Percentile Error Research Articles

Collocating wind data: A case study on the verification of the CERRA dataset

Abstract Wind resource assessment often relies on reanalysis data or meso-scale models due to the challenges and costs associated with on-site measurement campaigns, especially in offshore locations. While these datasets offer extensive spatial coverage, they often provide the desired parameters on a coarse grid only. The spatial averaging over the grid and the low frequency of the outputs hinders the model’s ability to capture the short-term wind speed variability and events with a time scale of a few minutes to a few dozen of minutes. In this paper we investigate the temporal scales of the wind events resolved by the Copernicus European Regional Re-Analysis (CERRA) dataset, ERA5, and an in-house regional down-scaling of ERA5 using the WRF model by comparing them against measurements. We propose a method to collocate the two sources of wind data (measured and modeled), focusing on the verification of CERRA with respect to first and second-order statistics in the North Sea. The results highlight the superiority of CERRA compared to ERA5 in estimating the wind speed distribution, improving the 95th percentile error from -0.85 to -0.19 ms−1 and the average bias from -0.286 to -0.004 m s−1 across all locations. The wind speed change over one hour is underestimated by both CERRA and ERA5 when collocated with 10-minute measurements at the full hour. However, hourly averaged measurements align well with CERRA’s wind speed variation predictions, while an averaging window of 3.5 hours is needed to align with ERA5’s hourly wind speed changes. The results showed that the synthetic wind speeds exhibit optimal correlation with the measurements when the measurements are averaged over 4.5 to 7 hours, depending on the modeled dataset. This time scale corresponds to a length scale of 160-250 km, assuming a mean wind speed of 10 m s−1, which falls within the meso-scale range. The study contributes to the understanding of model validation in offshore wind energy studies, with the potential to enhance wind resource assessment calculations and improve the results of long-term extrapolations.

A Global Evaluation of Radar‐Derived Digital Elevation Models: SRTM, NASADEM, and GLO‐30

AbstractThis study evaluates global radar‐derived digital elevation models (DEMs), namely the Shuttle Radar Topography Mission (SRTM), NASADEM and GLO‐30 DEMs. We evaluate their accuracy over bare‐earth terrain and characterize elevation biases induced by forests using global Lidar measurements from the Ice, Cloud, and Land Elevation Satellite (ICESat)'s Geoscience Laser Altimeter System (GLAS), the Global Ecosystem Dynamics Investigation (GEDI) and the ICESat‐2 Advanced Topographic Laser Altimeter System (ATLAS) instruments collected on locally flat terrain. Our analysis is based on error statistics calculated for each DEM tile, which are then summarized as global error percentiles, providing a regional characterization of DEM quality. We find NASADEM to be a significant improvement upon the SRTM V3. Over bare ground areas, the mean elevation bias and root mean square error (RMSE) improved from 0.68 to 2.50 m respectively to 0.00 and 1.5 m as compared to ICESat/GLAS. GLO‐30 is more accurate with bare ground elevation bias and RMSE were below 0.05 and 0.55 m. Similar improvements were observed when compared to GEDI and ICESat‐2 measurements. The DEM biases associated with the presence of vegetation vary linearly with canopy height, and more closely follow the percentile of Lidar Relative Height (RH50). Other factors such as canopy density, radar frequency and Lidar technology also contribute to observed elevation biases. This global analysis highlights the potential of various technologies for mapping of Earth's topography, and the need for more advanced remote sensing observations that can resolve vegetation structure and sub‐canopy ground elevation.

SWiLoc: Fusing Smartphone Sensors and WiFi CSI for Accurate Indoor Localization.

