The structural health monitoring of offshore wind turbine towers is critical for ensuring their operational integrity. The condition of the tower structure is characterized by its modal parameters, which serve as effective health indicators. These parameters are identified through modal identification techniques. In this study, an automated modal identification method is developed for tower vibration data. The identification method is based on Stochastic Subspace Identification (SSI) and Fast Search Density Peaks Clustering (FSDPC), which combines the reconstructed Hankel matrices for spurious pole elimination with FSDPC clustering for automated modal identification. This method is conducted on a 5-DOF numerical model for verification and robustness testing. A 4 MW wind turbine field structure is also introduced as the studying case. The automated modal identification method is further verified based on the offshore wind turbine tower vibration monitoring data. The identified frequencies measured from the monitoring data are analyzed and compared to those of the numerical method using eigenvalue analysis. The results show that the SSI-FSDPC algorithm enables fully automated, robust modal identification, providing valuable support for tower structural health monitoring.

The prediction accuracy of platform motion and mooring tension has a significant impact on the structural safety and survival ability of floating offshore wind turbines (FOWTs) in deep-sea environments. However, traditional numerical simulation calculations are costly, and existing single prediction models are unable to accurately predict the complex coupled dynamic responses. This study, based on convolutional neural networks (CNN), bidirectional long short-term memory networks (BiLSTM), and attention mechanism (AM), combined with Bayesian optimization algorithm, provides an accurate and efficient method for the prediction of FOWTs' motion and mooring tension. In the experimental stage, using the NREL 5 MW wind turbine on the OC4 semisubmersible platform as the object, a full-coupled simulation is conducted in the OpenFAST software to generate datasets covering rated and extreme sea conditions, and the proposed hybrid model is trained and tested. The results show that compared with seven benchmark models, the proposed model has a higher accuracy rate, especially in mooring tension prediction, with the coefficient of determination (R2) reaching 0.973. Moreover, compared with the traditional single-output method, the multi-input multi-output (MIMO) approach reduces the total training time by 64.9%. The results confirm the prediction accuracy, efficiency, and robustness of the model, and demonstrate its potential for engineering applications.

Abstract With the expansion of urban water supply networks, issues such as pipeline corrosion, sedimentation, and structural damage have become increasingly prominent. Vision-based inspection using pipeline robots has emerged as an effective approach for intelligent detection of internal defects. To address these issues, this paper proposes a pipeline defect detection method based on ADS-YOLOv8s, combined with adaptive region of interest integrated Otsu thresholding (Adaptive ROI Otsu). ADS-YOLOv8s improves YOLOv8s by adopting an enhanced bounding box regression loss (Alpha-IoU), designing a dynamic multi-scale detection head (DyHead), and integrating a feature enhancement module (SPPFCSPC) that combines spatial pyramid pooling with cross-stage partial connections. A dataset is built by extracting frames from pipeline inspection videos and integrating additional image data. The proposed model detects multiple defect types. Pipeline images are first segmented using Adaptive region of interest (ROI) Otsu for binarization and edge extraction, then the pipeline structure is modeled via least squares circle fitting to enable adaptive ROI localization. Experimental results show that ADS-YOLOv8s achieves a precision of 92.1%, a recall of 90.1%, and an F 1-score of 91.1%, outperforming several comparative YOLO models, while Adaptive ROI Otsu suppresses background interference and improves pipeline feature extraction accuracy. The method also provides reliable visual information for pipeline boundary extraction and geometric center localization, which may support subsequent robot positioning and centering tasks in fully flooded pipeline environments. Unlike conventional pipeline defect detection methods that mainly focus on object detection accuracy, the proposed method integrates adaptive ROI extraction, geometric center localization, and ADS-YOLOv8s-based defect detection into a unified visual perception framework for fully flooded pipeline inspection robots, enabling adaptation to various pipeline scenarios for detection tasks.

