Trained Convolutional Neural Network Model Research Articles

Quality Evaluation of Road Surface Markings with Uncertainty Aware Regression and Progressive Pretraining

Maintaining high-quality road markings is essential for both safety and traffic flow. However, there has been limited research on automating the process of evaluating the quality of these markings and identifying degraded ones that need to be fixed. Our paper introduces a new approach that uses uncertainty aware (UA) regression to evaluate the quality of road surface markings. The approach is based on deep learning models and a unique training method called “progressive pretraining (PPT).” We used a dataset of RGB images which we converted to binary masks. These masks were then augmented and used to train convolutional neural networks models with a PPT strategy. The results showed that both the hybrid and UA models managed to outperform the baseline model in some metrics such as mean average error which was at 24.38% and accuracy with 81.27%. Additionally, each model showed unique strengths across various performance metrics, highlighting the efficacy of integrating uncertainty and progressive learning in quality assessment tasks. This study presents a solid proof of concept for the application of UA methods in quality evaluation tasks in general, and surface marking quality evaluation in particular.

Web-based diagnostic platform for microorganism-induced deterioration on paper-based cultural relics with iterative training from human feedback

PurposePaper-based artifacts hold significant cultural and social values. However, paper is intrinsically fragile to microorganisms, such as mold, due to its cellulose composition, which can serve as a microorganisms’ nutrient source. Mold not only can damage papers’ structural integrity and pose significant challenges to conservation works but also may subject individuals attending the contaminated artifacts to health risks. Current approaches for strain identification usually require extensive training, prolonged time for analysis, expensive operation costs, and higher risks of secondary damage due to sampling. Thus, in current conservation practices with mold-contaminated artifacts, little pre-screening or strain identification was performed before mold removal, and the cleaning techniques are usually broad-spectrum rather than strain-specific. With deep learning showing promising applications across various domains, this study investigated the feasibility of using a convolutional neural network (CNN) for fast in-situ recognition and classification of mold on paper.MethodsMolds were first non-invasively sampled from ancient Xuan Paper-based Chinese books from the Qing and Ming dynasties. Strains were identified using molecular biology methods and the four most prevalent strains were inoculated on Xuan paper to create mockups for image collection. Microscopic images of the molds as well as their stains situated on paper were collected using a compound microscope and commercial microscope lens for cell phone cameras, which were then used for training CNN models with a transfer learning scheme to perform the classification of mold. To enable involvement and contribution from the research community, a web interface that actuates the process while providing interactive features for users to learn about the information of the classified strain was constructed. Moreover, a feedback functionality in the web interface was embedded for catching potential classification errors, adding additional training images, or introducing new strains, all to refine the generalizability and robustness of the model.Results & ConclusionIn the study, we have constructed a suite of high-confidence classification CNN models for the diagnostic process for mold contamination in conservation. At the same time, a web interface was constructed that allows recurrently refining the model with human feedback through engaging the research community. Overall, the proposed framework opens new avenues for effective and timely identification of mold, thus enabling proactive and targeted mold remediation strategies in conservation.

Pre- and post-fire forest canopy height mapping in Southeast Australia through the integration of multi-temporal GEDI data, satellite images, and Convolution Neural Network