Dead reckoning is a promising yet often overlooked smartphone-based indoor localization technology that relies on phone-mounted sensors for counting steps and estimating walking directions, without the need for extensive sensor or landmark deployment. However, misalignment between the phone's direction and the user's actual movement direction can lead to unreliable direction estimates and inaccurate location tracking. To address this issue, this paper introduces SWiLoc (Smartphone and WiFi-based Localization), an enhanced direction correction system that integrates passive WiFi sensing with smartphone-based sensing to form Correction Zones. Our two-phase approach accurately measures the user's walking directions when passing through a Correction Zone and further refines successive direction estimates outside the zones, enabling continuous and reliable tracking. In addition to direction correction, SWiLoc extends its capabilities by incorporating a localization technique that leverages corrected directions to achieve precise user localization. This extension significantly enhances the system's applicability for high-accuracy localization tasks. Additionally, our innovative Fresnel zone-based approach, which utilizes unique hardware configurations and a fundamental geometric model, ensures accurate and robust direction estimation, even in scenarios with unreliable walking directions. We evaluate SWiLoc across two real-world environments, assessing its performance under varying conditions such as environmental changes, phone orientations, walking directions, and distances. Our comprehensive experiments demonstrate that SWiLoc achieves an average 75th percentile error of 8.89 degrees in walking direction estimation and an 80th percentile error of 1.12 m in location estimation. These figures represent reductions of 64% and 49%, respectively for direction and location estimation error, over existing state-of-the-art approaches.

A correlation for U-value for laminar and turbulent flows in concentric tube heat exchangers

This paper introduces a novel empirical correlation designed to predict the overall heat transfer coefficient (U) in concentric tube heat exchangers. The study utilizes a comprehensive dataset of 2700 data points from CFD simulations, examining the impact of critical variables including hot and cold fluid Reynolds numbers (Reh, Rec), Prandtl numbers (Prh, Prc), and heat exchanger dimensions (tube hydraulic diameter Dh, annular space hydraulic diameter Dc, and overall length L). The analysis incorporates a range of fluid combinations in the tube and annular sections – air–air, air–water, water-air, and water–water – across Reynolds number ranges from 1000 to 20000, covering both laminar and turbulent flow regimes. The correlation was developed using the Adam optimizer to minimize the Weighted Mean Squared Error (WMSE), achieving an impressive convergence to an average prediction error of 6.7% compared to actual U values. The model categorizes flow into four distinct categories, each corresponding to a combination of flow patterns in the tube and annular spaces regimes – Laminar–Laminar, Laminar–Turbulent, Turbulent–Laminar, and Turbulent–Turbulent – integrating both hydrodynamic and thermal entry effects to enhance prediction accuracy. Rigorous error analysis revealed a median error of 5.18%, substantially outperforming existing models. Notably, 10% of the predictions deviate by less than 1%, with 90th percentile errors staying below 15%. Parameters were finely tuned through extensive numerical simulations, ensuring accurate application of the correlation across diverse operational conditions. The study concludes by highlighting the model’s potential to significantly enhance the design and efficiency of heat exchangers and proposes future enhancements to further improve its predictive accuracy.

Machine learning based prediction of Young's modulus of stainless steel coated with high entropy alloys

The High Entropy Alloy (HEA) coatings exhibit diverse properties contingent upon their composition and microstructure, addressing current industrial requirements. Machine Learning (ML) regression emerges as a proficient solution for predicting the properties of HEA coatings, offering a significant reduction in experimental work. The ML regressions including Support Vector Regression (SVR), Gaussian Process Regression (GPR), Ridge Regression (RR), and Polynomial Regression (PR), are effectively employed to predict Young's modulus of HEA coated Stainless Steel (SS) through a significant database. The statistical responses of the developed regression models are analyzed through evaluation indices of Coefficient of determination (R2), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). Among the regression models, the 2-degree PR model stands alone with a high prediction accuracy of R2-0.95, MAE-16.12, and RMSE-21.53. The 2-degree PR model demonstrates a significant correlation between the predicted and experimental Young's modulus, contributing to the accurate prediction of unknown Young's modulus of the HEA-coated SS. The prediction of Young's modulus by the PR model is more reliable, as proved by an error percentile of ±4.76 %, compared to the experimental values of Young's modulus.