As a crucial component of wind power generation systems, wind turbines must operate safely to prevent sudden failures. This requires effective identification of abnormal operational states. In this study, we propose a novel approach for categorizing abnormal states based on operational data from wind turbines, leveraging an improved density-based spatial clustering of applications with noise (DBSCAN) algorithm in conjunction with random forests. We establish a wind turbine performance model as a benchmark, employing a DBSCAN-bidirectional long short-term memory network (DBSCAN-BiLSTM) framework for anomaly detection. The random forest algorithm is then applied to accurately identify abnormal data points. Furthermore, we implement real-time adjustments to operational state thresholds based on identified anomalies, facilitating precise classification of abnormal operational data. A case study is presented, detailing steps including abnormal data cleaning, performance model construction, and abnormal state classification to validate our approach. Our results demonstrate that the DBSCAN-BiLSTM method significantly reduces the root mean square error (RMSE) by 41.5% and 49.7% compared to traditional LSTM and convolutional neural network (CNN) algorithms, respectively. This research provides an effective solution for identifying abnormal states in supervisory control and data acquisition (SCADA) data of wind turbines, which is vital for fault detection and maintenance in the wind power industry.

Offshore wind turbine tower acceleration data are often affected by significant noise and exhibit non-linear, non-stationary characteristics. Traditional Empirical Mode Decomposition (EMD) methods are prone to modal aliasing and the introduction of additional noise, which complicates modal analysis and effective vibration feature extraction. Moreover, the challenge of selecting valid signals using correlation coefficients further exacerbates these issues. This paper proposes a noise reduction method based on a parameter-adaptive time-varying filtering EMD (TVF-EMD) technique, combined with energy entropy increment and instantaneous frequency change rate dual thresholding. First, the Grey Wolf Optimization (GWO) algorithm is used to optimize the parameters of the TVF-EMD method, with the minimum sample entropy serving as the objective function. The vibration signal of the wind turbine tower is then decomposed using TVF-EMD to obtain a set of components. Energy entropy increments and instantaneous frequency change rates for all components are calculated, and a dual thresholding criterion is applied. Specifically, 10% of the total energy entropy increment and 10% of the maximum instantaneous frequency change rate are used to select valid components. The denoised signal is obtained by combining these selected components. Compared to wavelet transform and other denoising methods, the proposed method achieves a signal-to-noise ratio (SNR) improvement of over 6.68% and reduces the root mean square error (RMSE) by more than 12.2% across varying noise levels. Experimental validation demonstrates that the method effectively separates noise from the blade passing frequency, thereby facilitating subsequent modal analysis and vibration feature extraction.

A toroidal propeller distributes a generated vortex on the whole propeller blade, effectively preventing vortex leakage from the tip of the propeller, which causes the vortex to dissipate quickly in the water and reduces the propagation of hydrodynamic noise. However, owing to their complex structure, modeling toroidal propellers via conventional propeller modeling methods is difficult. In this paper, a low-noise toroidal propeller shape parameterization method is proposed, and the hydrodynamic performance and non cavitation noise are compared with those of a DTMB P4119 propeller. The results show that the modeled toroidal propeller has more thrust than the DTMB P4119 propeller under the same rotational speed conditions. Under the same thrust conditions, the hydrodynamic noise of the modeled toroidal propeller is lower than that of the DTMB P4119 propeller; the total noise SPL at the radial measurement point is reduced by approximately 4 dB, and the total noise SPL at the axial measurement point is reduced by approximately 6 dB compared with that of the DTMB P4119 propeller. This novel toroidal propeller shape parameterization method provides a theoretical basis for suppressing the hydrodynamic noise of ships and underwater vehicles.

