Figure 9 - uploaded by Jeanie A. Aird
Content may be subject to copyright.
Analysis of model failure modes when identifying and characterizing (a) wakes, (b) wake fragments and (c) obscured wakes. Values are for the testing data subset. Pie charts depict the failure modes as a percentage of total number of failures for fully visible and obscured wakes and wake fragments. Note: only four obscured wake fragments are included in the testing set and all four were identified successfully.
Source publication
We present a proof of concept of wind turbine wake identification and characterization using a region-based convolutional neural network (CNN) applied to lidar arc scan images taken at a wind farm in complex terrain. We show that the CNN successfully identifies and characterizes wakes in scans with varying resolutions and geometries, and can captur...
Contexts in source publication
Context 1
... all four wake fragments that are embedded in a highly heterogeneous data field as a result of the removal of many radial velocity estimates because of CNR filtering are correctly identified. The most common mode of failure for the CNN is a false negative-i.e., there is a wake or wake fragment in the scan and the model does not identify it as an object at all (Figure 9). The second most common mode of failure is returning a false positive-i.e., a non-wake (such as a wind turbine) structure is identified as a wake. ...
Context 2
... most common mode of failure for the CNN is a false negative-i.e., there is a wake or wake fragment in the scan and the model does not identify it as an object at all (Figure 9). The second most common mode of failure is returning a false positive-i.e., a non-wake (such as a wind turbine) structure is identified as a wake. ...
Context 3
... scan from Region B exhibits a larger area scanned than that of Region A because of a higher maximum range distance, which reduces the visibility of the wakes in the image (Figure 10c,d). Regardless, the CNN is able to successfully identify the wakes Figure 9. Analysis of model failure modes when identifying and characterizing (a) wakes, (b) wake fragments and (c) obscured wakes. ...
Citations
... Aird, J. A et al. [42] the first use of a Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization (R-CNNWTWC) using lidar arc scan data from a wind farm on hilly terrain. Numerical models of the shape, geographical extent, and fragment placements of the wake correspond quite closely with the results of visual inspections. ...
Automatic emotion identification from speech is a difficult problem that significantly depends on the accuracy of the speech characteristics employed for categorization. The display of emotions seen in human speech is inherently integrated with hidden representations of several dimensions and the fundamentals of human behaviour. This illustrates the significance of using auditory data gathered from discussions between people to determine people's emotions. In order to engage with people more closely, next-generation artificial intelligence will need to be able to recognize and express emotional states. Even though recovery of emotions from verbal descriptions of human interactions has shown promising outcomes, the accuracy of auditory feature-based emotion recognition from speech is still lacking. This paper suggests a unique method for Speech-based Emotion Recognition (SER) that makes use of Improved and a Faster Region-based Convolutional Neural Network (IFR-CNN). IFR-CNN employs Improved Intersection over Unification (IIOU) in the positioning stage with better loss function for improving Regions of Interest (RoI). With the help of a Recurrent Neural Network (RNN)-based model that considers both the dialogue structure and the unique emotional states; modern categorical emotion forecasts may be created quickly. In particular, IFR-CNN was developed to learn and store affective states, as well as track and recover speech properties. The effectiveness of the proposed method is evaluated with the help of real-time prediction capabilities, empirical evaluation, and benchmark datasets. From the speech dataset, we have extracted the Mel frequency cepstral coefficients (MFCC), as well as spectral characteristics and temporal features. Emotion recognition using retrieved information is the goal of the IFR-development. Quantitative analysis on two datasets, the Berlin Database of Emotional Speech (EMODB) and the Serbian Emotional Speech Database (GEES), revealed encouraging results. Specifically, for the EMODB, which represents 7 emotions, the IFR-CNN attained an accuracy of 89.5%. For the GEES dataset, which covers 5 emotions, the accuracy stood at 94.82%. These outcomes suggest that the proposed IFR-CNN method offers a significant improvement over existing models in emotion recognition from speech.
... Convolutional neural networks (CNN) have demonstrated wide success for use in image classification and segmentation tasks, such as facial recognition and object detection in videos and images [64][65][66]. CNNs have been implemented in a wide variety of applications, such as detecting cancerous cells in medical imaging, automated quantification of wind turbine wakes from doppler lidar scans, and object detection from geospatial data such as high-resolution satellite images [67][68][69][70]. ...
... Since the development of Mask R-CNN, studies have shown improvement in average precision values through alteration of the network architecture. Results are improved here by the use of a feature pyramid network (FPN) backbone [73] to enhance CNN performance through improved accuracy in segmentation of objects of varying sizes within the images (particularly important for the shallow damage class which is notably smaller than the deep damage class- Figure 2) and also the use of transfer learning to initialize the CNN weights before training (as in [68,69], weights derived from training the model on the MS COCO dataset are utilized to initialize the neural network). ...
