A Hybrid Nonlinear Forecasting Strategy for Short-Term Wind Speed

This contribution presents a new development in the design of control system based on evolving Gaussian process (GP) models . GP models provide a probabilistic, nonparametric modelling approach for black-box identification of nonlinear dynamic systems. They can highlight areas of the input space where prediction quality is poor, due to the lack of data or its complexity, by indicating the higher variance around the predicted mean. GP models contain noticeably less coefficients to be optimised than commonly used parametric models. While GP models are Bayesian models, their output is normal distribution, expressed in terms of mean and variance. Latter can be interpreted as a confidence in prediction and used in many fields, especially in control system. Evolving GP model is the concept approach within which various ways of model adaptations can be used. Successful control system needs as much as possible data about process to be controlled . If the prior knowledge about the system to be controlled is scarce or the system varies with time or operating region, this control problem can be solved with an iterative method which adapts model with information obtained with streaming data and concurrently optimises hyperparameter values. This contribution provides: a survey of adaptive control algorithms for dynamic systems described in publications where GP models have been used for control design, a novel and improved closed-loop controller with evolving GP models and an example for the illustration of proposed control algorithm.

Short-term wind speed prediction based on feature extraction with Multi-task Lasso and Multilayer Perceptron

Earth observation from satellite sensory data pose challenging problems, where machine learning is currently a key player. In recent years, Gaussian Process (GP) regression and other kernel methods have excelled in biophysical parameter estimation tasks from space. GP regression is based on solid Bayesian statistics, and generally yield efficient and accurate parameter estimates. However, GPs are typically used for inverse modeling based on concurrent observations and in situ measurements only. Very often a forward model encoding the well-understood physical relations is available though. In this work, we review three GP models that respect and learn the physics of the underlying processes in the context of inverse modeling. First, we will introduce a Joint GP (JGP) model that combines in situ measurements and simulated data in a single GP model. Second, we present a latent force model (LFM) for GP modeling that encodes ordinary differential equations to blend data-driven modeling and physical models of the system. The LFM performs multi-output regression, adapts to the signal characteristics, is able to cope with missing data in the time series, and provides explicit latent functions that allow system analysis and evaluation. Finally, we present an Automatic Gaussian Process Emulator (AGAPE) that approximates the forward physical model via interpolation, reducing the number of necessary nodes. Empirical evidence of the performance of these models will be presented through illustrative examples of vegetation monitoring and atmospheric modeling.

Wind speed/power has received increasing attention around the earth due to its renewable nature as well as environmental friendliness. With the global installed wind power capacity rapidly increasing, wind industry is growing into a large-scale business. Reliable short-term wind speed forecasts play a practical and crucial role in wind energy conversion systems, such as the dynamic control of wind turbines and power system scheduling. In this paper, an intelligent hybrid model for short-term wind speed prediction is examined; the model is based on cross correlation (CC) analysis and a support vector regression (SVR) model that is coupled with brainstorm optimization (BSO) and cuckoo search (CS) algorithms, which are successfully utilized for parameter determination. The proposed hybrid models were used to forecast short-term wind speeds collected from four wind turbines located on a wind farm in China. The forecasting results demonstrate that the intelligent hybrid models outperform single models for short-term wind speed forecasting, which mainly results from the superiority of BSO and CS for parameter optimization.

Quantitative prediction of subjective pain intensity from whole-brain fMRI data using Gaussian processes

Wind speed prediction in short term is required to asses the effect of wind on different objects in action in free space, like rockets, navigating ships and planes, guided missiles satellites in launch etc. Forecasting also helps in usage of wind energy as an alternative source of energy in Electrical power generation plants. The wind speed depends on the values of other atmospheric variables, such as pressure, moisture content, humidity, rainfall etc. This paper reports a Support Vector Machine model for short term wind speed prediction. The model uses the values of these parameters, obtained from a nearest weather station, as input data. The trained model is validated using a part of data. The model is then used to predict the wind speed, using the same meteorological information.

