Intelligent Operation Research Articles

A deep learning method for pointer meter reading recognition in inspection robots at refrigeration stations

Abstract To improve the intelligent operation and maintenance of large public building refrigeration stations, various automatic reading methods for pointer meters used in inspection robots have been developed. However, most methods exhibit low detection accuracy for pointer meters in refrigeration stations and poor segmentation of pointers and scale lines, ultimately leading to inaccurate readings. To address the challenges posed by the complex environments of refrigeration stations, this study proposes a deep learning-based image recognition method for precise pointer meter readings. Firstly, the pointer meters in the environment are detected and located by the YOLOV8-CBAM-Wise-IoU model. Secondly, an improved Fast Semantic Segmentation Network (Fast-SCNN) is used for image segmentation of the pointers and scale lines of the pointer meters, and a perspective transformation method is employed to correct images of tilted meters. Then, the image is transformed into a rectangular image using polar coordinate transformation. Finally, the accurate reading is calculated using the distance method, based on the relative distances between the pointers and scale lines. This research conducted experimental tests using an inspection robot in the actual environment of a large public building's refrigeration station. The results demonstrate that the proposed method for pointer meter reading recognition achieves a fiducial error within 0.41%, which can realize the inspection task of the pointer meter in refrigeration stations.

BIM Driving Innovation in Construction Industry: Advancements in Building Lifecycle Management and Facility Management

With the rapid development of the traditional construction industry, China's construction market is becoming saturated and overcapacity, and the global macro economy is facing uncertainty. Negatively impacted by global events and the climate crisis, the willingness of traditional architectural firms to invest in real estate, infrastructure and other projects needs to be higher. The enhancement of digital information technology has brought greater freedom and openness, and traditional architects are looking to BIM technology to stimulate creativity and build more dynamic buildings. This paper analyses the development and application of BIM technology from the perspective of the BIM+ intelligent management model. It discusses the importance of upgrading the traditional construction industry, especially in building life cycle management. Through BIM+ intelligent management, the traditional construction industry can handle the information in all project stages more efficiently and achieve the goal of resource integration. At the same time, the combination of BIM and facility management can realise intelligent operation and monitoring of building projects. This paper reviews related research and looks forward to the future research direction of combining BIM and building facility management.

Construction and Application of a Private 5G Standalone Medical Network in a Smart Health Environment: Exploratory Practice From China.

To date, the differentiated requirements for network performance in various health care service scenarios-within, outside, and between hospitals-remain a key challenge that restricts the development and implementation of digital medical services. This study aims to construct and implement a private 5G (the 5th generation mobile communication technology) standalone (SA) medical network in a smart health environment to meet the diverse needs of various medical services. Based on an analysis of network differentiation requirements in medical applications, the system architecture and functional positioning of the proposed private 5G SA medical network are designed and implemented. The system architecture includes the development of exclusive and preferential channels for medical use, as well as an ordinary user channel. A 3-layer network function architecture is designed, encompassing resource, control, and intelligent operation layers to facilitate management arrangements and provide network open services. Core technologies, including edge cloud collaboration; service awareness; and slicing of access, bearer, and core networks, are employed in the construction and application of the 5G SA network. The construction of the private 5G SA medical network primarily involves system architecture, standards, and security measures. The system, featuring exclusive, preferential, and common channels, supports a variety of medical applications. Relevant standards are adhered to in order to ensure the interaction and sharing of medical service information. Security is achieved through mechanisms such as authentication, abnormal behavior analysis, and dynamic access control. Three typical medical applications that rely on the 5G network in intrahospital, interhospital, and out-of-hospital scenarios-namely, mobile ward rounds, remote first aid, and remote ultrasound-were conducted. Testing of the 5G-enabled mobile ward rounds showed an average download rate of 790 Mbps and an average upload rate of 91 Mbps. Compared with 4G, the 5G network more effectively meets the diverse requirements of various business applications in prehospital emergency scenarios. For remote ultrasound, the average downlink rate of the 5G network is 4.82 Mbps, and the average uplink rate is 2 Mbps, with an average fluctuation of approximately 8 ms. The bandwidth, performance, and delay of the 5G SA network were also examined and confirmed to be effective. The proposed 5G SA medical network demonstrates strong performance in typical medical applications. Its construction and application could lead to the development of new medical service models and provide valuable references for the further advancement and implementation of 5G networks in other industries, both in China and globally.