... Previous studies have exhibited sensitivity in CNN results to object orientation, particularly when a robust training dataset of objects in varying orientations is unavailable or limited [68,69]. Further, previous studies conducted during model development in [68,69] indicated that colormap choice and inclusion of color hue (i.e., color images compared to grayscale images) improves CNN results for certain applications. Thus, lowest CNN accuracy/precision rates in LEE quantification and classification may be expected for the CNN trained and tested on black and white, unrotated images ( , and highest accuracy/precision rates may be expected for the CNN trained and tested on color images rotated such that the Eighty images are used for training the Mask R-CNN model, thirty for validation to prevent overfitting and thirty for testing. ...
Wind turbine blade leading edge erosion is a major source of power production loss and early detection benefits optimization of repair strategies. Two machine learning (ML) models are developed and evaluated for automated quantification of the areal extent, morphology and nature (deep, shallow) of damage from field images. The supervised ML model employs convolutional neural networks (CNN) and learns features (specific types of damage) present in an annotated set of training images. The unsupervised approach aggregates pixel intensity thresholding with calculation of pixel-by-pixel shadow ratio (PTS) to independently identify features within images. The models are developed and tested using a dataset of 140 field images. The images sample across a range of blade orientation, aspect ratio, lighting and resolution. Each model (CNN v PTS) is applied to quantify the percent area of the visible blade that is damaged and classifies the damage into deep or shallow using only the images as input. Both models successfully identify approximately 65% of total damage area in the independent images, and both perform better at quantifying deep damage. The CNN is more successful at identifying shallow damage and exhibits better performance when applied to the images after they are preprocessed to a common blade orientation.
... Wind power has become an important source of global renewable energy [1,2]. As wind turbines (WTs) often fail in extreme environments, including sleet, wind gusts, and lightning strikes [3], wind turbine blades (WTBs) monitoring, such as fault prognostics, health monitoring, and early failure warning, etc., is deemed an important task to ensure their maintenance of normal operation [4][5][6][7][8]. Since machine vision techniques have shown great advantages in object detection and recognition, installing visual systems onboard unmanned aerial vehicles (UAVs) is a promising labor-saving and remote sensing approach for WTB surface inspection [5,9]. ...
Unmanned air vehicle (UAV) based imaging has been an attractive technology to be used for wind turbine blades (WTBs) monitoring. In such applications, image motion blur is a challenging problem which means that motion deblurring is of great significance in the monitoring of running WTBs. However, an embarrassing fact for these applications is the lack of sufficient WTB images, which should include better pairs of sharp images and blurred images captured under the same conditions for network model training. To overcome the challenge of image pair acquisition, a training sample synthesis method is proposed. Sharp images of static WTBs were first captured, and then video sequences were prepared by running WTBs at different speeds. The blurred images were identified from the video sequences and matched to the sharp images using image difference. To expand the sample dataset, rotational motion blurs were simulated on different WTBs. Synthetic image pairs were then produced by fusing sharp images and images of simulated blurs. Finally, a total of 4000 image pairs were obtained. To conduct motion deblurring, a hybrid deblurring network integrated with DeblurGAN and DeblurGANv2 was deployed. The results show that the integration of DeblurGANv2 and Inception-ResNet-v2 provides better deblurred images, in terms of both metrics of signal-to-noise ratio (80.138) and structural similarity (0.950) than those obtained from the comparable networks of DeblurGAN and MobileNet-DeblurGANv2.
A convolutional neural network is applied to lidar scan images from three experimental campaigns to identify and characterize wind turbine wakes. Initially developed as a proof-of-concept model and applied to a single data set in complex terrain, the model is now improved and generalized and applied to two other unique lidar data sets, one located near an escarpment and one located offshore. The model, initially developed using lidar scans collected in predominantly westerly flow, exhibits sensitivity to wind flow direction. The model is thus successfully generalized through implementing a standard rotation process to scan images before input into the convolutional neural network to ensure the flow is westerly. The sample size of lidar scans used to train the model is increased, and along with the generalization process, these changes to the model are shown to enhance accuracy and robustness when characterizing dissipating and asymmetric wakes. Applied to the offshore data set in which nearly 20 wind turbine wakes are included per scan, the improved model exhibits a 95% success rate in characterizing wakes and a 74% success rate in characterizing dissipating wake fragments. The improved model is shown to generalize well to the two new data sets, although an increase in wake characterization accuracy is offset by an increase in model sensitivity and false positive wake identifications.
A convolutional neural network (CNN) autoencoder model has been developed to generate 3D realizations of time-averaged velocity in the wake of the wind turbines at the Sandia National Laboratories Scaled Wind Farm Technology (SWiFT) facility. Large-eddy simulations (LES) of the SWiFT site are conducted using an actuator surface model to simulate the turbine structures to produce training and validation datasets of the CNN. The simulations are validated using the SpinnerLidar measurements of turbine wakes at the SWiFT site and the instantaneous and time-averaged velocity fields from the training LES are used to train the CNN. The trained CNN is then applied to predict 3D realizations of time-averaged velocity in the wake of the SWiFT turbines under flow conditions different than those for which the CNN was trained. LES results for the validation cases are used to evaluate the performance of the CNN predictions. Comparing the validation LES results and CNN predictions, we show that the developed CNN autoencoder model holds great potential for predicting time-averaged flow fields and the power production of wind turbines while being several orders of magnitude computationally more efficient than LES.