ABSTRACT This study leveraged Convolutional Neural Network (CNN) models to estimate canopy height in Southeast Australian forests before and after the 2019–2020 bushfire event, using inputs from Sentinel-1, Sentinel-2, spectral indices, and terrain features and GEDI canopy height (rh99) as target variables. Our research consisted of three primary objectives: (1) an evaluation of GEDI rh99 canopy height in relation to an nDSM derived from airborne LiDAR, (2) the development of two CNN models for pre- and post-fire scenarios, and (3) utilizing the trained CNN models to generate pre- and post-fire canopy height maps and canopy height change maps across the continuous landscape. GEDI rh99 accuracy analysis revealed R 2 = 0.85, RMSE = 7.25 m and nRMSE = 0.23 for pre-fire GEDI and R 2 = 0.78, RMSE = 10.1 m and nRMSE = 0.3 for post-fire GEDI. The pre-fire CNN model achieved R 2 = 0.75, RMSE = 7.29 m and nRMSE = 0.22, and post-fire CNN model achieved R 2 = 0.62, RMSE = 11.67 m, and nRMSE = 0.34 when validated against nDSM. Bias analysis was conducted, revealing that GEDI rh99 underestimated nDSM across all canopy height ranges, and slope has no significant impact on GEDI accuracy. Notably, post-fire GEDI rh99 exhibited significant underestimation in high to extreme fire severity. The CNN models’ prediction displayed overestimation in short forests < 20 m and increasing underestimation in taller forests > 20 m and underestimation in high and extreme fire severity. Subsequent investigation indicated that canopy layer thinning in high fire severity conditions resulted in weak GEDI waveform signals and consequent bias. Additionally, the GEDI rh99 bias was propagated to the CNN models, together with insensitivity of spectral band values to canopy height in dense and burnt forests potentially contributed to models’ bias. Finally, we demonstrated CNN models’ ability to monitor forest recovery by generating canopy height maps spanning East Gippsland from 2019 to 2023.

Composition and state prediction of lithium-ion cathode via convolutional neural network trained on scanning electron microscopy images

High-throughput materials research is strongly required to accelerate the development of safe and high energy-density lithium-ion battery (LIB) applicable to electric vehicle and energy storage system. The artificial intelligence, including machine learning with neural networks such as Boltzmann neural networks and convolutional neural networks (CNN), is a powerful tool to explore next-generation electrode materials and functional additives. In this paper, we develop a prediction model that classifies the major composition (e.g., 333, 523, 622, and 811) and different states (e.g., pristine, pre-cycled, and 100 times cycled) of various Li(Ni, Co, Mn)O2 (NCM) cathodes via CNN trained on scanning electron microscopy (SEM) images. Based on those results, our trained CNN model shows a high accuracy of 99.6% where the number of test set is 3840. In addition, the model can be applied to the case of untrained SEM data of NCM cathodes with functional electrolyte additives.

Real Time Emotion Based Music Player

Abstract: In this paper, we present the development and implementation of an Emotion-based Music Player System utilizing facial emotion data, Convolutional Neural Networks (CNN), Flask, OpenCV for face detection, and Spotify API for music playlist integration. The system aims to provide users with personalized music recommendations based on their current emotional state, detected through real-time analysis of facial expressions. The proposed system consists of several key components, including data collection, preprocessing, CNN algorithm implementation for emotion classification, integration with Flask for web application development, and real-time emotion detection using OpenCV. The trained CNN model is capable of accurately classifying emotions such as anger, disgust, fear, happiness, neutrality, sadness, and surprise. Results from the implementation demonstrate the system's effectiveness in accurately detecting emotions from facial expressions and providing corresponding music recommendations. The CNN model achieved a final accuracy of 62.44% after 10 epochs of training. Realtime emotion detection using OpenCV successfully identifies facial expressions, allowing for dynamic adjustments to the music playlist. Overall, the Emotion-based Music Player System presents a novel approach to enhancing user experience by leveraging facial emotion data and advanced machine learning techniques to deliver personalized music recommendations tailored to individual emotional states.

Deep learning approaches to recover the plasma current density profile from the safety factor based on Grad–Shafranov solutions across multiple tokamaks

Many magnetohydrodynamic stability analyses require generation of a set of equilibria with a fixed safety factor q-profile while varying other plasma parameters. A neural network (NN)-based approach is investigated that facilitates such a process. Both multilayer perceptron (MLP)-based NN and convolutional neural network (CNN) models are trained to map the q-profile to the plasma current density J-profile, and vice versa, while satisfying the Grad–Shafranov radial force balance constraint. When the initial target models are trained, using a database of semi-analytically constructed numerical equilibria, an initial CNN with one convolutional layer is found to perform better than an initial MLP model. In particular, a trained initial CNN model can also predict the q- or J-profile for experimental tokamak equilibria. The performance of both initial target models is further improved by fine-tuning the training database, i.e. by adding realistic experimental equilibria with Gaussian noise. The fine-tuned target models, referred to as fine-tuned MLP and fine-tuned CNN, well reproduce the target q- or J-profile across multiple tokamak devices. As an important application, these NN-based equilibrium profile convertors can be utilized to provide a good initial guess for iterative equilibrium solvers, where the desired input quantity is the safety factor instead of the plasma current density.