Segmentation of tunnel water leakage based on a lightweight DeepLabV3+ model

Abstract The accurate and efficient detection of water leakage with complex backgrounds is crucial for the safety of metro operations. A lightweight segmentation method for metro tunnel water leakage based on transfer learning is proposed. Firstly, this is based on the Deeplabv3+ model and adopts MobileNetv3-Large as the backbone feature extraction network, which significantly reduces the network parameters and improves the detection speed; secondly, it incorporates the efficient channel attention mechanism, which enables the model to adaptively adjust the weights of the channel features and capture the inter-channel relationships in the image, which significantly improves the model’s ability for feature extraction ability; furthermore, for the problem of severe imbalance between positive and negative samples in the dataset, the recognition accuracy of complex samples is increased by optimizing the loss function; finally, the training method of transfer learning is utilized to solve the problem of scarcity of water leakage dataset, and to improve the model’s accuracy and generalization ability. The results show that the model has more significant detection accuracy and segmentation speed advantages than today’s mainstream semantic segmentation model. With strong generalization ability in complex environments (e.g., low illumination and multiple obstructions), the model can be used for intelligent operation and maintenance in metro tunnel projects.

Development of an Intelligent Coal Production and Operation Platform Based on a Real-Time Data Warehouse and AI Model

Smart mining solutions currently suffer from inadequate big data support and insufficient AI applications. The main reason for these limitations is the absence of a comprehensive industrial internet cloud platform tailored for the coal industry, which restricts resource integration. This paper presents the development of an innovative platform designed to enhance safety, operational efficiency, and automation in fully mechanized coal mining in China. This platform integrates cloud edge computing, real-time data processing, and AI-driven analytics to improve decision-making and maintenance strategies. Several AI models have been developed for the proactive maintenance of comprehensive mining face equipment, including early warnings for periodic weighting and the detection of common faults such as those in the shearer, hydraulic support, and conveyor. The platform leverages large-scale knowledge graph models and Graph Retrieval-Augmented Generation (GraphRAG) technology to build structured knowledge graphs. This facilitates intelligent Q&A capabilities and precise fault diagnosis, thereby enhancing system responsiveness and improving the accuracy of fault resolution. The practical process of implementing such a platform primarily based on open-source components is summarized in this paper.

Anomaly detection and confidence interval‐based replacement in decay state coefficient of ship power system

AbstractThe anomaly detection and predictive replacement of the degradation decay state coefficient (Desc) of ship power system (SPS) are crucial for ensuring their operational safety and maintenance efficiency. This study introduces the YC3Model, a model based on a dynamic triple sliding window mechanism, and Gaussian process regression) to address this challenge. It combines the temporal variation characteristics of the decay state coefficient's original data, first‐order, and second‐order differential data in both normal and abnormal trend intervals. The model calculates three local statistical measures within each sliding window and employs the Z‐score method for anomaly detection. The combination of three sliding windows reduces false positives and negatives, enhancing the precision of anomaly detection. For detected anomalies, Gaussian process regression is used for prediction and replacement, providing confidence intervals to increase the reliability of the predicted values. Experimental results demonstrate that the YC3Model exhibits superior anomaly detection accuracy and adaptability in the degradation process of SPS, surpassing traditional methods across a range of evaluation metrics. This confirms the potential of YC3Model in health monitoring and predictive maintenance of SPS, offering reliable data input for the intelligent operation and maintenance (IO&M) of SPS.

The intelligent operation and maintenance method and experimental research of CNC machine tool bearings

As the core components of high-precision rotary machine tools, machine tool bearings are prone to failure or damage, and their running state directly affects the safety and reliability of the entire equipment. Therefore, the paper proposes an intelligent operation and health management method for machine tool bearings based on Bayes-VMD-KPCA, Bayes-CNN-SVM, and health status evaluation index. Aiming at the complex working environment and high nonlinear degree of fault features of machine tool bearings, the Bayes-VMD-KPCA algorithm is used to perform variational mode decomposition of the original signal of the bearing fault, which avoids the influence of background noise and solves the problem of high dimension of multi-domain fault feature set. Second, the CNN-SVM fault diagnosis model uses a convolutional neural network (CNN) to deeply mine the extracted multi-domain fault features, and a support vector machine (SVM) can be used as a fault classifier to complete the diagnosis of various fault types of machine tool bearings. At the same time, the Bayes algorithm is introduced to tune the hyperparameters in the CNN-SVM model to further improve the prediction performance of the CNC machine tool-bearing fault diagnosis model. The results show that the intelligent operation and maintenance and health management method of CNC machine tool bearings proposed in this paper can effectively complete the fault type diagnosis and health status assessment of bearings, and its accuracy index reaches 99.33%. Compared with the single SVM model, Bayes-CNN model, Bayes-SVM model, and CNN-SVM model, the health effect is evaluated. The accuracy is increased by 6.33%, 4.33%, 2.66%, and 1%, respectively. It can be seen that the proposed method can more effectively realize the monitoring and health management of bearing fault.

