Deep learning–based automatic bird identification system for offshore wind farms

Practical deterrent methods are needed to prevent collisions between birds and wind turbine blades for offshore wind farms. It is improbable that a single deterrent method would work for all bird species in a given area. An automatic bird identification system is required in order to develop bird species–level deterrent methods. This system is the first and necessary part of the entirety that is eventually able to automatically monitor bird movements, identify bird species, and launch deterrent measures. A prototype system has been built on Finnish west coast. In the proposed system, a separate radar system detects birds and provides WGS84 coordinates to a steering system of a camera. The steering system consists of a motorized video head and our software to control it. The steering system tracks flying birds in order to capture series of images by a digital single‐lens reflex camera. Classification is based on these images, and it is implemented by convolutional neural network trained with a deep learning algorithm. We applied to the images our data augmentation method in which images are rotated and converted into different color temperatures. The results indicate that the proposed system has good performance to identify bird species in the test area. Aiming accuracy for the video head was 88.91 %. Image classification performance as true positive rate was 0.8688.     

基于鸟鸣功率谱密度的输电线路涉鸟故障相关鸟种智能识别

鸟类活动对架空输电线路的运行影响较大,不同鸟种引起的线路故障差异明显,为了提高渉鸟故障防治的针对性,需分类识别输电线路相关鸟种。以鸟粪类、鸟巢类、鸟体短接类、鸟啄类四种渉鸟故障对应的8种代表性鸟种为例,通过分帧加窗、端点检测等预处理,采用离散傅里叶变换和Welch算法获取不同鸟鸣信号的功率谱密度(PSD)图,从中提取129个频率点的PSD特征,并基于随机森林建立分类识别模型,以PSD特征作为输入量,开展模型训练和不同鸟种分类测试,8种鸟类的识别准确率为83.3%~100.0%,提高了渉鸟故障防治措施的精准性。     

Effects of sea surface temperature accuracy on offshore wind resource assessment using a mesoscale model

Offshore wind simulations were performed with the Weather Research and Forecasting (WRF) model driven by three different sea surface temperature (SST) datasets for Japanese coastal waters to investigate the effect of the SST accuracies on offshore wind simulations. First, the National Centers for Environmental Prediction Final analysis (FNL) (1° × 1° grid resolution) and the Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) (0.05° × 0.05° grid resolution) datasets were compared with in situ measurements. The results show a decrease in accuracy of these datasets toward the coast from the open ocean. Aiming at an improved accuracy of SST data, we developed a new high‐resolution SST dataset (0.02° × 0.02° grid resolution). The new dataset referred to as MOSST is based on the Moderate Resolution Imaging Spectroradiometer (MODIS) product, provided by the Japan Aerospace Exploration Agency (JAXA). MOSST was confirmed to be more accurate than FNL and OSTIA for the coastal waters. Then, WRF simulations were carried out for 1 year with a 2 km grid resolution and by using the FNL, OSTIA and MOSST datasets. The use of the OSTIA dataset for a WRF simulation was found to improve the accuracy when compared with the FNL dataset, and further improvement was obtained when the MOSST dataset was applied. The sensitivity of wind speed and wind energy density to SST is also discussed. We conclude that the use of an accurate SST is a key factor not only for realistic offshore wind simulations near the surface but also for accurate wind resource assessments at the hub height of wind turbines. Copyright © 2014 John Wiley & Sons, Ltd.     

Advanced analytics for detection and diagnosis of false alarms and faults: A real case study

Onshore and offshore wind farms require a high level of advanced maintenance. Supervisory control and data acquisition (SCADA) and condition monitoring systems are now being employed, generating large amounts of data. They require robust and flexible approaches to convert dataset into useful information. This paper presents a novel approach based on the correlations of SCADA variables to detect and identify faults and false alarms in wind turbines. A correlation matrix between all the SCADA variables is used for pattern recognition. A new method based on curve fittings is employed for detecting false alarms and abnormal behaviours or faults in the components. The study is done in a real case study, validated with false alarms.     

江西电网涉鸟故障分析及防范

为分析江西电网涉鸟故障原因及规律,通过统计2009~2016年涉鸟故障,发现涉鸟故障主要发生在农田或邻近水源低海拔地区,4~7月为涉鸟故障发生的高峰期,且鸟巢类故障主要发生在凌晨和早晨,而鸟粪类故障主要发生在夜间至凌晨。同时,涉鸟故障发生的概率还与杆塔的电压等级、杆塔结构型式、排列方式、相别和天气有关。为更好地防范涉鸟故障的发生,综合考虑鸟类分布、人类干扰度、地理环境和运行经验等要素,提出了涉鸟故障差异化防治措施。研究成果对于降低涉鸟故障率,保障电网安全运行有重要意义。     

