with the L-BFGS-B algorithm improved model accuracy and computational efficiency. Experimental results show that the conventional Gaussian plume model yields substantial prediction errors in the near-source and ceiling-reflection regions, with an average RMSE of 0.49. By introducing momentum jet, buoyancy coupling, and a temporal correction factor, the semi-empirical closed-form model reduces RMSE to 0.27 and raises R2 to 0.91, confirming high fidelity across all spatial and temporal scales. Through optimized sensor placement, the final scheme achieves full deployment within 5 min and attains over 90 % leakage-detection accuracy, demonstrating clear advantages over traditional computational fluid dynamics (CFD) approaches in engineering practice. The method's short computation time and low hardware requirements make it well suited for on-site deployment.

Subsea Christmas tree (XT) systems are critical assets in offshore production. Among their subsystems, the hydraulic control system—comprising the Subsea Control Module (SCM) and multiple functional valves—must remain reliable under deep-sea conditions. In practice, malfunctions of SCM control valves can induce symptoms indistinguishable from intrinsic failures of downstream functional valves, creating diagnostic ambiguity. This study develops a dynamic fault-diagnosis framework based on Dynamic Bayesian Networks (DBN) that models component degradation over time and integrates discretized sensor evidence and HMI alarm/status flags as diagnostic inputs. The framework is activated after an anomaly is detected by the HMI system, performing fault verification, cause differentiation, and reliability evaluation through probabilistic reasoning. A key capability is the differentiation between upstream-induced and intrinsic valve failures, enabling precise fault attribution even when system-level symptoms are similar. Weibull-based priors capture component degradation trends, while temporal dependencies within the DBN enable both retrospective and predictive inference, supporting validation of current alarms and estimation of future fault probabilities. Case studies on a representative subsea XT system demonstrate the framework’s feasibility and strong potential for practical application, using quantitative time-based indicators, including time to warning and time to fault confirmation, to evaluate diagnostic timeliness, degradation progression, and fault cause identification under uncertain and degraded conditions.

Offshore oil well blowout preventer （BOP） is a core safety guarantee for offshore drilling operations， and the reliability of its hydraulic control system directly affects the safety of platform and personnel. However， the reliability assessment is challenged by data scarcity， information imbalance， and modeling complexity. This paper presents an assessment method based on multi-source data fusion and dynamic Bayesian network. This method can enhance the accuracy of reliability assessment by integrating historical and test data to overcome data scarcity problem， while incorporating multistate degradation and correlation features. First， the Bayesian method was used to integrate data from different sources， and the posterior distribution was obtained by Markov chain Monte Carlo （MCMC） simulation. Then， the DBN was used to establish a multi-level degradation state assessment model of components and analyze the influence of maintenance strategy. Finally， the DBN was used to conduct reliability inference. The results show that the reliability of the system is the lowest in the tenth year， reaching 81.31%. After implementing the maintenance strategy， the usability is improved to 91.81%. This method is superior to the traditional method relying solely on test data， and effectively realizes the reliability modeling and assessment of complex systems. This method performs outstandingly in the reliability assessment of small sample complex systems， provides important theoretical and technical support for the maintenance of hydraulic control system of deepwater BOPs， and has high engineering application value.

The spent fuel pool is constructed by welding stainless steel and is susceptible to corrosion, damage, degradation, and micro-leakage due to prolonged exposure to boric acid and high-radiation environments. Accurate three-dimensional reconstruction of defects is challenging with traditional methods because of the coarse grains in stainless steel, severe scattering attenuation, high background noise, and the complex morphology and small scale of micron-sized pores. In this study, we propose a novel method for detecting and 3D reconstructing internal micropores in stainless steel using an ultrasound microscope. The proposed method involves three key steps: (1) signal denoising and filtering using the adaptive sparrow search algorithm combined with variational mode decomposition (ASFSSA-VMD), (2) slice-by-slice threshold segmentation for image mask extraction, and (3) 3D reconstruction of the segmented masks using the Visualization Toolkit (VTK). The results show that this workflow effectively addresses the challenges of noise interference and segmentation accuracy, enabling high-quality 3D visualization of minor defect. The 3D microleak hole in the thin stainless plate can be shown visually. The results indicate that the maximum reconstruction error in the stainless steel base material is 0.2 mm, while the maximum reconstruction error in the stainless steel cladding layer is 0.5 mm. It provides robust technical support for the safety maintenance and defect characterization of spent fuel pools.

The electro-hydraulic composite control system is widely used in industrial equipment. It has the advantages of fast response speed and high load level. Sudden faults would lead to shutdown of the production. These would cause production losses, even the safety accidents. Fault diagnosis can obtain the states of equipment faults, promptly alert when faults occur, reduce accidents, and minimize production losses. Due to the large number of components and few sensors in the electro-hydraulic composite system, the faults of many components often exhibit the same characteristics. This increases the difficulty of troubleshooting. To identify faults that require similar features, a digital twin-driven group fault diagnosis method is proposed for faults with similar symptoms. A functional fault grouping strategy was introduced for grouping faults for diagnosis. A fault diagnosis model was established for inferring faulty components during the diagnosis and re diagnosis process. A digital twin model was established to simulate diagnostic results and verify their correctness. The proposed method is evaluated through application to an electro-hydraulic system. The results indicate that the grouping diagnosis strategy improves diagnostic accuracy by approximately 3%-5%.

The blowout preventer (BOP) is a safety-critical device in offshore oil and gas production, and the electrical control system (ECS) of it is essential for safe operation. Reliable assessment of the ECS is therefore of great importance. Existing methods mainly rely on precise modeling of hardware degradation to achieve accurate reliability evaluation. However, they generally neglect the influence of human factors on equipment, leading to discrepancies between theoretical predictions and practical conditions. To address this limitation, a reliability assessment method integrating human-machine interaction is proposed. Through a multi-scenario human reliability assessment model, external environmental factors and operator-related factors are simultaneously incorporated. Through a hardware reliability model, multi-state degradation, hardware redundancy, and hierarchical dependency relationships are simultaneously described. A human-machine interaction is established to represent the influence of human-factor states on hardware reliability and the feedback effect of equipment states on human reliability. A subsea BOP in the South China Sea is used as a case study to validate the proposed method. The results indicate that incorporating human reliability yields more realistic and reliable reliability assessment results.