An artificial neural network‐based condition monitoring method for wind turbines,with application to the monitoring of the gearbox

Major failures in wind turbines are expensive to repair and cause loss of revenue due to long downtime. Condition‐based maintenance, which provides a possibility to reduce maintenance cost, has been made possible because of the successful application of various condition monitoring systems in wind turbines. New methods to improve the condition monitoring system are continuously being developed. Monitoring based on data stored in the supervisory control and data acquisition (SCADA) system in wind turbines has received attention recently. Artificial neural networks (ANNs) have proved to be a powerful tool for SCADA‐based condition monitoring applications. This paper first gives an overview of the most important publications that discuss the application of ANN for condition monitoring in wind turbines. The knowledge from these publications is utilized and developed further with a focus on two areas: the data preprocessing and the data post‐processing. Methods for filtering of data are presented, which ensure that the ANN models are trained on the data representing the true normal operating conditions of the wind turbine. A method to overcome the errors from the ANN models due to discontinuity in SCADA data is presented. Furthermore, a method utilizing the Mahalanobis distance is presented, which improves the anomaly detection by considering the correlation between ANN model errors and the operating condition. Finally, the proposed method is applied to case studies with failures in wind turbine gearboxes. The results of the application illustrate the advantages and limitations of the proposed method. Copyright © 2017 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.     

Copula‐based model for wind turbine power curve outlier rejection

Power curve measurements provide a conventional and effective means of assessing the performance of a wind turbine, both commercially and technically. Increasingly high wind penetration in power systems and offshore accessibility issues make it even more important to monitor the condition and performance of wind turbines based on timely and accurate wind speed and power measurements. Power curve data from Supervisory Control and Data Acquisition (SCADA) system records, however, often contain significant measurement deviations, which are commonly produced as a consequence of wind turbine operational transitions rather than stemming from physical degradation of the plant. Using such raw data for wind turbine condition monitoring purposes is thus likely to lead to high false alarm rates, which would make the actual fault detection unreliable and would potentially add unnecessarily to the costs of maintenance. To this end, this paper proposes a probabilistic method for excluding outliers, developed around a copula‐based joint probability model. This approach has the capability of capturing the complex non‐linear multivariate relationship between parameters, based on their univariate marginal distributions; through the use of a copula, data points that deviate significantly from the consolidated power curve can then be removed depending on this derived joint probability distribution. After filtering the data in this manner, it is shown how the resulting power curves are better defined and less subject to uncertainty, whilst broadly retaining the dominant statistical characteristics. These improved power curves make subsequent condition monitoring more effective in the reliable detection of faults. Copyright © 2013 John Wiley & Sons, Ltd.     

基于深度卷积自编码器的风电机组故障预警方法研究

通过风电机组状态监测进行故障预警,可防止故障进一步发展,降低风场运维成本。为充分挖掘风电机组监控与数据采集（SCADA）各状态参数时序信息,以及不同参数之间的非线性关系,该文将深度学习中自动编码器（AE）与卷积神经网络（CNN）相结合,提出基于深度卷积自编码（DCAE）的风电机组状态监测故障预警方法。首先基于历史SCADA数据离线建立基于DCAE的机组正常运行状态模型,然后分析重构误差确定告警阈值,使用EMWA控制图对实时对机组状态监测并进行故障预警。以北方某风电场2 MW双馈型风电机组叶片故障为实例进行实验分析,结果表明该文提出DCAE状态监测故障预警方法,可有效对机组故障提前预警,且优于现有基于深度学习的风电机组故障预警方法,可显著提升重构精度、减少模型参数和训练时间。     

Wind turbine condition monitoring by the approach of SCADA data analysis

Wind turbines are being increasingly deployed in remote onshore and offshore areas due to the richer wind resource there and the advantages of mitigating the land use and visual impact issues. However, site accessing difficulties and the shortage of proper transportation and installation vehicles/vessels are challenging the operation and maintenance of the giants erected at these remote sites. In addition to the continual pressure on lowering the cost of energy of wind, condition monitoring is being regarded as one of the best solutions for the maintenance issues and therefore is attracting significant interest today. Much effort has been made in developing wind turbine condition monitoring systems and inventing dedicated condition monitoring technologies. However, the high cost and the various capability limitations of available achievements have delayed their extensive use. A cost-effective and reliable wind turbine condition monitoring technique is still sought for today. The purpose of this paper is to develop such a technique through interpreting the SCADA data collected from wind turbines, which have already been collected but have long been ignored due to lack of appropriate data interpretation tools. The major contributions of this paper include: (1) develop an effective method for processing raw SCADA data; (2) propose an alternative condition monitoring technique based on investigating the correlations among relevant SCADA data; and (3) realise the quantitative assessment of the health condition of a turbine under varying operational conditions. Both laboratory and site verification tests have been conducted. It has been shown that the proposed technique not only has a potential powerful capability in detecting incipient wind turbine blade and drive train faults, but also exhibits an amazing ability in tracing their further deterioration.     