Two-stage correction prediction of wind power based on numerical weather prediction wind speed superposition correction and improved clustering

The large-scale integration of wind power into the grid has led to a reduction in system inertia, threatening frequency stability. There is an urgent need to accurately assess the inertia support capability of wind turbines, providing a theoretical basis for grid inertia dispatch and supporting grid frequency stability. However, due to factors such as wake effects, time-delay effects, and wind shear effects, the rotational speeds of different wind turbines within a wind farm under certain wind speed conditions exhibit probabilistic distribution characteristics. Existing research on wind turbine inertia assessment rarely accounts for the rotational speed randomness. To address this, this paper proposes a probabilistic assessment method for the available inertia of wind turbines that considers rotational speed randomness, establishes a joint probability model for wind speed and rotational speed, deriving the conditional probability density function of rotational speed. By substituting this into the frequency-domain inertia model, we achieve probabilistic inertia assessment. Using operational data from a wind farm in China, a practical case study is constructed, verifying the accuracy of the proposed probabilistic assessment method. At a wind speed of 6 m/s, the proposed method accurately captures the actual system inertia within its 90% confidence interval, in contrast to a conventional approach which yielded a significant 6.5% error.

In short-term wind speed prediction, different features in the time window have different degrees of influence on the prediction results. However, the importance of different features in the time window is mostly treated equally, which results in the deficiency of responding promptly to the time-varying characteristics of wind speed and treating differently the time window feature weights that affect the prediction accuracy. Hence, a novel model called WTCGRU is proposed to improve the prediction accuracy by fully considering the influence of different features in the time window during wind speed prediction. In WTCGRU, the weights of different features in the time window are sampled through an adaptive genetic approach, which is called importance sampling, and fed into a hybrid model composed of GRU and TCN, to predict more precisely the short-term wind speed; This step is executed repeatedly until WTCGRU's prediction accuracy is higher than a predefined threshold. Finally, theoretical experiments are carried out on the wind speed dataset from Szeged of Hungary, and it is shown that the WTCGRU has outperformed the mainstream short-term wind speed prediction model TCN, BILSTM, CNN-LSTM, and ensemble-CNN.

Turbine-specific short-term wind speed forecasting considering within-farm wind field dependencies and fluctuations

هدف از انجام‏دادن ‏این پژوهش بررسی عملکرد پایگاه داده بازکاوی ECMWF برای توزیع زمانی-مکانی تندی باد در شرق ایران و روند آن است. به این منظور، از داده‏های بازکاوی ECMWF با تفکیک افقی 125/0×125/0 درجة قوسی استفاده شد؛ عملکرد داده‏ها با استفاده از 11 ایستگاه سینوپتیک با دورة آماری ۱۹۸۵-2015 و به‏کارگیری نمایة آماری RMSE، MBE، MAE، و R2 بررسی شد. نتایج نشان داد Interim برای بررسی تندی باد از عملکرد بالا و مناسبی برخوردار است. متوسط بلندمدت تندی باد در منطقة مورد مطالعه 56/3 m/s است؛ بیشینه و کمینة تندی باد به‏ترتیب در جولای و دسامبر اتفاق افتاده است. آرایش اصلی باد در شرق ایران شرقی و شمالی است؛ ارتباط بین تندی باد با ارتفاع معکوس و با طول جغرافیایی مستقیم و در سطح 05/0 معنی‏دار است. همچنین، ارتباط بین عرض جغرافیایی و تندی باد نشان داد که این ارتباط در ماه‏های سرد سال معکوس و در ماه‏های گرم سال مستقیم است. بررسی روند تندی باد با استفاده از آزمون من- کندال (M-K) نشان داد متوسط روند تندی باد در هفت ماه سال مثبت و در پنج ماه منفی است. همچنین، روند تندی باد در زمان آغاز (جون) باد 120روزه مثبت (۱۹۵/۰)و در زمان خاتمه (اکتبر) آن منفی (-۱۵۲/۰) است.