This study presents a comprehensive investigation into multi-directional fatigue damage characteristics of fixed offshore wind turbine tower roots through comparative analysis using FAST (3.5.0) and Bladed (4.3) software platforms. The research methodology encompasses three principal phases: First, a stochastic wind field model was developed through statistical analysis of historical wind speed measurements, achieving superior correlation (R2 = 0.983) in goodness-of-fit tests. Subsequently, the rain flow counting technique was employed to characterize equivalent cyclic load spectra. Building upon these foundations, an integrated predictive fatigue life evaluation framework was formulated by synergistically combining S–N curve principles with Palmgren–Miner’s linear cumulative damage theory. The methodology was further validated through cross-platform verification with Bladed software, revealing only a 7.4% deviation in predicted fatigue lives between the two computational models, confirming the technical feasibility of the proposed simplified model.

To expand the detectable range of the linear array in the near-field acoustic emission source localization process to improve the localization accuracy and efficiency, this paper proposes a method of near-field acoustic emission source localization based on orthogonal matching pursuit under nonuniform linear array. Firstly, according to the propagation characteristics of acoustic emission signals, a narrowband signal decomposition method for broadband acoustic emission signals is proposed. Then, a near-field source dimension reduction method based on fourth-order cumulants under nonuniform array is proposed to realize the separation and estimation of source angle and distance parameters. Subsequently, an acoustic emission source localization method based on orthogonal matching pursuit is proposed to maximize the source localization performance of the nonuniform linear array. Finally, the performance comparison between three different nonuniform linear arrays and different methods is carried out. Meantime, the finite element simulation and acoustic emission localization experiments are used to analyze the localization accuracy law and verify the effectiveness of the method.

The ultrasonic guided wave-based method for multi-damage localization has been widely proposed. However, the precision of this method is directly correlated with both the quantity of sensors employed and the intricacy of the implementation process. This relationship poses a challenge in striking a balance between the accuracy and efficiency. To improve the computational efficiency under the premise of ensuring the accuracy of multi-damage localization, this paper proposes a near-field weighted subspace fitting algorithm based on niche-particle swarm optimization. Firstly, the fitting relationship between the signal subspace of the diffraction wave and the array steering vector is established under the uniform linear array. Secondly, a multidimensional solution space search algorithm based on niche-particle swarm optimization is proposed to improve the search efficiency of damage. Finally, the algorithm is verified by performance comparison, finite element simulation and experiment. The results show that compared with the same type of method, the algorithm improves the computational efficiency by nearly threefold under the identifiable multi-damage conditions. Additionally, the angle error is 1 - 6 degrees, and the distance error is 1 - 20 mm.

In recent years, ultra-large-scale offshore wind turbines have attracted widespread attention. However, accurately evaluating the motion responses of offshore wind turbines under extreme conditions, especially for semisubmersible floating off-shore wind turbines, is often challenging. In order to assess the operational behavior of wind turbines under wind and wave loads, this paper adopted a numerical analysis method to solve the motion responses under extreme conditions. It specifically examines the motion responses of the IEA 15 MW wind turbine in terms of surge, heave, and pitch direction, focusing on environmental loads that occur once every 50 years. The results show that the wind turbine can still operate normally under the Ultimate condition. However, the average amplitude increased by 7% in the pitch direction and decreased by 4% in the heave direction compared to the rated condition. Under extreme conditions (occurring once every 50 years), with the wind turbine parked, the average amplitude in the surge direction reduced by 33%, while the average amplitude in the pitch direction reduced by 106%. Thus, it is essential to pitch the blades and brake the generator in extreme environmental conditions to ensure the safety of the wind turbine.

This paper proposes a topology optimization method considering fatigue constraints for the jacket support structure of offshore wind turbine. Jacket support structure is an important component and supports the whole wind turbine weight under the dynamic environmental loads, which would induce serious fatigue problems. To obtain the optimal structural layout, the topology optimization method considering fatigue constrained is introduced for the jacket support structure under the dynamic load. The fatigue lifetime is constrained base on the Miner's cumulative damage rule. The design sensitivity of the objective and constraint functions are evaluated analytically, while the equivalent static load approach is applied. Some typical topology optimization problems are introduced to verify the effectiveness of the proposed method, and this method can obtain the appropriate layouts with stable convergence. A case study on the OC4 reference jacket indicates that the method can achieve a reasonable layout structure under the dynamic load, which meets the fatigue constraint. And the final design is reconstructed based on the topology optimization layout. Finally, a lighter jacket structure has good performance on the ultimate bearing capacity, eigenfrequency, and buckling, which is suited for long operation life.

