scholarly journals Unmanned Aerial Vehicle Pitch Control Using Deep Reinforcement Learning with Discrete Actions in Wind Tunnel Test

Aerospace ◽  
2021 ◽  
Vol 8 (1) ◽  
pp. 18
Author(s):  
Daichi Wada ◽  
Sergio A. Araujo-Estrada ◽  
Shane Windsor
Keyword(s):  
Reinforcement Learning ◽  
Wind Tunnel ◽  
Attitude Control ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Experimental Investigations ◽  
Pitch Control ◽  
Discrete Action ◽  
Control Application ◽  
Mean Square Errors

Deep reinforcement learning is a promising method for training a nonlinear attitude controller for fixed-wing unmanned aerial vehicles. Until now, proof-of-concept studies have demonstrated successful attitude control in simulation. However, detailed experimental investigations have not yet been conducted. This study applied deep reinforcement learning for one-degree-of-freedom pitch control in wind tunnel tests with the aim of gaining practical understandings of attitude control application. Three controllers with different discrete action choices, that is, elevator angles, were designed. The controllers with larger action rates exhibited better performance in terms of following angle-of-attack commands. The root mean square errors for tracking angle-of-attack commands decreased from 3.42° to 1.99° as the maximum action rate increased from 10°/s to 50°/s. The comparison between experimental and simulation results showed that the controller with a smaller action rate experienced the friction effect, and the controllers with larger action rates experienced fluctuating behaviors in elevator maneuvers owing to delay. The investigation of the effect of friction and delay on pitch control highlighted the importance of conducting experiments to understand actual control performances, specifically when the controllers were trained with a low-fidelity model.

scholarly journals Unmanned Aerial Vehicle Pitch Control under Delay Using Deep Reinforcement Learning with Continuous Action in Wind Tunnel Test

Aerospace ◽  
2021 ◽  
Vol 8 (9) ◽  
pp. 258
Author(s):  
Daichi Wada ◽  
Sergio A. Araujo-Estrada ◽  
Shane Windsor
Keyword(s):  
Neural Networks ◽  
Reinforcement Learning ◽  
Wind Tunnel ◽  
Real World ◽  
Time Delays ◽  
Wind Tunnel Test ◽  
The Real ◽  
Pitch Control ◽  
Controller Performance ◽  
The Neural Networks

Nonlinear flight controllers for fixed-wing unmanned aerial vehicles (UAVs) can potentially be developed using deep reinforcement learning. However, there is often a reality gap between the simulation models used to train these controllers and the real world. This study experimentally investigated the application of deep reinforcement learning to the pitch control of a UAV in wind tunnel tests, with a particular focus of investigating the effect of time delays on flight controller performance. Multiple neural networks were trained in simulation with different assumed time delays and then wind tunnel tested. The neural networks trained with shorter delays tended to be susceptible to delay in the real tests and produce fluctuating behaviour. The neural networks trained with longer delays behaved more conservatively and did not produce oscillations but suffered steady state errors under some conditions due to unmodeled frictional effects. These results highlight the importance of performing physical experiments to validate controller performance and how the training approach used with reinforcement learning needs to be robust to reality gaps between simulation and the real world.

Design of a Wind Tunnel Test Section for Single Airfoil Configuration With Real-Time Lift Force Control

2017 ◽  
Author(s):  
Aline Aguiar da Franca ◽  
Dirk Abel
Keyword(s):  
Wind Tunnel ◽  
Wind Turbine ◽  
Real Time ◽  
Force Control ◽  
Test Section ◽  
Lift Force ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Wiener Model ◽  
Dynamic Matrix

This article presents a concept of test section for a closed-return wind tunnel, where the lift force of an airfoil, which depends on the angle of attack, is controlled in real-time. This airfoil is used to represent a wind turbine blade. The lift force of the blades is what produces the rotor torque of the wind turbine. This torque determines the amount of energy that will be captured by the wind turbine. The linear dynamics of the motor used to change the angle of attack and the static non-linearity of the airfoil are modeled as a Wiener model. The Quadratic Dynamic Matrix Controller based on Wiener model with linearizing pre-compensation is implemented to keep the lift force constant, which is desirable to avoid mechanical loads for wind turbine applications.

The Aerodynamic Analysis for Iced Four Bundled Transmission Lines with Typical Crescent

2014 ◽  
Vol 8 (1) ◽  
pp. 84-89 ◽  
Author(s):  
Liu Yuejun ◽  
Tang Ai P. ◽  
Liu Ke T. ◽  
Tu Jie W.
Keyword(s):  
Numerical Simulation ◽  
Wind Tunnel ◽  
Transmission Lines ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Simulation Method ◽  
Single Wire ◽  
Wind Tunnel Tests ◽  
Absolute Value ◽  
Aerodynamic Test

Despite the fact that the wind tunnel tests have been carried out on iced transmission lines subjected to wind load, it is not practical to do wind tunnel tests due to its high cost. This paper describes a detailed numerical simulation method that can be used to instead of wind tunnel tests. Based on the galloping mechanism of iced transmission lines, the aerodynamic test was simulated with the typical crescent super-large thickness iced four bundled conductors. One of the results highlighted in this study is that the wind angle of attack had significant influence on the aerodynamics of iced conductors. The Den-Hartog and O.Nigol coefficient were calculated to determine galloping of iced transmission lines, comparing with the reference of wind tunnel test in the Zhejiang university, the range of the wind angle of attack to the bundled conductor which can lead to gallop is larger than single wire, but the absolute value of amplitude is less than the single conductor, split conductor is more likely to gallop than single conductor.