Translating Test Responses to Images for Test-termination Prediction via Multiple Machine Learning Strategies

Failure diagnosis is a software-based data-driven procedure. Collecting an excessive amount of fail data not only increases the overall test cost but can also potentially reduce diagnostic resolution. Test-termination prediction is thus proposed to dynamically determine which failing test pattern to terminate testing, producing an amount of test data that is sufficient for an accurate diagnosis analysis. In this work, we describe a set of novel methods utilizing advanced machine learning techniques for efficient test-termination prediction. To implement this approach, we first generate images representing failing test responses from failure-log files. These images are then used to train a multi-layer convolutional neural network (CNN) incorporating a residual block. The trained CNN model leverages the images and known diagnostic results to determine the optimal test-termination strategy within the testing process, ensuring efficient and high-quality diagnosis. In addition to the integration of test response-to-image translation, our approach harnesses two cutting-edge learning strategies to enhance fail data and boost performance in subsequent tasks. The first strategy is transfer learning, which utilizes sample-label information from one circuit to guide the decision of whether to continue or stop testing for another circuit lacking labels. The second strategy involves the use of a generative deep model to generate fail data in the form of synthetic images. This technique increases the modeling effectiveness by expanding the volume of training samples. Experimental results conducted on actual failing chips and standard benchmarks validate that our proposed method surpasses existing approaches. Our method creates opportunities to harness the power of recent advances in machine learning for improving test and diagnosis efficiency.

Detection Of Deep Fakes Using CNN

The rise of technology has emphasized the importance of identifying altered images and videos to preserve trust and authenticity in digital content. Convolutional Neural Networks (CNNs) have emerged as tools, for tasks like detecting fakes. This study introduces a method for identifying fakes using CNNs. Our strategy involves training a CNN model on a dataset containing both synthetic visuals. The network is structured to learn features that differentiate between manipulated content. We use a mix of layers pooling layers and connected layers to extract and process features from the input data. To strengthen the reliability of our model we employ techniques like data augmentation and transfer learning. Data augmentation includes applying transformations such as rotation, scaling and cropping to the training data to enhance its variety. Transfer learning allows us to utilize trained CNN models and adjust them for deep fake detection purposes. We assess the effectiveness of our approach on datasets. Compare it with existing techniques. Our experiments show that our CNN based method achieves accuracy in spotting fake content in different scenarios. Additionally we examine how well the model withstands attacks and variations, in input data quality.

Damage Identification of Simple Supported Bridges Under Moving Loads Based on Variational Mode Decomposition and Deep Learning

Aiming at rapid and economical damage detection of a large number of simple supported bridges, a new structural damage identification method under moving load based on variational mode decomposition (VMD) and deep learning is proposed. Firstly, a moving vehicle is used as an exciting load to invoke structural damage feature and enhance the signal-to-noise ratio, and the structural vertical acceleration response is extracted by a finite element simulating analysis under various damage cases. In order to simulate the influence of noise and expand the samples, Gaussian white noise is added to the extracted data, and then the response signal is decomposed into a series of intrinsic mode functions (IMFs) using VMD, and the optimal IMF component is selected as the damage sample of the structure. Then, a one-dimensional convolutional neural network (CNN) model is built and trained by the various samples of damage. The vibration response of the practical bridge is processed and inputted by the trained CNN model to identify the location of the damage and degree of the structure. Finally, the effectiveness and anti-noise performance of the proposed method are verified through numerical analysis and a simply supported beam bridge model experiment. The results show that the average identification accuracy of the numerical simulations and experimental is 93.4% and 86.8% with 20% Gaussian white noise, respectively. Sensors at different locations have almost the same identification effect for various cases of damage, so it is possible to identify structural damage only using a small amount of accelerometer.