220 kV线路鸟害故障分析及防治对策

针对巴彦淖尔河套地区近年来220kV线路连续不断发生的鸟害故障,通过分析比较,阐述了鸟类对架空电力线路的危害,分析了河套地区鸟害故障的原因特征及故障规律,并结合工作实际提出了具体的防鸟害措施。     

架空电力线路鸟害分析与预防

针对鸟类的频繁活动极易造成电力架空线路故障甚至事故,从鸟害的形成机理上进行分析和归类,并提出了相应的防治措施。     

Comparative performance of AI methods for wind power forecast in Portugal

Because wind has a high volatility and the respective energy produced cannot be stored on a large scale because of excessive costs, it is of utmost importance to be able to forecast wind power generation with the highest accuracy possible. The aim of this paper is to compare 1‐h‐ahead wind power forecasts performance using artificial intelligence‐based methods, such as artificial neural networks (ANNs), adaptive neural fuzzy inference system (ANFIS), and radial basis function network (RBFN). The latter was implemented using three different learning algorithms: stochastic gradient descent (SGD), hybrid, and orthogonal least squares (OLS). The application dataset is the injected wind power in the Portuguese power systems throughout the years 2010–2014. The network architecture optimization and the learning algorithms are presented. An initial data analysis showed data seasonality; therefore, the wind power forecasts were performed according to the seasons of the year. The results showed that ANFIS was the best performer method, and ANN and RBFN‐OLS also showed strong performances. RBFN‐Hybrid and RBFN‐SGD performed poorly. In general, all methods outperformed persistence.     

Estimation of variability and predictability of large‐scale wind energy in The Netherlands

This paper presents a data‐driven approach for estimating the degree of variability and predictability associated with large‐scale wind energy production for a planned integration in a given geographical area, with an application to The Netherlands. A new method is presented for generating realistic time series of aggregated wind power realizations and forecasts. To this end, simultaneous wind speed time series—both actual and predicted—at planned wind farm locations are needed, but not always available. A 1‐year data set of 10‐min averaged wind speeds measured at several weather stations is used. The measurements are first transformed from sensor height to hub height, then spatially interpolated using multivariate normal theory, and finally averaged over the market resolution time interval. Day‐ahead wind speed forecast time series are created from the atmospheric model HiRLAM (High Resolution Limited Area Model). Actual and forecasted wind speeds are passed through multi‐turbine power curves and summed up to create time series of actual and forecasted wind power. Two insights are derived from the developed data set: the degree of long‐term variability and the degree of predictability when Dutch wind energy production is aggregated at the national or at the market participant level. For a 7.8 GW installed wind power scenario, at the system level, the imbalance energy requirements due to wind variations across 15‐min intervals are ±14% of the total installed capacity, while the imbalance due to forecast errors vary between 53% for down‐ and 56% for up‐regulation. When aggregating at the market participant level, the balancing energy requirements are 2–3% higher. Copyright © 2008 John Wiley & Sons, Ltd.     

海上风电场对鸟类的影响及其危害预防

目的 风能作为公认的最成熟的可再生能源技术之一,近年来发展迅速。中国在近年开发了大量的海上风电项目。但风电场对环境,尤其是对鸟类的影响引发了人们大量的担忧和研究。在碳达峰实现之前,我国的海上风能产业将持续增长,需要全面了解风电场对鸟类的影响。 方法 通过总结现有研究,对海上风电场引起的鸟类问题进行了综述,并讨论了可将对鸟类不利环境影响降至最低的预防和缓解措施。 结果 研究表明：危害鸟类生存的风电问题主要有风力涡轮机叶片撞击和风机运行噪音两方面的原因,尚无明确的证据表明鸟类会受到电磁场的影响。风电场的运行可能造成鸟类栖息地的变迁、繁殖和交流的受阻以及种群结构的改变等。人们可以通过风电场选址、风电场形态及风机叶片和桩机结构设计、遥感与视频监控等方式减少风电场对鸟类的不利影响。 结论 对风电场鸟类开展更加深入的研究,将有利于我们掌握海上风电场开发建设及运营相关的鸟类生态学基本规律,形成降低生态环境负面影响的科学决策。     

Validation of a hybrid modeling approach to floating wind turbine basin testing

Hybrid modeling combining physical tests and numerical simulations in real time opens new opportunities in floating wind turbine research. Wave basin testing is an important validation step for floating support structure design, but current methods are limited by scaling problems in the aerodynamic loadings. Applying wind turbine loads with an actuation system controlled by a simulation that responds to the basin test offers a way to avoid scaling problems and reduce cost barriers for floating wind turbine design validation in realistic coupled conditions. In this work, a cable‐based hybrid coupling approach is developed and implemented for 1:50‐scale wave basin tests with the DeepCwind semisubmersible floating wind turbine. Tests are run with thrust loads provided by a numerical wind turbine model. Matching tests are run with physical wind loads using an above‐basin wind maker. When the numerical submodel is set to match the aerodynamic performance of the physical scaled wind turbine, the results show good agreement with purely physical wind‐wave tests, validating the hybrid model approach. Further hybrid model tests with simulated true‐to‐scale dynamic thrust loads and wind turbulence show noticeable differences and demonstrate the value of a hybrid model approach for improving the true‐to‐scale realism of floating wind turbine basin tests.     