Research on fault diagnosis for three-phase rectifier device in direct current transmission system based on continuous wavelet transform and vision transformer

Abstract As a core component of photovoltaic power generation systems, the three-phase rectifier device plays a crucial role, and its failure can potentially reduce energy conversion efficiency and output quality. Presently, the performance of fault time-domain signal diagnosis methods based on three-phase rectifier circuits is challenging to enhance. This paper proposes a novel fault detection method for three-phase rectifier devices based on Vision Transformer, referred to as CWT-ViT, to address this issue. This method transforms time-domain fault signals into images through Continuous Wavelet Transform (CWT), subsequently inputting these images into a Vision Transformer model. Relying on its powerful self-attention mechanism and fully connected layers, it realizes the extraction and learning of rectifier device image features. A Simulink simulation model of a three-phase bridge controllable rectifier circuit is established for fault injection to collect fault signals. Fault diagnosis experiments demonstrate that the proposed diagnostic method achieves a prediction accuracy of 98.6%, maintaining a relatively high precision level. In comparison to four excellent classification models currently available: AlexNet, RepVGG, ResNet, and GoogLeNet, the proposed method demonstrates superior diagnostic performance. Additionally, this paper conducts ablation experiments to meticulously analyze the impact of each module in the fault diagnosis process. This research achieves more precise and efficient fault diagnosis in photovoltaic power generation systems, thereby reducing downtime and maintenance costs for actual equipment and enhancing the stability of photovoltaic power generation systems. This research provides an innovative, intelligent solution for the intelligent operations and maintenance of photovoltaic power generation.

Design and stability performance optimization of a novel hybrid inspection robot walking device for smart grid applications

Inspecting in high-altitude windy environments requires maintaining a stable attitude to avoid significant oscillations, which is a crucial factor in advancing the use of hybrid power transmission line inspection robots (PTLIRs) in real-world scenarios for the electrical power and energy industry. We developed and optimized a novel walking device to improve the safety of hybrid PTLIRs operating in strong wind conditions. First, we investigate the operational conditions of transmission line inspection and design a walking device capable of transporting a flight device to be suspended under the line for inspection purposes. Second, a force equilibrium analysis is performed to explore the relationship between the key structural parameters of the walking device and the wind-induced deflection of the robot. A multiobjective optimization model is formulated to minimize the maximum deflection angle of the robot along three axes under wind loads. Finally, an optimization algorithm, using an improved non-dominated sorting particle swarm optimization (NSPSO) approach, is proposed to solve the model with good global search ability and fast convergence speed. A test platform is constructed to collect the wind deflection angle of a self-developed hybrid PTLIR under the transmission line. The test results demonstrate that the walking device can carry 30 kg to move along the transmission line. The average deflection angle of the robot around the Z axis is the highest under the influence of wind force level 7 (wind speed of 15.5 m/s) perpendicular to the windward surface of the robot. After optimization, the maximum deflection angle is limited to 10.38°, aligning within safety thresholds. The optimization algorithm effectively reduces the maximum deflection angles around the X, Z, and Y axes by 33.19 %, 39.21 %, and 13.23 %, with minimal average errors of 2.40 %, 3.87 %, and 2.13 %, respectively. The research in this paper can solve the bottleneck problem of robot practicality, which has important theoretical significance and practical application value for the safety, stability, and intelligent operation and maintenance management of power transmission lines.

LightGBM integration with modified data balancing and whale optimization algorithm for rock mass classification

The accurate prediction of uneven rock mass classes is crucial for intelligent operation in tunnel-boring machine (TBM) tunneling. However, the classification of rock masses presents significant challenges due to the variability and complexity of geological conditions. To address these challenges, this study introduces an innovative predictive model combining the improved EWOA (IEWOA) and the light gradient boosting machine (LightGBM). The proposed IEWOA algorithm incorporates a novel parameter l for more effective position updates during the exploration stage and utilizes sine functions during the exploitation stage to optimize the search process. Additionally, the model integrates a minority class technique enhanced with a random walk strategy (MCT-RW) to extend the boundaries of minority classes, such as Classes II, IV, and V. This approach significantly improves the recall and F1-score for these rock mass classes. The proposed methodology was rigorously evaluated against other predictive algorithms, demonstrating superior performance with an accuracy of 94.74%. This innovative model not only enhances the accuracy of rock mass classification but also contributes significantly to the intelligent and efficient construction of TBM tunnels, providing a robust solution to one of the key challenges in underground engineering.