Adapting UWB AoA estimation towards unseen environments using transfer learning and data augmentation

Ultra-wideband technology has become increasingly prevalent in localization systems, particularly with the emergence of multi-antenna systems capable of estimating both distance and angle of arrival (AoA) for incoming signals. However, most scientific research analyzes the accuracy of AoA in one specific environment for which the estimator is trained. In this paper, we analyze the performance of various AoA estimation algorithms, such as deep convolutional neural networks (DCNN), multiple signal classification (MUSIC), and phase difference of arrival (PDoA), in unseen environments. Prior work already demonstrated the superior performance of ML for AoA estimation compared to PDoA or MUSIC. We show that MUSIC, PDoA and ML solutions suffer from degradation in unseen environments at the 90th percentile of error, with ML-based AoA estimation degrading by about 14 degrees in unseen environments compared to 4 degrees for PDoA. We demonstrate that while PDoA more effectively corrects AoA at the median level in unseen environments, ML-based methods excel at correcting higher-percentile AoA errors, including outliers. Finally, we propose a novel framework to further improve the correction of AoA outliers for DCNN-based AoA estimators using data augmentation and transfer learning, resulting in a median angular error of only 5 degrees in unseen environments, even considering a field of view up to 90 degrees.

Reliable unit strength correlations to predict the compressive strength of grouted concrete masonry

Compressive strength of grouted concrete masonry is an important parameter to design reinforced/grouted concrete masonry walls. The design standards stipulate two methods to determine the compressive strength of masonry (1) using tabulated unit strength and mortar type, and (2) testing representative masonry prisms. The compressive strength prediction of grouted concrete masonry is influenced by compressive strength values of hollow blocks, mortar and grout, and their geometries. Therefore, a multi-level approach was employed in this study to improve the existing unit strength correlations of the standards for more reliable prediction of compressive strengths of grouted concrete masonry. The existing methods to determine the compressive strength of grouted masonry were critically appraised and a database of compression tests of grouted concrete masonry prisms/wallettes was developed. This database was then used to evaluate the correlations between the compressive strengths of block, mortar, grout and masonry. The applicability of existing unit strength correlations from the design standards and literature were assessed and their relevancy and limitations are highlighted. Subsequently, updated sets of unit strength correlations are proposed in this study, through statistical reliability analyses of the predictions against the experimental results included in the database. The proposed unit strength correlations were classified according to the mortar type/strengths (≤ 10 MPa and > 10 MPa). It has been shown that the new correlations are more structurally reliable than the existing unit strength correlations through comparing the 95th percentile error values.

Investigating the Reliability of an Existing Top Angle and Seat Pad Semi-Rigid Connection System Using Advanced Modelling Techniques

The investigation introduces an advanced model procedure for evaluating the structural reliability of semi-rigid frame connections in existing aged buildings. Specifically, the focus is on the top angle and seat pad (TA–SP) semi-rigid connection, which was not initially considered in the current design standards. The approach employs a plastic hinge model to predict the ultimate strength of the connection and its beam-to-column behaviour. In order to increase computational efficiency, the investigation leverages the nonlinear behaviour of the finite element (FE) model to validate critical parameters. The statistical properties of the existing connection were obtained based on past experimental data, highlighting the weakest elements in the system. The first-order and second-order reliability methods and Monte Carlo simulations were employed to estimate the reliability index. Percentile errors were assessed to understand their impact on higher-order interactions. This new technique of identification quantifies the probability of the system failure interactions. Notably, a 45% lesser error aligns with the target reliability index, while a 114.5% larger error indicates a significant deviation from actual failure probability values. Each methodology introduced adheres to the current standard, and the system reliability analysis provides a vigorous conclusions scheme framework for assessing the existing TA–SP semi-rigid connection.