Wind tunnel test and numerical simulation of wind characteristics at a bridge site in mountainous terrain

2016 ◽  
Vol 20 (8) ◽  
pp. 1223-1231 ◽  
Author(s):  
Yongle Li ◽  
Xinyu Xu ◽  
Mingjin Zhang ◽  
Youlin Xu
Keyword(s):  
Wind Speed ◽  
Wind Tunnel ◽  
Bridge Deck ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Vertical Direction ◽  
Steep Slope ◽  
Mountainous Terrain ◽  
Bridge Site ◽  
Wind Characteristics

Wind tunnel test and computational fluid dynamics simulation were conducted to study the wind characteristics at a bridge site in mountainous terrain. The upstream terrains were classified into three types: open terrain, open terrain with a steep slope close to the bridge, and open terrain with a ridge close to the bridge. Results obtained from the two methods were compared, including mean speed profiles in the vertical direction and variations of wind speed and angle of attack along the bridge deck. In addition, turbulence intensities at the bridge site obtained from wind tunnel test were discussed. For mean speed profiles in the vertical direction, two methods are reasonably close for open terrain, while mountain shielding effects are evident for open terrain with a steep slope for both the methods, but the extents of effects appear different. Wind speed and angle of attack along the bridge deck are mainly influenced by the local terrain. Strong downslope wind is generated at the lee slope for the case of wind normal to top of the ridge. The comparative results are expected to provide useful references for the study of wind characteristics in mountainous terrain in the future.

scholarly journals Identification of Sharp Edge Non-Slender Delta Wing Aerodynamic Coefficient Using Neural Network

2021 ◽  
Vol 2129 (1) ◽  
pp. 012086
Author(s):  
Fahmi Izzuddin Abdul Rahman ◽  
Shabudin Mat ◽  
Nor Haizan Mohamed Radzi ◽  
Mohd Nazri Mohd Nasir ◽  
Roselina Sallehudin
Keyword(s):  
Neural Network ◽  
Wind Tunnel ◽  
Transfer Functions ◽  
Delta Wing ◽  
Flow Analysis ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Training Data ◽  
Vortical Flow ◽  
Sharp Edge

Abstract Delta wing formed a vortical flow on its surface which produced higher lift compared to conventional wing. The vortical flow is complex and non-linear which requires more studies to understand its flow physics. However, conventional flow analysis (wind tunnel test and computational flow dynamic) comes with several significant drawbacks. In recent times, application of neural network as alternative to conventional flow analysis has increased. This study is about utilization of Multi-Layer Perceptron (MLP) neural network to predict the coefficient of pressure (Cp ) on a delta wing model. The physical model that was used is a sharp edge non-slender delta wing. The training data was taken from wind tunnel tests. 70% of data is used as training, 15% is used as validation and another 15% is used as test set. The wind tunnel test was done at angle of attack from 0°-18° with increment of 3°. The flow velocity was set at 25m/s which correspond to 800,000 Reynolds number. The inputs are angle of attack and location of pressure tube (y/cr) while the output is Cp . The MLP models were fitted with 3 different transfer functions (linear, sigmoid, and tanh) and trained with Lavenberg-Marquadt backpropagation algorithm. The results of the models were compared to determine the best performing model. Results show that large amount of data is required to produce accurate prediction model because the model suffer from condition called overfitting.

Rotor aeromechanics study using two different blade property data sets

2016 ◽  
Vol 88 (6) ◽  
pp. 873-884
Author(s):  
Jae-Sang Park ◽  
Young Jung Kee
Keyword(s):  
Wind Tunnel ◽  
Natural Frequencies ◽  
Wind Tunnel Test ◽  
Test Cases ◽  
Data Sets ◽  
Elastic Deformations ◽  
Content Type ◽  
Pitch Control ◽  
Property Data ◽  
Blade Section