The Use of eXplainable Artificial Intelligence and Machine Learning Operation Principles to Support the Continuous Development of Machine Learning-Based Solutions in Fault Detection and Identification

Machine learning (ML) revolutionized traditional machine fault detection and identification (FDI), as complex-structured models with well-designed unsupervised learning strategies can detect abnormal patterns from abundant data, which significantly reduces the total cost of ownership. However, their opaqueness raised human concern and intrigued the eXplainable artificial intelligence (XAI) concept. Furthermore, the development of ML-based FDI models can be improved fundamentally with machine learning operations (MLOps) guidelines, enhancing reproducibility and operational quality. This study proposes a framework for the continuous development of ML-based FDI solutions, which contains a general structure to simultaneously visualize and check the performance of the ML model while directing the resource-efficient development process. A use case is conducted on sensor data of a hydraulic system with a simple long short-term memory (LSTM) network. Proposed XAI principles and tools supported the model engineering and monitoring, while additional system optimization can be made regarding input data preparation, feature selection, and model usage. Suggested MLOps principles help developers create a minimum viable solution and involve it in a continuous improvement loop. The promising result motivates further adoption of XAI and MLOps while endorsing the generalization of modern ML-based FDI applications with the HITL concept.

Application of digital twin technology in monitoring system of pump turbine

The advent of advanced productive forces has catalyzed the ongoing evolution of intelligent and digital technologies within the pump-turbine sector. The integration of digital technologies has not only enhanced production efficiency but also facilitated intelligent operation management, offering novel solutions for pump-turbine design and operation, thereby accelerating the digital transformation in fluid machinery. This study first established the theoretical framework of the digital twin system for pump-turbines, followed by the development of a mathematical model and the construction of a twin virtual model utilizing Proper Orthogonal Decomposition (POD) reduction theory. The model’s accuracy was validated through real-time pressure pulsation data, and exploratory investigations into cavitation prediction were subsequently carried out. To enable the visualization of the digital twin system, this study incorporated Open3D (Three-dimensional computer graphics tools) point cloud technology, effectively rendering the state cloud map of the pump-turbine. Finally, by synthesizing the proposed theories and technologies, the digital twin system’s visualization and monitoring for pump-turbines were successfully implemented through the Unity3D simulation platform. This study offers novel insights into intelligent monitoring and prediction of pump-turbines, holding significant implications for the modernization and intelligent advancement of the hydropower industry.

DESIGN and be SMART: Eleven engineering challenges to achieve sustainable air transportation under safety assurance in the year 2050

The aviation industry faces various challenges in meeting long-term sustainability goals amidst surging demand for air travel and growing environmental concerns of the general public. The year 2050 is set as an ambitious goal for net zero emissions, a substantial reduction in carbon dioxide emissions per passenger kilometer flown, major improvements in aircraft energy efficiency, and a development towards autonomous, intelligent operations. This review explores the pivotal role of advancements in engineering for achieving sustainability in aviation. Through a comprehensive review of existing literature and case studies, our work highlights how innovations in all aspects of aircraft engineering coupled with operations-related technologies, offer promising solutions to mitigate environmental impact, enhance efficiency, and ensure long-term sustainability in aviation operations. To discuss the necessary advances, we promote the so-called ‘DESIGN and be SMART’ framework, consisting of eleven complementary engineering challenges towards reaching sustainability. To address the high safety levels reached in air transportation, our DESIGN and be SMART framework also addresses the safety assurance challenge that is overarching each of the eleven engineering challenges. We believe that through an orchestrated integration of hardware advancements with innovative software solutions, and novel safety assurance methods, the aviation industry can realize synergistic benefits that drive sustainable growth of air transportation. Our review contributes to such an orchestration by describing the status quo and research challenges ahead.