海上风电场对鸟类的影响及其危害预防

[目的]风能作为公认的最成熟的可再生能源技术之一,近年来发展迅速.中国在近年开发了大量的海上风电项目.但风电场对环境,尤其是对鸟类的影响引发了人们大量的担忧和研究.在碳达峰实现之前,我国的海上风能产业将持续增长,需要全面了解风电场对鸟类的影响.[方法]通过总结现有研究,对海上风电场引起的鸟类问题进行了综述,并讨论了可将...     

Properties of the offshore low level jet and rotor layer wind shear as measured by scanning Doppler Lidar

The atmospheric flow phenomenon known as the Low Level Jet (LLJ) is an important source of wind power production in the Great Plains. However, due to the lack of measurements with the precision and vertical resolution needed, particularly at rotor heights, it is not well‐characterized or understood in offshore regions being considered for wind‐farm development. The present paper describes the properties of LLJs and wind shear through the rotor layer of a hypothetical wind turbine, as measured from a ship‐borne Doppler lidar in the Gulf of Maine in July–August 2004. LLJs, frequently observed below 600 m, were mostly during nighttime and transitional periods, but they were also were seen during some daytime hours. The presence of a LLJ significantly modified wind profiles producing vertical wind speed shear. When the wind shear was strong, the estimates of wind power based upon wind speeds measured at hub‐height could have significant errors. Additionally, the inference of hub‐height winds from near‐surface measurements may introduce further error in the wind power estimate. The lidar dataset was used to investigate the uncertainty of the simplified power‐law relation that is often employed in engineering approaches for the extrapolation of surface winds to higher elevations. The results show diurnal and spatial variations of the shear exponent empirically found from surface and hub‐height measurements. Finally, the discrepancies between wind power estimates using lidar‐measured hub‐height winds and rotor equivalent winds are discussed. Copyright © 2016 John Wiley & Sons, Ltd.     

Deep learning with knowledge transfer for explainable anomaly prediction in wind turbines

The last decade has witnessed an increased interest in applying machine learning techniques to predict faults and anomalies in the operation of wind turbines. These efforts have lately been dominated by deep learning techniques which, as in other fields, tend to outperform traditional machine learning algorithms given sufficient amounts of training data. An important shortcoming of deep learning models is their lack of transparency—they operate as black boxes and typically do not provide rationales for their predictions, which can lead to a lack of trust in predicted outputs. In this article, a novel hybrid model for anomaly prediction in wind farms is proposed, which combines a recurrent neural network approach for accurate classification with an XGBoost decision tree classifier for transparent outputs. Experiments with an offshore wind turbine show that our model achieves a classification accuracy of up to 97%. The model is further able to generate detailed feature importance analyses for any detected anomalies, identifying exactly those components in a wind turbine that contribute to an anomaly. Finally, the feasibility of transfer learning is demonstrated for the wind domain by porting our “offshore” model to an unseen dataset from an onshore wind farm. The latter model achieves an accuracy of 65% and is able to detect 85% of anomalies in the unseen domain. These results are encouraging for application to wind farms for which no training data are available, for example, because they have not been in operation for long.     

基于深度迁移学习的小样本光伏热斑识别方法

针对训练传统深度学习模型需大量数据而热斑效应样本数量相对少且不易采集的问题,提出基于深度迁移学习的小样本光伏热斑识别方法.在Inception-v3模型的基础上构建深度迁移学习模型,然后在负样本多分类的小样本热斑数据集上完成训练,得到可用于热斑识别的网络模型.实验结果表明,在样本数量不充足的情况下深度迁移学习方法训练出...     

Predicting wind power variability events using different statistical methods driven by regional atmospheric model output

Variability in power generation from wind farms is an important issue in the energy industry. If sub‐hour variability events can be predicted, potential disruptions to the grid operations might be mitigated. Using 4 years of 5 min wind power data from the Australian Energy Market Operator for an 80 MW wind farm in south‐east Australia, we fit statistical models of variability on meteorological reanalysis data from the US National Centers for Environmental Prediction. The National Centers for Environmental Prediction fields were transformed into spatial empirical orthogonal functions, and 6 h projections onto these became explanatory covariates for generalized linear, random forest (RF), gradient boosting and support vector machine classification models. Other covariates considered were local wind speed and 6 h‐lagged empirical orthogonal function differences. Models were selected by minimizing cross‐validated misclassification rate and assessed using area under the receiver operating characteristic curve and reliability score. Considering performance and ease of tuning, RFs were preferred. Performance was poorer for larger ramps. The RFs accurately predicted their performance on the validation set. For asymmetric costs (miss‐to‐false alarm cost ratio = 10), RFs yielded competitive low‐cost models. Support vector machines produced slightly superior models but needed to be tuned manually. RF models using atmospheric model output provide a robust approach to predicting wind power variability and relatively large ramp events. We recommend the RF models as a practical and skilful method to feed into an early warning system for energy/electricity operators. Copyright © 2014 John Wiley & Sons, Ltd.     