Purpose This paper aims to compare the comprehensive rotorcraft analyses using the two different blade section property data sets for the blade natural frequencies, airloads, elastic deformations, the trimmed rotor pitch control angles and the blade structural loads of a small-scale model rotor in a blade vortex interaction (BVI) phenomenon. Design/methodology/approach The two different blade section property data sets for the first Higher-harmonic control Aeroacoustic Rotor Test (HART-I) are considered for the present rotor aeromechanics analyses. One is the blade property data set using the predicted values which is one of the estimated data sets used for the previous validation works. The other data set uses the measured values for an uninstrumented blade. A comprehensive rotorcraft analysis code, CAMRAD II (comprehensive analytical model of rotorcraft aerodynamics and dynamics II), is used to predict the rotor aeromechanics such as the blade natural frequencies, airloads, elastic deformations, the trimmed rotor pitch control angles and the blade structural loads for the three test cases with and without higher-harmonic control pitch inputs. In CAMRAD II modelling with the two different blade property data sets, the blade is represented as a geometrically nonlinear elastic beam, and the multiple-trailer wake with consolidation model is used to consider more elaborately the BVI effect in low-speed descending flight. The aeromechanics analysis result sets using the two different blade section property data sets are compared with each other as well as are correlated with the wind-tunnel test data. Findings The predicted blade natural frequencies using the two different blade section property data sets at non-rotating condition are quite similar to each other except for the natural frequency in the fourth flap mode. However, the natural frequencies using the predicted blade properties at nominal rotating condition are lower than those with the measured blade properties except for the second lead-lag frequency. The trimmed collective pitch control angle with the predicted blade properties is higher than both the wind-tunnel test data and the result using the measured blade properties in all the three test cases. The two different blade property data sets both give reasonable predictions on the blade section normal forces with BVI in the three test cases, and the two analysis results are reasonably similar to each other. The blade elastic deformations at the tip using the measured blade properties are correlated more closely with the wind-tunnel test data than those using the predicted blade properties in most correlation examples. In addition, the predictions of blade structural loads can be slightly or moderately improved by using the measured blade properties particularly for the oscillatory flap bending moments. Finally, the movement of the sectional centre of gravity location of the uninstrumented blade has a moderate influence on the blade elastic twist at the tip in the baseline case and the oscillatory flap bending moment in the minimum noise case. Practical implications The present comparison study on rotor aeromechanics analyses using the two different blade property data sets will show the influence of blade section properties on rotor aeromechanics analysis. Originality/value This paper is the first attempt to compare the aeromechanics analysis results using the two different blade section property data sets for all three test cases (baseline, minimum noise and minimum vibration) of HART-I in low-speed descending flight.

The Evaluation of a Rigid Sail of Ship Using Wind Tunnel Test

2014 ◽  
Vol 493 ◽  
pp. 287-293
Author(s):  
Aries Sulisetyono
Keyword(s):  
Renewable Energy ◽  
Wind Tunnel ◽  
Aspect Ratio ◽  
Thrust Force ◽  
Angle Of Attack ◽  
Wind Tunnel Test ◽  
Uniform Flow ◽  
The Other ◽  
Test Results ◽  
Sailing Ship

This paper described the evaluation of rigid sail performances by using the wind tunnel test. The rigid sail models were developed in the three variations of shape geometry which was having the same value of aspect ratio. The tests were performed to investigate the performances of sail models in terms of lift, drag, resultant, driving and heeling coefficient. The three sail models were tested at the variation of angle of attack such as 15o, 20o and 25o respectively, and it was fluided by a uniform flow of air with three different speeds. The comparisons of test results were evaluated to look for which sail models had the best performance. Based on the test results, Model 2 which is a triangle shape had generated a maximal efficiency and thrust force compare to the other models.Keywords: rigid sail, wind tunnel, sailing ship, renewable energy

scholarly journals About the suitability of different numerical methods to reproduce model wind turbine measurements in a wind tunnel with high blockage ratio

2017 ◽  
Author(s):  
Annette Claudia Klein ◽  
Sirko Bartholomay ◽  
David Marten ◽  
Thorsten Lutz ◽  
George Pechlivanoglou ◽  
...  
Keyword(s):  
Wind Tunnel ◽  
Wind Turbine ◽  
Angle Of Attack ◽  
Flow Fields ◽  
Experimental Investigations ◽  
Far Field ◽  
Bending Moments ◽  
Validation Data ◽  
The University ◽  
Lifting Line

Abstract. Numerical and experimental investigations of a model wind turbine with a diameter of 3.0 m are described in the present paper. The objectives of the study are the provision of validation data, the comparison and evaluation of methods of different fidelity and the assessment of the influence of the wind tunnel walls by comparison of measurements to simulations with and without wind tunnel walls. The experiments were carried out in the large wind tunnel of the TU Berlin. With the Lifting Line Free Vortex Wake (LLFVW) code QBlade, the turbine was simulated under far field conditions at the TU Berlin. URANS simulations were performed at the University of Stuttgart with the CFD code FLOWer for far field condition to draw a comparison to QBlade. Moreover, CFD simulations of the turbine in a wind tunnel were carried out, as the walls have a significant influence on the turbine performance. Comparisons between experiment, the LLFVW code and CFD include on-blade velocities, angle of attack and bending moments. Comparisons of flow fields are drawn between experiment and the CFD code. A good accordance was achieved for the flow fields, the on-blade velocity and the angle of attack, whereas deviations occur for the bending moments.

Sign in / Sign up

Export Citation Format

Share Document