Guest Editorial Special Section on Advancing Power Electronics Reliability: Components, Systems, and Intelligent Operation

Guest Editorial Special Section on Advancing Power Electronics Reliability: Components, Systems, and Intelligent Operation

Unsung Hero: Artificial Lift Boosts Production Without Breaking the Bank

_ In the quest to cost-effectively boost production, operators are experimenting with artificial lift methods and how digital tech can boost its performance. Artificial lift experts are increasingly adding artificial intelligence (AI), machine learning (ML), and autonomous operations into their workflows because these data-driven technologies are demonstrating they can generate uplift when they are thoughtfully developed and deployed. Operators are also looking at the performance of high-pressure gas lift (HPGL) wells compared to wells using electrical submersible pumps (ESPs) for artificial lift. Finally, operators sometimes have to decide whether to risk replacing equipment that seems to be functioning, even if at subpar levels, in a bid to increase production. Digital Boost Optimized gas lift workflows and autonomously implemented ESP set points are helping ExxonMobil and Vital Energy produce more oil. Gas lift is a popular artificial lift method partly because the system is simple, robust, and inexpensive to operate, but optimizing it has historically been time-consuming. SPE 219553 outlines ExxonMobil’s approach to automating gas lift optimization for wells in the Permian Basin. The resulting workflow, which relies on AI and ML, generated a 2.2% uplift in production at the more than 1,300 wells where it was deployed, without requiring changes to surface or downhole equipment or generating additional work for personnel. Derek Burmaster, production optimization engineer for ExxonMobil Upstream Integrated Solutions, said during SPE’s Artificial Lift Conference and Exhibition (ALCE) in August that one of the big benefits of applying the closed-loop iterative well-by-well gas lift optimization workflow is that it removes the need for an engineer or specialist to have to go into the field repeatedly to change injection rates and analyze the well’s data. “For each well, that might take a few hours, at least,” he said. “We’ve done this on hundreds and hundreds of wells, and it happens every month without anybody, any manpower, or anything. So that’s one of the huge benefits of doing this in a data-driven manner.” Gas lift is so robust that it can be challenging to diagnose if the system is operating in a suboptimal state, the paper noted. Changes in a well’s water cut or productivity can cause a previously optimized system to lose efficiency, which may not otherwise be known for years. ExxonMobil was seeking a method to optimize gas lift that would continuously monitor the state of the system and respond immediately. The resulting closed-loop software iteratively performs multirate tests, analyzes the results, and remotely implements set-point changes to optimize the wells. “The optimization process is most effective in an iterative form, in which each multirate test produces a model that suggests a given change,” the paper stated. While each model produces its own optimal injection rate, subsequent models may point to the same injection rate, and thus changes are not always made, Burmaster noted. Even if an initially suggested set point is not optimal, it will “push the well in the right direction,” which improves the starting point for the next test and helps the well reach an optimal set point over the course of a few iterations, according to the paper. The workflow was tested in two different stages, with the first 3-month stage requiring manual approval of the recommended set points.

Real-time identification of precursors in commercial aviation using multiple-instance learning

This research pioneers the application of precursor concepts to preemptively identify and prevent aviation safety incidents using Machine Learning (ML). Airlines and governing organizations, such as the Federal Aviation Administration (FAA) in the United States, have been trying to prevent safety incidents during routine operations. However, this task is challenging due to the lack of timestep-wise event annotation in flights and the complexity involved in the timely identification of incidents prior to their occurrence. To address these issues, we propose a real-time precursor identification methodology combining Multiple-Instance Learning (MIL) and feature-based Knowledge Distillation (KD) learning. Our two-stage approach, involving deep MIL for labeling and a KD-based model for real-time warnings, demonstrates state-of-the-art performance and a time delay of 2.99ms using a dataset of 23,549 real flights. Further experiments using t-distributed Stochastic Neighbor Embedding (t-SNE) and occlusion method confirm our model’s transparency, enabling the generation of reliable quantitative precursor scores and facilitating reasoning about the causes of safety incidents at the parameter level. Additionally, statistical analysis of precursors reveals varying evolution times for different safety events, which indicates that pilots have at least 8 s to react after receiving a warning. In conclusion, our research provides a theoretical foundation and technical support for the next generation of online risk warning systems, enhancing flight safety and offering a pathway towards more intelligent and secure flight operations.