For the challenging task of modeling actual complex motions, we propose a new class of Gaussian process (GP) models that are data-driven and also take into account prior knowledge of the motion intention. As a theoretical basis, we show that the GP regression is mathematically equivalent to regularized least-squares estimation for random functions with known prior means. Compared with the popular GP models in machine learning literature, the proposed GP motion model priors with conditional kernels have at least two advantages: 1) they are nonstationary and more applicable to represent complex motions by integrating the basic kinematic principles; 2) conditional kernels are further devised by incorporating the motion intent so that the resultant GP models are more versatile and would expectedly entail more accurate trajectory prediction. A superior property of the GP models with conditional kernels is found to improve the computational efficiency. Finally, illustrative examples are provided to show the superiority of the proposed motion models and to verify the theoretical results given in the paper.

Horizontal axis wind turbines (HAWT's) routinely suffer significant time varying aerodynamic loads that exercise adverse effects on structures, mechanical components, and power production. As lighter and more flexible wind energy machines are designed to reduce overall cost of energy, greater accuracy and reliability will become even more crucial in future aerodynamics models. Aerodynamics modeling advances will, in turn, rely on improved understanding of the three-dimensional, unsteady, vortical flows that reside on turbine blades under high load conditions. To experimentally characterize these flows, the National Renewable Energy Laboratory (NREL) Phase VI Unsteady Aerodynamics Experiment (UAE) turbine was erected in the NASA Ames National Full- Scale Aerodynamics Complex (NFAC) 80 foot x 120 foot wind tunnel. Then, under strictly controlled inflow conditions, turbine blade surface pressures and local inflow velocities were acquired at multiple radial locations. Surface pressure histories and normal force records were used to characterize dynamic stall vortex kinematics and normal forces. Stall vortices occupied approximately two-thirds of the aerodynamically active blade span, and persisted for nearly one-fourth of the blade rotation cycle. Stall vortex convection varied dramatically along the blade radius, yielding pronounced, dynamic stall vortex deformation. Analysis of these data revealed systematic alterations to vortex kinematics due to changes in wind speed, yaw error, and blade span location. NOMENCLATURE

With the development of Wind power grid connection technology, the ultra-short-term wind speed prediction technology plays an important role in power system operation and energy scheduling. This paper proposes a median regression prediction model based on Gaussian process(GPMR). This model can accurately analyze and predict the time series of ultra-short-term wind speed from several minutes to an hour. Based on the assumption of prior distribution by Gaussian process, this method solves the problem of nonlinear processing in the conventional time series method. It also combines with the selection of median, overcoming the shortcoming that the previous mean point prediction is easily affected by large error deviation, making the prediction result more representative. In the part of example analysis, two machine learning prediction methods commonly used in recent years are compared, and the simulation results show that the method proposed in this paper is more accurate in the ultra-short-term prediction of wind speed.

The main purpose of the paper is to use Computational Fluid Dynamics (CFD) in 3-D analysis of aerodynamic forces of a Horizontal Axis Wind Turbine (HAWT) blade and compare the 3-D results with the 2-D experimental results. The National Renewable Energy Laboratory (NREL) Phase VI wind blade profile is used as a model for the analysis. The results are compared with the experimental data obtained by NREL at NASA Ames Research Center for the NREL Phase VI wind turbine blade. The aerodynamic forces are evaluated using 3-D Computational Fluid Dynamics (CFD) simulation. The commercial ANSYS CFX and parameterized 3-D CAD model of NREL Phase VI are used for the analysis. The Shear Stress Transport (SST) Gamma-Theta turbulence model and 0-degree yaw angle condition are adopted for CFD analysis. For the case study seven varying wind speeds (5 m/s, 7 m/s, 10 m/s, 13 m/s, 15 m/s, 20 m/s, 25 m/s) with constant blade rotational speed (72 rpm) are considered. To evaluate the 3-D aerodynamic effect sectional pressure coefficient (Cp) and integrated forces about primary axis such as normal, tangential, thrust and torque are evaluated for each of the seven wind speed cases and compared with the NREL experimental values. The numerical difference of values on wind blade surface between this study and 3-D results of NREL wind tunnel test are found negligible. The paper represents an important comparison between the 3-D lift & drag coefficient with the NREL 2-D experimental data. The results shows that though the current study is in good agreement with NREL 3-D experimental values there is large deviation between the NREL 2-D experimental data and current 3-D study which suggests that in case of 3-D analysis of aerodynamic force of blade surface it is better to use NREL 3-D values instead of 2-D experimental values.