轴箱轴承是高速动车组(EMU)走行部的关键部件,往往处于高速甚至超高速的运行状态,其轴承的磨损是一个非常突出的问题.建立转向架轴承的磨损模型,模拟在轨道激励载荷因素作用下滚动轴承的磨损状态,进行轴承的磨损损伤分析具有重要意义.针对动车组转向架轴箱轴承磨损监测难、分析难的问题,基于数值模拟方法,对轴箱轴承的磨损损伤分析方法进行了探究,为降低有限元模型的分析计算时间成本,提出了一种轴承切片式半解耦损伤分析方法.首先,将轴承三维模型径向等距切分成有限数量的二维切片,对二维轴承切片模型进行了磨损分析;然后,基于任意拉格朗日-欧拉(ALE)自适应网格偏移技术以及UMESHMOTION子程序,提出了轴承模型磨损有限元分析方法(流程);最后,利用高斯过程非线性拟合完成了三维轴承磨损状态复现,得到了轴承内圈的磨损分布规律,并对该模型的高效性与收敛性进行了对比分析.研究结果表明:利用二维切片拟合磨损分析方法的计算效率获得了大幅提高,同时该方法的准确性也较为可靠,与传统方法相比,其误差在10%以内;另外,轴承内圈磨损较为严重的区域出现在滚子与内滚道接触的边缘位置,而接触区域中心位置的磨损程度与边缘相比较轻.

在中国海洋渔业中,网衣清洗机器人对网箱养殖有害物附着的清洗具有重要作用.为了提高网衣清洗机器人在海流干扰下工作的稳定性,本文根据网衣清洗机器人的工作特点,提出了采用串级-前馈比例积分微分(PID)控制方法对海流干扰下的网衣清洗机器人进行控制,通过仿真实验证明该方法是可行的.为进一步提高控制效果,在串级-前馈PID控制的基础上加入模糊控制,仿真结果表明,改进后的控制器一定程度上减少了网衣清洗机器人运动稳定时间和超调量.

This paper proposes a digital twin method based on multi-source crack growth prediction data fusion. In this method, two different prediction methods based on theoretical model correction and machine learning model correction are constructed, which avoids the inapplicability of a single method in practical applications. Meantime, based on the consistency retention method corresponding to each model, the influence of uncertainty factors on crack growth prediction is gradually reduced by inputting crack detection data. Subsequently, by fusing the historical data and prediction data, the crack growth prediction result with the smallest deviation and higher reliability is output. The verification results show that the digital twin model proposed in this paper can effectively reduce the influence of uncertainty factors on crack growth prediction and realize the dynamic prediction of crack growth.

自主学习和创新能力的培养一直是我国高等教育的重点,课程教学是其中重要环节之一.探讨在机械类研究生课程教学中存在的问题,提出以项目设计为导向的课程教学模式,并以振动分析与动态测试课程为例,进行具体的项目设计和教学实施,利用具体的项目设计为导向,与课程理论知识内容融会贯通.在实际教学中该教学形式取得良好的教学成果,能够激发学生学习创新的积极性和主动性,实现预期效果.

At present, ML has become an effective method to solve the prediction problem of fatigue crack growth. To reduce the inaccurate prediction caused by uncertain factors in crack growth, this paper proposes a fatigue crack growth prediction method based on the ML model correction. This method improves the accuracy of crack growth prediction by using real crack data to correct the ML model. In the research process, the prediction performance of the three ML methods is compared, and the CGR-ML model for crack growth is established. Subsequently, dynamic correction strategy for the CGR-ML model is proposed while selecting crack detection points by using the nonlinear crack length selection method. Finally, the effectiveness of the method is verified by the central crack growth and the crack growth experiment under mixed-mode multi-step loading. It can be seen from the comparison with the previously proposed fatigue crack growth prediction method based on the theoretical model correction that the method proposed in this paper can achieve a better prediction effect.