A Real-world Toxicity Atlas Shows that Adverse Events of Combination Therapies Commonly Result in Additive Interactions.

Combination therapies are a promising approach for improving cancer treatment, but it is challenging to predict their resulting adverse events in a real-world setting. We provide here a proof-of-concept study using 15 million patient records from the FDA Adverse Event Reporting System (FAERS). Complex adverse event frequencies of drugs or their combinations were visualized as heat maps onto a two-dimensional grid. Adverse event frequencies were shown as colors to assess the ratio between individual and combined drug effects. To capture these patterns, we trained a convolutional neural network (CNN) autoencoder using 7,300 single-drug heat maps. In addition, statistical synergy analyses were performed on the basis of BLISS independence or χ2 testing. The trained CNN model was able to decode patterns, showing that adverse events occur in global rather than isolated and unique patterns. Patterns were not likely to be attributed to disease symptoms given their relatively limited contribution to drug-associated adverse events. Pattern recognition was validated using trial data from ClinicalTrials.gov and drug combination data. We examined the adverse event interactions of 140 drug combinations known to be avoided in the clinic and found that near all of them showed additive rather than synergistic interactions, also when assessed statistically. Our study provides a framework for analyzing adverse events and suggests that adverse drug interactions commonly result in additive effects with a high level of overlap of adverse event patterns. These real-world insights may advance the implementation of new combination therapies in clinical practice.

Development of a convolutional neural network based geomechanical upscaling technique for heterogeneous geological reservoir

Geomechanical assessment using coupled reservoir-geomechanical simulation is becoming increasingly important for analyzing the potential geomechanical risks in subsurface geological developments. However, a robust and efficient geomechanical upscaling technique for heterogeneous geological reservoirs is lacking to advance the applications of three-dimensional (3D) reservoir-scale geomechanical simulation considering detailed geological heterogeneities. Here, we develop convolutional neural network (CNN) proxies that reproduce the anisotropic nonlinear geomechanical response caused by lithological heterogeneity, and compute upscaled geomechanical properties from CNN proxies. The CNN proxies are trained using a large dataset of randomly generated spatially correlated sand-shale realizations as inputs and simulation results of their macroscopic geomechanical response as outputs. The trained CNN models can provide the upscaled shear strength (R2 > 0.949), stress-strain behavior (R2 > 0.925), and volumetric strain changes (R2 > 0.958) that highly agree with the numerical simulation results while saving over two orders of magnitude of computational time. This is a major advantage in computing the upscaled geomechanical properties directly from geological realizations without the need to perform local numerical simulations to obtain the geomechanical response. The proposed CNN proxy-based upscaling technique has the ability to (1) bridge the gap between the fine-scale geocellular models considering geological uncertainties and computationally efficient geomechanical models used to assess the geomechanical risks of large-scale subsurface development, and (2) improve the efficiency of numerical upscaling techniques that rely on local numerical simulations, leading to significantly increased computational time for uncertainty quantification using numerous geological realizations.

Monitoring manufacturing systems using AI: A method based on a digital factory twin to train CNNs on synthetic data

Modern cyber–physical production systems provide advanced solutions to enhance factory throughput and efficiency. However, monitoring its behaviour and performance becomes challenging as the complexity of a manufacturing system increases. Artificial Intelligence (AI) provides techniques to manage not only decision-making tasks but also to support monitoring. The integration of AI into a factory can be facilitated by a reliable Digital Twin (DT) that enables knowledge-based and data-driven approaches. While computer vision and convolutional neural networks (CNNs) are crucial for monitoring production systems, the need for extensive training data hinders their adoption in real factories. The proposed methodology leverages the Digital Twin of a factory to generate labelled synthetic data for training CNN-based object detection models. Regarding their position and state, the focus is on monitoring entities in manufacturing systems, such as parts, components, fixtures, and tools. This approach reduces the need for large training datasets and enables training when the actual system is unavailable. The trained CNN model is evaluated in various scenarios, with a real case study involving an industrial pilot plant for repairing and recycling Printed Circuit Boards (PCBs).