基于主动探测技术的输电线路驱鸟策略研究

为解决鸟害引起的输配电线路故障,通过总结鸟害故障机理,分析鸟害事故较严重的华北地区鸟害的时间规律,提出了基于光电探测技术、激光立体扫描和能源优化设计的驱鸟策略。研究表明,鸟害故障一般发生在春秋两季,且06:00～08:00时间段故障率较高;光电探测技术可探测与探测平面成60°范围内的飞鸟;激光器可实现上下左右10°范围内立体式驱鸟,覆盖范围广;驱鸟策略采用定扫模式和巡航模式,系统综合功耗优化至38AH,鸟害故障率降低了60%。研究结果为输配电线路鸟害故障治理提供了参考。     

Generating wind time series as a hybrid of measured and simulated data

Certain applications, such as analysing the effect of a wind farm on grid frequency regulation, require several years of wind power data measured at intervals of a few seconds. We have developed a method to generate days to years of non‐stationary wind speed time series sampled at high rates by combining measured and simulated data. Measured wind speed data, typically 10–15 min averages, capture the non‐stationary characteristics of wind speed variation: diurnal variations, the passing of weather fronts, and seasonal variations. Simulated wind speed data, generated from spectral models, add realistic turbulence between the empirical data. The wind speed time series generated with this method agree very well with measured time series, both qualitatively and quantitatively. The power output of a wind turbine simulated with wind data generated by this method demonstrates energy production, ramp rates and reserve requirements that closely match the power output of a turbine simulated turbine with measured wind data. Copyright © 2011 John Wiley & Sons, Ltd.     

A general method to estimate wind farm power using artificial neural networks

An artificial neural network (ANN) is trained and validated using a large dataset of observations of wind speed, direction, and power generated at an offshore wind farm (Lillgrund in Sweden). In its traditional form, the ANN is used to generate a new two‐dimensional power curve, which predicts with high accuracy (bias ～?0.5% and absolute error ～2%) the power of the entire Lillgrund wind farm based on wind speed and direction. By contrast, manufacturers only provide one‐dimensional power curves (i.e., power as a function of wind speed) for a single turbine. The second innovative application of the ANN is the use of a geometric model (GM) to calculate two simple geometric properties to replace wind direction in the ANN. The resulting GM‐ANN has the powerful feature of being applicable to any wind farm, not just Lillgrund. A validation at an onshore wind farm (Nørrekær in Denmark) demonstrates the high accuracy (bias ～?0.7% and absolute error ～6%) and transfer‐learning ability of the GM‐ANN.     

One‐way mesoscale‐microscale coupling for simulating a wind farm in North Texas: Assessment against SCADA and LiDAR data

One‐way nested mesoscale to microscale simulations of an onshore wind farm have been performed nesting the Weather Research and Forecasting (WRF) model and our in‐house high‐resolution large‐eddy simulation code (UTD‐WF). Each simulation contains five nested WRF domains, with the largest domain spanning the north Texas Panhandle region with a 4 km resolution, while the highest resolution (50 m) nest simulates microscale wind fluctuations and turbine wakes within a single wind farm. The finest WRF domain in turn drives the UTD‐WF LES higher‐resolution domain for a subset of six turbines at a resolution of ～5 m. The wind speed, direction, and boundary layer profiles from WRF are compared against measurements obtained with a met‐tower and a scanning Doppler wind LiDAR located within the wind farm. Additionally, power production obtained from WRF and UTD‐WF are assessed against supervisory control and data acquisition (SCADA) system data. Numerical results agree well with the experimental measurements of the wind speed, direction, and power production of the turbines. UTD‐WF high‐resolution domain improves significantly the agreement of the turbulence intensity at the turbines location compared with that of WRF. Velocity spectra have been computed to assess how the nesting allows resolving a wide range of scales at a reasonable computational cost. A domain sensitivity analysis has been performed. Velocity spectra indicate that placing the inlet too close to the first row of turbines results in an unrealistic peak of energy at the rotational frequency of the turbines. Spectra of the power production of a single turbine and of the cumulative power of the array have been compared with analytical models.