Jack-up platforms are widely used for the offshore oil and gas exploration, constructions and offshore wind farms. The jack-up platform legs are import component, which not only support the whole weight but also withstands various external loads, such as wind, current and wave load. A topology optimization algorithm-BESO (bi-directional evolutionary structural optimization) is proposed in this paper for the jack-up leg optimization. Based on the sensitivity analysis, the optimization strategy with sensitivity filtering and updated have been applied. And a simple numerical example is used to verify the proposed optimization method. A case jack-up platform is introduced to implement the leg structure optimization. After a series of iterations, a new leg structure is obtained. And the ultimate analysis, Eigen analysis and buckling of optimized leg structure optimized structure are all discussed compared with the traditional structures (K-type, the reverse(rev) K-type and X-type). To further evaluate the performance of the optimized leg structures, an experiment test is designed in the wave tank considering the current and wave load. Also, the optimized leg structure and the traditional ones are all manufactured to perform the experiment. The results show that the optimized leg structure has better ultimate bearing capacity, Eigen analysis and dynamic response with lighter weight under the same environmental loads. It is indicated that the BESO topology method is superior in finding best structure of the jack-up platform leg.

针对近海风机在风、波浪和地震3种载荷作用下的振动控制问题,提出了基于磁流变弹性体(magnetorheological elastomer,MRE)调谐质量阻尼器(tuned mass damper,TMD)的海上风机半主动控制方法.首先,介绍了MR E的变刚度磁致力学特性及面向海上减振需求的MR E-T MD结构设计原理;其次,建立了海上风机-T MD动力学模型,计算了风、波浪及地震激励载荷;再次,应用半主动控制算法跟踪识别风机塔筒顶端响应的频率,实时调节MRE-TMD的刚度,进而对海上风机进行振动控制.通过分析导管架式近海风机在多种载荷作用下的动力响应可知,采用基于MRE-TMD的控制方法能够有效衰减导管架式海上风机在多种载荷作用下的振动响应.与被动TMD相比,MRE-TMD具有较好的减振效果,为海上风机振动控制提供了一种新的解决思路.

With the development of artificial intelligence, big data, Internet of Things, and other technologies, digital twin has gained great attention and become a current research topic. Using digital twin technology, the digital twin model can be constructed in the cyber space that is fully equivalent to the physical entity. It is always consistent with the physical entity in the operation process, which greatly improves the dynamic perception and prediction ability of the real world. After the development in recent years, digital twin has gradually changed from the initial concept discussion to the study of model framework and implementation method. However, because the research objects in different industries have great differences in their own composition, service conditions, and application scenarios, they have personalized characteristics in modeling strategies and usage methods. Therefore, based on different industries, this paper reviews the current articles on digital twins and distinguishes the focus of digital twin modeling research; subsequently, the relevant supporting techniques and methods are summarized according to their different importance for digital twin modeling. Based on the review in this paper, future researchers can conduct targeted research on digital twin technology in term of the characteristics of the objects in their industry.

Jacket support structure is the main support form in medium water depth for bottom fixed offshore wind turbine. Topology optimization is presented in this paper for an efficient method to optimize the jacket structure by altering the structural layout to improve its design, minimize weight and ultimately reduce the cost. Together with size and shape optimization, an innovative jacket structure is obtained with constraints on strength, stiffness, stability, ultimate strength, and natural frequency. The optimization has been solved and the overall mass of the optimized structure decreases by 38.24% compared with the primary structure. Moreover, ultimate limit state analysis, eigenanalysis and buckling are used to evaluate the optimized structure. The results indicate that the proposed optimized method is an effective way to reduce the weight and stress concentration. It also shows that topology optimization provides useful insights for the conceptual design phase and provides a better starting point for the further size and shape optimization.