Alzheimer's MRI Recognition Based on Convolutional Neural Networks

Alzheimer's disease (AD) is a progressively worsening neurodegenerative disorder, where early diagnosis is crucial for implementing effective disease management strategies. In multimodal diagnostic pathways, Magnetic Resonance Imaging (MRI) plays a pivotal role as a non-invasive imaging technique in disease identification and progression monitoring. Deep learning, particularly Convolutional Neural Networks (CNN), can extract key biomarkers from complex imaging data. By training CNN models to automatically interpret MRI scans, radiology experts can utilize these advanced analytical tools for more efficient and consistent pathological assessments, thereby enhancing clinical decision support systems. This study develops a deep learning-based method for the automatic classification of Alzheimer's brain MRI images, using convolutional neural networks to categorize MRI images into four stages: no dementia, very mild dementia, mild dementia, and moderate dementia. A CNN model is constructed, learning distinctive features of the images through multi-level feature extraction and performing feature map visualization. Early stopping is employed to prevent overfitting. The model is trained on a training set and evaluated on a test set, with performance metrics including confusion matrix, accuracy, precision, recall, F1 score, Kappa coefficient, Matthew’s coefficient, ROC curve with AUC value, and PQ curve with AP value. The results show that the proposed model effectively differentiates between various categories of MRI images, providing valuable tools for early diagnosis and condition monitoring.

A review of convolutional neural networks in computer vision

In computer vision, a series of exemplary advances have been made in several areas involving image classification, semantic segmentation, object detection, and image super-resolution reconstruction with the rapid development of deep convolutional neural network (CNN). The CNN has superior features for autonomous learning and expression, and feature extraction from original input data can be realized by means of training CNN models that match practical applications. Due to the rapid progress in deep learning technology, the structure of CNN is becoming more and more complex and diverse. Consequently, it gradually replaces the traditional machine learning methods. This paper presents an elementary understanding of CNN components and their functions, including input layers, convolution layers, pooling layers, activation functions, batch normalization, dropout, fully connected layers, and output layers. On this basis, this paper gives a comprehensive overview of the past and current research status of the applications of CNN models in computer vision fields, e.g., image classification, object detection, and video prediction. In addition, we summarize the challenges and solutions of the deep CNN, and future research directions are also discussed.

Abstract 7169: A real-world toxicity-atlas shows that adverse events of combination-therapies commonly result in additive interactions

Abstract Purpose: Combination-therapies are a promising approach for improving cancer treatment, but it is challenging to predict their resulting adverse events in an real-world setting. Experimental Design: We provide here a proof-of-concept study using 15 million patient records from the US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS). Complex adverse-event frequencies of drugs or their combinations were visualized as heatmaps onto a 2D UMAP grid. Adverse event frequencies were shown as colors to assess the ratio between individual and combined drug effects. To capture these patterns, we trained a convolutional neural network (CNN) autoencoder using 7,300 single-drug heatmaps. Additionally, statistical synergy analyses were performed based on BLISS independence or Chi-square testing. Results: The trained CNN model was able to decode patterns, showing that adverse events occur in global rather than isolated and unique patterns. Patterns were not likely to be attributed to disease symptoms given their relatively limited contribution to drug associated adverse events. Pattern recognition was validated using trial data from ClinicalTrials.gov and drug combination data. We examined the adverse-event interactions of 140 drug combinations known to be avoided in the clinic and found that near all of them showed additive rather than synergistic interactions, also when assessed statistically. Conclusion: Our study provides a framework for analyzing adverse events and suggest that adverse drug interactions commonly result in additive effects with a high level of overlap of adverse event patterns. These real-world insights may advance the implementation of new combination therapies in clinical practice. Citation Format: Asli Küçükosmanoglu, Silvia I. Scoarta, Bart Westerman. A real-world toxicity-atlas shows that adverse events of combination-therapies commonly result in additive interactions [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7169.