Applicability of data-driven methods in modeling electricity demand-climate nexus: A tale of Singapore and Hong Kong

Predicting electricity demand at city level is vital for stakeholders. There are various data-driven methods in electricity consumption prediction, but their applicability to forecast monthly electricity demand has yet to be systematically explored. This study examines several widely used data-driven methods, namely multiple linear regression (MLR), machine learning (ML) method including support vector machine (SVM) and random forest (RF), and deep learning (DL) method including long short-term memory network (LSTM) and LSTM-gated recurrent unit (GRU), for long-term energy prediction using climatic and historical electricity datasets in Singapore (2005–2019) and Hong Kong (1975–2019). In Singapore, ML outperforms other methods in terms of statistical criteria and time series plots, and SVM provides the best accuracy with mean absolute percentile error (MAPE) of 2.55 %, while MLR exhibits the worst accuracy with MAPE of 3.1 %. In Hong Kong, DL surpasses ML in terms of statistical criteria, time series stability, and generalization ability. MLR achieves the best overall prediction accuracy with R2 up to 0.95 but shows poor ability for predicting peak and low electricity consumption, while LSTM exhibits the smallest bias in these months. RF shows strong overfitting issues in both cities. Overall, SVM and LSTM are recommended for small and large datasets, respectively.

Vital Sign Monitoring in Dynamic Environment via mmWave Radar and Camera Fusion

Contact-free vital sign monitoring, which uses wireless signals for recognizing human vital signs (i.e, breath and heartbeat), is an attractive solution to health and security. However, the subject's body movement and the change in actual environments can result in inaccurate frequency estimation of heartbeat and respiratory. In this paper, we propose a robust mmWave radar and camera fusion system for monitoring vital signs, which can perform consistently well in dynamic scenarios, e.g., when some people move around the subject to be tracked, or a subject waves his/her arms and marches on the spot. Three major processing modules are developed in the system, to enable robust sensing. Firstly, we utilize a camera to assist a mmWave radar to accurately localize the subjects of interest. Secondly, we exploit the calculated subject position to form transmitting and receiving beamformers, which can improve the reflected power from the targets and weaken the impact of dynamic interference. Thirdly, we propose a weighted multi-channel Variational Mode Decomposition (WMC-VMD) algorithm to separate the weak vital sign signals from the dynamic ones due to subject's body movement. Experimental results show that, the 90 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${^{th}}$</tex-math></inline-formula> percentile errors in respiration rate (RR) and heartbeat rate (HR) are less than 0.5 RPM (respirations per minute) and 6 BPM (beats per minute), respectively.

Remote sensing and monitoring of water resources: A comparative study of different indices and thresholding methods

Water resources are essential for the ecological system and the development of civilization. Water is imperative factor for health preservation and sustaining various human activities, including industrial production, agriculture, and daily life. Remote sensing provides a cost-effective and practical means to detect and monitor water bodies, offers valuable insights into the impact of climatic events on water structures, especially in coastal lake regions. The research primarily utilizes Landsat-9 OLI-2 satellite images to evaluate the effectiveness of various water indices (WRI, NWI, MNDWI, NDWI) in combination with global automatic thresholding methods (K-Means, Zhenzhou's, Adaptive, Intermodes, Prewitt and Mendelsohn's Minimum, Maximum Entropy, Median, Concavity, Percentile, Intermeans, Kittler and Illingworth's Minimum Error, Tsai's Moments, Otsu's, Huang's fuzzy, Triangle, Mean, IsoData, Li’s). The study was carried out on Lake Nazik, Lake Iznik, and Lake Beyşehir, which have unique geographical characteristics, and examined the adaptability and robustness of the selected indices and thresholding methods. MNDWI consistently stands out as a robust index for water extraction, delivering accurate results across different thresholding methods in regions all three lakes. As a result of quite extensive analysis, it is obtained that MNDWI and NDWI are reliable choices for water feature extraction in various lake environments, but the specific index should consider the thresholding method and unique lake characteristics. The Minimum thresholding method stands out as the most effective thresholding technique, demonstrating impressive results across different lakes. Specifically, it achieved an average Peak Signal-to-Noise Ratio (PSNR) of 78.97 and Structural Similarity Index (SSIM) of 99.37 for Lake Nazik, 74.08 PSNR and 98.34 SSIM for Lake Iznik, and 63.96 PSNR and 93.61 SSIM for Lake Beyşehir.