风力机叶片覆冰状态监测基准值与分级诊断标准研究

以2 MW风力机为研究对象,基于实际风力机状态(SCADA)系统大数据,选取叶片正常状态和覆冰状态下的风速、功率、桨距角和偏航角数据,采用核密度-均值数据处理方法,得到叶片覆冰状态监测基准值及其定量表达式。同时,根据叶片不同覆冰时期桨距角和功率值随风速的变化情况,提出叶片覆冰状态分级诊断标准。应用结果表明,根据桨距角随风速的变化情况可判断在叶片覆冰过程中机组最大功率追踪情况以及气动性能损失情况,根据风速-功率值分布情况可较准确地判别叶片的覆冰状态。     

An engineering condition indicator for condition monitoring of wind turbine bearings

Condition monitoring (CM) of wind turbine becomes significantly important part of wind farms in order to cut down operation and maintenance costs. The large amount of CM system vibration data collected from wind turbines are posing challenges to operators in signal processing. It is crucial to design sensitive and reliable condition indicator (CI) in wind turbine CM system. Bearing plays an important role in wind turbine because of its high impact on downtime and component replacement. CIs for wind turbine bearing monitoring are reviewed in the paper, and the advantages and disadvantages of these indicators are discussed in detail. A new engineering CI (ECI), which combined the energy and kurtosis representation of the vibration signal, is proposed to meet the requirement of easy applicability and early detection in wind turbine bearing monitoring. The quantitative threshold setting method of the ECI is provided for wind turbine CM practice. The bearing run‐to‐failure experiment data analysis demonstrates that ECI can evaluate the overall condition and is sensitive to incipient fault of bearing. The effectiveness in engineering of ECI is validated though a certain amount of real‐world wind turbine generator and gearbox bearing vibration data.     

Condition based maintenance optimization for wind power generation systems under continuous monitoring

By utilizing condition monitoring information collected from wind turbine components, condition based maintenance (CBM) strategy can be used to reduce the operation and maintenance costs of wind power generation systems. The existing CBM methods for wind power generation systems deal with wind turbine components separately, that is, maintenance decisions are made on individual components, rather than the whole system. However, a wind farm generally consists of multiple wind turbines, and each wind turbine has multiple components including main bearing, gearbox, generator, etc. There are economic dependencies among wind turbines and their components. That is, once a maintenance team is sent to the wind farm, it may be more economical to take the opportunity to maintain multiple turbines, and when a turbine is stopped for maintenance, it may be more cost-effective to simultaneously replace multiple components which show relatively high risks. In this paper, we develop an optimal CBM solution to the above-mentioned issues. The proposed maintenance policy is defined by two failure probability threshold values at the wind turbine level. Based on the condition monitoring and prognostics information, the failure probability values at the component and the turbine levels can be calculated, and the optimal CBM decisions can be made accordingly. A simulation method is developed to evaluate the cost of the CBM policy. A numerical example is provided to illustrate the proposed CBM approach. A comparative study based on commonly used constant-interval maintenance policy demonstrates the advantage of the proposed CBM approach in reducing the maintenance cost.     

A new seismic‐based monitoring approach for wind turbines

Wind turbine performance and condition monitoring play vital roles in detecting and diagnosing suboptimal performance and guiding operations and maintenance. Here, a new seismic‐based approach to monitoring the health of individual wind turbine components is presented. Transfer functions are developed linking key condition monitoring properties (drivetrain and tower acceleration) to unique, robust, and repeatable seismic signatures. Predictive models for extreme (greater than 99th percentile) drivetrain and tower acceleration based on independent seismic data exhibit higher skill than reference models based on hub‐height wind speed. The seismic models detect extreme drivetrain and tower acceleration with proportions correct of 96% and 93%, hit rates of 91% and 82%, and low false alarm rates of 4% and 6%, respectively. Although new wind turbines incorporate many diagnostic sensors, seismic‐based condition/performance monitoring may be particularly useful in extending the productive lifetime of previous generation wind turbines.     