Reusing Convolutional Neural Network Models through Modularization and Composition

With the widespread success of deep learning technologies, many trained deep neural network (DNN) models are now publicly available. However, directly reusing the public DNN models for new tasks often fails due to mismatching functionality or performance. Inspired by the notion of modularization and composition in software reuse, we investigate the possibility of improving the reusability of DNN models in a more fine-grained manner. Specifically, we propose two modularization approaches named CNNSplitter and GradSplitter, which can decompose a trained convolutional neural network (CNN) model for N -class classification into N small reusable modules. Each module recognizes one of the N classes and contains a part of the convolution kernels of the trained CNN model. Then, the resulting modules can be reused to patch existing CNN models or build new CNN models through composition. The main difference between CNNSplitter and GradSplitter lies in their search methods: the former relies on a genetic algorithm to explore search space, while the latter utilizes a gradient-based search method. Our experiments with three representative CNNs on three widely used public datasets demonstrate the effectiveness of the proposed approaches. Compared with CNNSplitter, GradSplitter incurs less accuracy loss, produces much smaller modules (19.88% fewer kernels), and achieves better results on patching weak models. In particular, experiments on GradSplitter show that (1) by patching weak models, the average improvement in terms of precision, recall, and F1-score is 17.13%, 4.95%, and 11.47%, respectively, and (2) for a new task, compared with the models trained from scratch, reusing modules achieves similar accuracy (the average loss of accuracy is only 2.46%) without a costly training process. Our approaches provide a viable solution to the rapid development and improvement of CNN models.

A comparative analysis of damage identification methods based on multi-channel data fusion and convolutional neural network for frame structures

Vibration-based damage identification methods are crucial for structural health monitoring. In recent years, deep learning methods, such as convolutional neural networks (CNNs), have been widely used for damage identification due to their superior performance. To enhance the accuracy of damage identification for complex frame structures and effectively utilize data measured from multiple sensors, this study proposes and compares four different damage identification methods based on CNN and four different kinds of multi-channel data fusion approaches, i.e., data splicing, matrix reconstruction, data dimension elevation, and sub-model integration. Firstly, the original one-dimensional vibration signals of the frame structures are measured. Then, multi-channel data fusion is performed on the signals to construct datasets used as input to train the CNN model. Finally, the sensitive damage features are automatically extracted using the trained CNN models after parameter tuning, and the structural damage patterns are identified. The effectiveness of these damage identification methods is evaluated using the numerical model of IASC-ASCE Benchmark structure, the Qatar University Grandstand Simulator test, and a three-layer stainless steel frame structure test. The efficiency of damage identification methods using multi-channel data is compared to those using only single-channel data. Additionally, the impact of noise on damage identification accuracy is also analyzed. The results demonstrate that the proposed four different damage identification methods based on CNN and multi-channel data fusion, outperform the methods that solely rely on single-channel data. Furthermore, these methods significantly enhance the accuracy of damage identification.

SeptoSympto: a precise image analysis of Septoria tritici blotch disease symptoms using deep learning methods on scanned images

BackgroundInvestigations on plant-pathogen interactions require quantitative, accurate, and rapid phenotyping of crop diseases. However, visual assessment of disease symptoms is preferred over available numerical tools due to transferability challenges. These assessments are laborious, time-consuming, require expertise, and are rater dependent. More recently, deep learning has produced interesting results for evaluating plant diseases. Nevertheless, it has yet to be used to quantify the severity of Septoria tritici blotch (STB) caused by Zymoseptoria tritici—a frequently occurring and damaging disease on wheat crops.ResultsWe developed an image analysis script in Python, called SeptoSympto. This script uses deep learning models based on the U-Net and YOLO architectures to quantify necrosis and pycnidia on detached, flattened and scanned leaves of wheat seedlings. Datasets of different sizes (containing 50, 100, 200, and 300 leaves) were annotated to train Convolutional Neural Networks models. Five different datasets were tested to develop a robust tool for the accurate analysis of STB symptoms and facilitate its transferability. The results show that (i) the amount of annotated data does not influence the performances of models, (ii) the outputs of SeptoSympto are highly correlated with those of the experts, with a similar magnitude to the correlations between experts, and (iii) the accuracy of SeptoSympto allows precise and rapid quantification of necrosis and pycnidia on both durum and bread wheat leaves inoculated with different strains of the pathogen, scanned with different scanners and grown under different conditions.ConclusionsSeptoSympto takes the same amount of time as a visual assessment to evaluate STB symptoms. However, unlike visual assessments, it allows for data to be stored and evaluated by experts and non-experts in a more accurate and unbiased manner. The methods used in SeptoSympto make it a transferable, highly accurate, computationally inexpensive, easy-to-use, and adaptable tool. This study demonstrates the potential of using deep learning to assess complex plant disease symptoms such as STB.