Resilient Multi-Sensor UAV Navigation with a Hybrid Federated Fusion Architecture

Future UAV (unmanned aerial vehicle) operations in urban environments demand a PNT (position, navigation, and timing) solution that is both robust and resilient. While a GNSS (global navigation satellite system) can provide an accurate position under open-sky assumptions, the complexity of urban operations leads to NLOS (non-line-of-sight) and multipath effects, which in turn impact the accuracy of the PNT data. A key research question within the research community pertains to determining the appropriate hybrid fusion architecture that can ensure the resilience and continuity of UAV operations in urban environments, minimizing significant degradations of PNT data. In this context, we present a novel federated fusion architecture that integrates data from the GNSS, the IMU (inertial measurement unit), a monocular camera, and a barometer to cope with the GNSS multipath and positioning performance degradation. Within the federated fusion architecture, local filters are implemented using EKFs (extended Kalman filters), while a master filter is used in the form of a GRU (gated recurrent unit) block. Data collection is performed by setting up a virtual environment in AirSim for the visual odometry aid and barometer data, while Spirent GSS7000 hardware is used to collect the GNSS and IMU data. The hybrid fusion architecture is compared to a classic federated architecture (formed only by EKFs) and tested under different light and weather conditions to assess its resilience, including multipath and GNSS outages. The proposed solution demonstrates improved resilience and robustness in a range of degraded conditions while maintaining a good level of positioning performance with a 95th percentile error of 0.54 m for the square scenario and 1.72 m for the survey scenario.

Short-term wind power forecasting based on dual attention mechanism and gated recurrent unit neural network

Accurate wind power forecasting is essential for both optimal grid scheduling and the massive absorption of wind power into the grid. However, the continuous changes in the contribution of various meteorological features to the forecasting of wind power output under different time or weather conditions, and the overlapping of wind power sequence cycles, make forecasting challenging. To address these problems, a short-term wind power forecasting model is established that integrates a gated recurrent unit (GRU) network with a dual attention mechanism (DAM). To compute the contributions of different features in real time, historical wind power data and meteorological information are first extracted using a feature attention mechanism (FAM). The feature sequences collected by the FAM are then used by the GRU network for preliminary forecasting. Subsequently, one-dimensional convolution employing several distinct convolution kernels is used to filter the GRU outputs. In addition, a multi-head time attention mechanism (MHTAM) is proposed and a Gaussian bias is introduced to assign different weights to different time steps of each modality. The final forecast results are produced by combining the outputs of the MHTAM. The results of the simulation experiment show that for 5-h, 10-h, and 20-h short-term wind power forecasting, the established DAM-GRU model performs better than comparative models on the basis of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), R-squared (R2), Square sum error (SSE), Mean absolute percentile error (MAPE), and Relative root mean square error (RRMSE) index.

Visible Light Positioning-Based Robot Localization and Navigation

Visible light positioning or VLP has been identified as a promising technique for accurate indoor localization utilizing pre-existing lighting infrastructure. Robot navigation is one of the many potential applications of VLP. Recent literature shows a small number of works on robots being controlled by fusing location information acquired via VLP that uses a rolling shutter effect camera as a receiver with other sensor data. This paper, in contrast, reports on the experimental performance of a cartesian robot that was controlled solely by a VLP system using a cheap photodiode-based receiver rigidly attached to the robot’s end-effector. The receiver’s position was computed using an inverse-Lambertian function for ranging followed by multi-lateration. We developed two novel methods to leverage the VLP as an online navigation system to control the robot. The position acquired from the VLP was used by the algorithms to determine the direction the robot needed to move. The developed algorithms guided the end-effector to move from a starting point to target/destination point(s) in a discrete manner, determined by a pre-determined step size. Our experiments consisted of the robot autonomously repeating straight line-, square- and butterfly-shaped paths multiple times. The results show median errors of 27.16 mm and 26.05 mm and 90 percentile errors of 37.04 mm and 47.48 mm, respectively, for the two methods.