基于长短时记忆网络融合SCADA数据的风电齿轮箱状态监测

针对不具有时间记忆能力的机器学习方法融合风电机组数据采集与监控系统(SCADA)的时序数据而导致风电齿轮箱状态预测精度不高的问题,提出基于长短时记忆(LSTM)网络融合SCADA数据的风电齿轮箱状态预测模型。选择能表征风电齿轮箱运行状态的某个监测量作为模型的输出量,基于灰色关联度选择与该监测量关联密切的SCADA参数作为预测模型的输入量;使用正常状态下的SCADA数据训练LSTM预测模型,得出预测值和残差,通过3σ准则计算出上下预警阈值,用于风电齿轮箱状态监测和故障预警。某风电场风电齿轮箱的SCADA数据验证表明所提出的方法能有效预警风电齿轮箱故障。     

Wake impacts on downstream wind turbine performance and yaw alignment

Aerodynamic wake interaction between commercial scale wind turbines can be a significant source of power losses and increased fatigue loads across a wind farm. Significant research has been dedicated to the study of wind turbine wakes and wake model development. This paper profiles influential wake regions for an onshore wind farm using 6 months of recorded SCADA (supervisory control and data acquisition) data. An average wind velocity deficit of over 30% was observed corresponding to power coefficient losses of 0.2 in the wake region. Wind speed fluctuations are also quantified for an array of turbines, inferring an increase in turbulence within the wake region. A study of yaw data within the array showed turbine nacelle misalignment under a range of downstream wake angles, indicating a characteristic of wind turbine behaviour not generally considered in wake studies. The turbines yaw independently in order to capture the increased wind speeds present due to the lateral influx of turbulent wind, contrary to many experimental and simulation methods found in the literature. Improvements are suggested for wind farm control strategies that may improve farm‐wide power output. Additionally, possible causes for wind farm wake model overestimation of wake losses are proposed.Copyright © 2012 John Wiley & Sons, Ltd.     

基于运行数据的风电机组转矩控制性能评估

风电机组的性能评估方法具有多样性及复杂性的特点,基于风电场SCADA系统中采集的大量风电机组运行数据,对风电机组转矩控制的性能评估方法进行了研究。在深入分析风电机组中发电机转速与发电机转矩关系的基础上,提出了风电机组在最佳风能利用系数Cp(max)跟踪区内的转矩优化控制的性能评估方法。通过筛选有效数据,拟合计算出风电机组的实际运行转矩增益系数;再通过与理论最优转矩增益系数进行对比,找出风能捕获能力较弱的风电机组,进而采取措施提高其发电量。通过软件仿真及案例分析表明,该方法在不增加设备及成本的情况下,可有效识别因转矩控制的性能差而影响发电量的风电机组,以便及时进行控制策略调校,维护风电场的利益。     

1.5MW风力发电场SCADA系统设计

设计了1.5 MW风场控制系统,实现对风场风力机运行的集中监测与控制,具备设备控制、显示、报表、历史、报警、趋势等功能,同时实现风力设备远程监控。该系统允许用户通过办公室电脑、远程电脑、或直接从风机塔底柜触摸屏上查看风机运行数据,可以在风机本地,实现对风机实时运行数据的监视、风机控制以及调试功能,还可以将实时采集的风机相关数据发送到中央监控室。     

Fatigue distribution optimization for offshore wind farms using intelligent agent control

A novel control approach is proposed to optimize the fatigue distribution of wind turbines in a large‐scale offshore wind farm on the basis of an intelligent agent theory. In this approach, each wind turbine is considered to be an intelligent agent. The turbine at the farm boundary communicates with its neighbouring downwind turbines and organizes them adaptively into a wind delivery group along the wind direction. The agent attributes and the event structure are designed on the basis of the intelligent agent theory by using the unified modelling language. The control strategy of the intelligent agent is studied using topology models. The reference power of an individual wind turbine from the wind farm controller is re‐dispatched to balance the turbine fatigue in the power dispatch intervals. In the fatigue optimization, the goal function is to minimize the standard deviation of the fatigue coefficient for every wind turbine. The optimization is constrained such that the average fatigue for every turbine is smaller than what would be achieved by conventional dispatch and such that the total power loss of the wind farm is restricted to a few percent of the total power. This intelligent agent control approach is verified through the simulation of wind data from the Horns Rev offshore wind farm. The results illustrate that intelligent agent control is a feasible way to optimize fatigue distribution in wind farms, which may reduce the maintenance frequency and extend the service life of large‐scale wind farms. Copyright © 2012 John Wiley & Sons, Ltd.     