Localization method of subsynchronous oscillation source based on high-resolution time-frequency distribution image and CNN

The penetration of new energy sources such as wind power is increasing, which consequently increases the occurrence rate of subsynchronous oscillation events. However, existing subsynchronous oscillation source-identification methods primarily analyze fixed-mode oscillations and rarely consider time-varying features, such as frequency drift, caused by the random volatility of wind farms when oscillations occur. This paper proposes a subsynchronous oscillation source-localization method that involves an enhanced short-time Fourier transform and a convolutional neural network (CNN). First, an enhanced STFT is performed to secure high-resolution time-frequency distribution (TFD) images from the measured data of the generation unit ports. Next, these TFD images are amalgamated to form a subsynchronous oscillation feature map that serves as input to the CNN to train the localization model. Ultimately, the trained CNN model realizes the online localization of subsynchronous oscillation sources. The effectiveness and accuracy of the proposed method are validated via multimachine system models simulating forced and natural oscillation events using the Power Systems Computer Aided Design platform. Test results show that the proposed method can localize subsynchronous oscillation sources online while considering unpredictable fluctuations in wind farms, thus providing a foundation for oscillation suppression in practical engineering scenarios.

Study on the Consistency Between Automatic Measurement Based on Convolutional Neural Network Technology and Manual Visual Evaluation in Intracavitary Ultrasonic Cine of Anterior Pelvic.

This study was to evaluate the application of automatic measurement based on convolutional neural network (CNN) technology in intracavitary ultrasound cine of anterior pelvic. A total of 500 patients who underwent pelvic floor ultrasound examination at Peking University Shenzhen Hospital from July 2021 to February 2022 were retrospectively retrieved by the picture archiving and communication system (PACS) system, and 300 cases were used as a training set. The training set was labeled by three experienced ultrasound physicians to train CNN models and develop an automatic measurement software. The remaining 200 cases were used as a test set. Automatic measurement software identified relevant anatomical structures frame by frame and determined the two frames with the greatest difference, calculated the bladder neck descent (BND), urethral rotation angle (URA), and retrovesical angle (RA). Meanwhile, two experienced ultrasound physicians evaluated the resting frame and the maximum Valsalva frame on the cines by manual visual evaluation, labeled the anatomical structures in the corresponding frame, such as the inferoposterior margin of pubic symphysis, the mid-axis of pubic symphysis, bladder contour, and urethra in the front, and calculated BND, URA, and RA. Considering that the residual urine volume (RUV) in the bladder may affect the results, enrolled patients were grouped according to the RUV (10-50 mL, 50-100 mL, and >100 mL). The consistency of the results by automatic measurement and manual visual evaluation was evaluated using the intraclass correlation coefficient (ICC) and the Bland-Altman graph. Of the 200 cases in the test set, 120 cases were successfully identified by the CNN automatic software with a 60% recognition rate. In the case of successful identification, the ICC of manual visual evaluation measurement and automatic measurement was 0.936 (BND), 0.911 (URA), 0.756 (RA in rest), and 0.877 (RA at maximum Valsalva), respectively. In addition, the RUV had a negligible effect on the consistency. The Bland-Altman plot shows the proportion of samples outside the limit was below 5%. CNN-based automatic measurement software exhibited high reliability in anterior pelvic measurement, which results in a significantly enhanced measurement efficiency.