A human-centered intelligent platform for holographic control and management of 3D printers

Under the wave of emerging human-centric concepts such as Industry 5.0 and Human-Centered Intelligent Manufacturing, Mixed Reality (MR) has emerged as a pivotal enabling technology with profound impacts on human–machine interaction. Simultaneously, the affordability of Fused Deposition Modeling (FDM) 3D printers has expanded their accessibility beyond industrial applications to homes and public spaces. However, the operational and management complexities of different 3D printers and their varying human–machine interfaces (HMI) present challenges for users, especially beginners. In this paper, we presented a human-centered intelligent platform for FDM 3D printer control and management enabled by Mixed Reality. A flexible system architecture is developed in the paper with two essential communication protocols, MQTT and WebSocket, for stable data transfer. In addition, a task-oriented HMI is explored on the MR headset to simplify the operation procedure with little training effort and allow parallel supervision of multiple machines. By leveraging the advantages of our architecture, real-time monitoring, failure detection, and hologram visualization were able to be integrated into our platform for intelligent operation and management. Importantly, the flexibility of the architecture via a holographic interface enables our platform to easily expand to various machining operations in the future.

Wavelet-based spatiotemporal sparse quaternion dictionary learning for reconstruction of multi-channel vibration data

Accurate reconstruction of multi-source data information plays an essential role in remote intelligent operation, and prognostic and health maintenance (RIO-PHM) of industrial systems. However, traditional signal reconstruction methods such as compressed sensing fail to capture the spatio-temporal characteristics and coupling nature of the multi-source or multi-channel data. To alleviate this bottleneck, in this paper, a new wavelet-based spatiotemporal sparse quaternion dictionary learning (WSTS-QDL) algorithm is formulated for the reconstruction of multi-channel vibration data for the first time. Specifically, the wavelet-based spatiotemporal sparse low-rank matrix (SLRM) algorithm is elaborated and the sparse coefficient matrix of the multi-channel signals can be estimated using the split Bregman iteration (SBI) technique. Then, quaternion-based dictionary learning is introduced for multi-channel data reconstruction according to the updated sparse coefficient matrix during quaternion dictionary learning. Eventually, two equipmental scenarios including run-to-failure data of rolling bearing in bearing test bench and vibration signal of the failured bearing in corn thresher, are utilized for verification. The experimental results demonstrate that the highest recovery accuracy performance with scoring function and the lowest errors are achieved by the proposed method in contrast with the state-of-the-art benchmarks such as orthogonal matching pursuit (OMP) method and spatiotemporal sparse Bayesian learning (SSBL), and the waveform of phase space trajectory and points cloud distribution of the Poincare section in the original signal are similar/same to that obtained by the proposed method.

A novel wheel wear indicator in regard to wheel-rail contact parameters and vehicle hunting stability

It is well known that the abnormal equivalent conicity caused by wheel and rail wear is the main root of hunting instability, but it is still difficult to trace the underlying factor when the dynamic instability occurs, that is, wheel wear induced, rail wear induced or both. To face this challenge problem, this paper constructed a novel wheel wear indicator based on the moved wear area difference. The long-term tracking test data of different wheel profiles were used to analyze the correlation between the wear indicators and the wheel-rail contact parameters. A fully detailed dynamic model of one typical high-speed railway vehicle was developed to investigate the evolution of vehicle hunting stability under the variation of the proposed wear indicator. The results indicate that the proposed wear indicator based on the moved wear area difference has a strong correlation with the equivalent conicity and can be used for the estimation of wheel-rail contact conditions. Furthermore, the hunting stability and the instability form is closely related to the value of the proposed wheel wear indicator, which is helpful to trace the underlying factor of hunting instability, thereby supporting intelligent operation and maintenance of railway vehicles.

Efficient anomaly detection method for offshore wind turbines

Time-series anomaly detection plays a crucial role in the operation of offshore wind turbines. Various wind turbine monitoring systems rely on time-series data to monitor and identify anomalies in real-time, as well as to initiate early warning processes. However, for offshore wind turbines with a high data density, conventional methods have high computational overhead in detecting anomalies while failing to accurately detect anomalies due to variations in data scales. To address this challenge, we propose an efficient anomaly detection method with contrastive learning, called Hawkeye. Hawkeye is based on residual clustering, an unsupervised anomaly detection method for multivariate time-series data. To ensure accurate anomaly detection, a trend-capturing prediction module is also combined with an automatic labeling module. As a result, the most common information can be learned from multivariate time-series data to reconstruct data trends. By evaluating Hawkeye on public datasets and real offshore wind turbine operation datasets, the results show that Hawkeye’s F1-score improves by an average of 14% compared with Isolation Forest, and its size shrinks by up to 11.5 times on the largest dataset compared with other methods. The proposed Hawkeye is potential to real-time monitoring and early warning systems for wind turbines, accelerating the development of intelligent operation and maintenance.