Wind shear estimation model using load measurement of wind turbine tower and surrogate model

A wind shear estimation method based on fore‐aft moment is proposed to estimate wind shear strength without a Doppler lidar. We construct wind shear estimation models (WSEMs) using surrogate models whose input is the time‐averaged fore‐aft moment and various supervisory control and data acquisition (SCADA) system data. Learning data for the WSEMs are generated by numerical simulation or field measurement of a real turbine using SCADA, strain gauges, and Doppler lidar. By using simulation data, we construct 20 WSEMs with various input combinations and surrogate methods to select a model with the highest accuracy. The best WSEM is constructed with the universal Kriging surrogate model and uses the fore‐aft moment and wind speed as its input. Subsequently, the best WSEM is applied to a real turbine to validate its accuracy in real wind conditions, and we confirm that the WSEM has reasonable accuracy. However, the estimation error in the real wind condition is about twice as high as that in the simulation due to the real wind shear not completely corresponding to the assumed wind profile and a large yaw error. Further improvement in wind shear estimation accuracy will be achieved by adding yaw error and turbulence intensity to the input variables and applying the WSEM to wind farms on simple terrain or offshore wind farms where wind profile error decreases.     

Possible power of down‐regulated offshore wind power plants: The PossPOW algorithm

This paper proposes a method for real‐time estimation of the possible power of an offshore wind power plant when it is down‐regulated. The main purpose of the method is to provide an industrially applicable estimate of the possible (or reserve) power. The method also yields a real‐time power curve, which can be used for operation monitoring and wind farm control. Currently, there is no verified approach regarding estimation of possible power at wind farm scale. The key challenge in possible power estimation at wind farm level is to correct the reduction in wake losses, which occurs due to the down‐regulation. Therefore, firstly, the 1‐second wind speeds at the upstream turbines are estimated, since they are not affected by the reduced wake. Then they are introduced into the wake model, adjusted for the same time resolution, to correct the wake losses. To mitigate the uncertainties due to dynamic changes within the large offshore wind farms, the algorithm is updated at every turbine downstream, considering the local axial and lateral turbulence effects. The PossPOW algorithm uses only 1‐Hz turbine data as inputs and provides possible power output. The algorithm is trained and validated in Thanet and Horns Rev‐I offshore wind farms under nominal operation, where the turbines are following the optimum power curve. The results indicate that the PossPOW algorithm performs well; in the Horns Rev‐I wind farm, the strict power system requirements are met more than 70% of the time over the 24‐hour data set on which the algorithm was evaluated.     

基于随机组合赋权模糊评价的风电机组健康状态评估

由于健康指标权重随机性会导致风电机组状态评估灵敏度降低,提出一种评估风电机组健康状态的随机组合赋权模糊评价方法。首先,通过相关性、方差、偏度等多角度分析风电场采集与监视控制系统（SCADA）数据,结合IEC61400-1标准建立机组健康状态评估指标架构,并基于随机因子优化组合权重得到赋权公式,提高评估指标层权重的准确性。其次,为充分覆盖评估指标数据劣化度,基于岭型分布函数建立健康指标劣化隶属度计算函数。结合随机组合权重和隶属度函数,构建风电机组健康状态模糊综合评价数学模型。通过分层评估风电机组健康状态指标架构,得到机组健康等级并实现故障预警。最后,对大连驼山风电场多台机组进行评估试验,结果表明：该文方法能准确评估出风电机组健康状态等级,相比组合赋权云模型方法,灵敏度提高了1.85%。     

数据驱动的风电机组变桨系统状态监测

针对数据采集与监视控制(SCADA)系统存在误报、故障报警滞后等问题,提出一种基于单分类模型的风电机组变桨系统在线状态监测方法.首先,从SCADA数据中提取出与变桨系统相关的特征参数并进行特征重构以进一步提取出更值得关注的桨叶之间的差异化信息.其次,基于单分类支持向量机对历史数据的分析确定变桨系统运行数据的健康边界,进...     

基于改进劣化度模型的风电机组日常运行状态评估

为准确地掌握整台风电机组的运行状态,提出一种基于改进劣化度模型的评估方法。首先求得风电机组层次结构中各状态参数的劣化度;其次通过组合赋权法确定各参数的权重,利用岭型分布隶属度函数确定各参数的隶属度矩阵;最后利用模糊综合评判法和彩色图谱对机组的运行状态进行评估和展示。案例机组的日常运行状态评估表明：所提方法可较SCADA系统提前发现机组发电机非驱动端轴承的异常,且误报次数较少。     

Nonlinear system identification for model-based condition monitoring of wind turbines

This paper proposes a data driven model-based condition monitoring scheme that is applied to wind turbines. The scheme is based upon a non-linear data-based modelling approach in which the model parameters vary as functions of the system variables. The model structure and parameters are identified directly from the input and output data of the process. The proposed method is demonstrated with data obtained from a simulation of a grid-connected wind turbine where it is used to detect grid and power electronic faults. The method is evaluated further with SCADA data obtained from an operational wind farm where it is employed to identify gearbox and generator faults. In contrast to artificial intelligence methods, such as artificial neural network-based models, the method employed in this paper provides a parametrically efficient representation of non-linear processes. Consequently, it is relatively straightforward to implement the proposed model-based method on-line using a field-programmable gate array.