Unmanned aerial vehicle localization using angle of departure from a single base station and dead‐reckoning

AbstractLocalization in GNSS‐denied environments for unmanned aerial vehicles (UAVs) has recently gained significant interest from the research community. Most of the research is focused primarily on visual localization. This paper examines an algorithm which employs angle of departure and UAVs payload sensor data for UAV localization. First the algorithm uses multiple angle of departures from a single base station and a travel calculated by applying dead‐reckoning on the UAVs inertial measurement unit (IMU), to compute UAV location in 2D coordinates. The 2D location estimate is then fed into a modified extended Kalman filter, which employs the estimate, IMU and barometer data to compute the 3D coordinates for UAV. For the simulation, simulation‐in‐the‐loop accompanied by Arducopter and MAVLink is used to simulate different trajectories and collect the required data for the algorithm. The algorithm is validated by comparing the extended Kalman filter estimates with IMU dead‐reckoned positions. Three simulations were performed, consisting of linear, zigzag, and curved trajectories. A percentile error of 2.5 and 4 m is achieved for the x‐coordinate and y‐coordinate, respectively, on the zigzag and curved trajectories. Interestingly, the linear trajectory showed a larger localization error in its y‐coordinate.

New Simple Analytical Surge/Swab Pressure Model for Power-Law and Modified Yield-Power-Law Fluid in Concentric/Eccentric Geometry

The axial movement of pipe in and out of the well generates positive (surge) and negative (swab) pressures that will impact the well pressure. When the swab and surge effects cause well pressures outside the allowable operational limits, wellbore instability (well collapse/well fracture), kick, and induced drill string sticking issues will occur. The problems increase the operational and nonproductive time-related costs. Consequently, the drilling budget rises significantly. It is therefore, imperative to predict the differential pressures in order to mitigate the problems. Even though several models have been developed in the past, models work for the considered experimental setup and conditions. In this paper, a simple analytical model was derived for eccentric/concentric annuli. The fluid rheological behaviors were assumed to be described by power law and yielded power-law. The model is derived based on a steady state condition, and the effects of tripping speed, the power-law fluids, the yield-power-law fluids rheological parameters, and well geometries are considered. The model is compared with experimental data from the literature and with the existing model. Parametric sensitivity studies have been conducted. Results show that the model prediction exhibited quite good performance, with an average percentile error deviation of 9.9% and 6.2% for the power-law and yield-power-law fluids, respectively. However, more testing is required to determine the model’s limitations and application.

Evaluation of PTV margins and plan robustness for single isocentre multiple target stereotactic radiosurgery

PurposeRobustness to residual setup errors and linac delivery errors of BrainLab Elements single-isocentre-multiple-target stereotactic radiosurgery was evaluated. MethodsResidual setup errors of 13 patients were evaluated. Linac delivery error was quantified through multi-metastases-Winston-Lutz measurements. PTV margins were calculated using the van Herk recipe. Patient scans were translated and rotated by the median and 95th percentile of the combined uncertainties, and plans were recalculated subsequently. Previous patients’ plans were then replanned with the derived margins, effects on GTV coverage and normal brain doses were assessed. ResultsMean (±stdev) coverage of all targets in the original plans were 99.4% (±0.9%) and 98.9% (±1.0%) for 1 and 3-fraction patients respectively. Median geometrical errors did not result in significant differences. A statistically significant reduction in coverage to 91.4% (±10.4%) and 93.0% (±9.6%) was seen under 95th percentile errors. Applying the derived optimal margin of 0.5 mm resulted in 78% of the GTVs retaining a coverage of 98% or above even in the presence of 95th percentile errors, compared to only 30% if no margins were applied. Replanning with margins also caused no significant increase to local normal brain doses, however global dose increases varied according to the number of metastases. ConclusionsPlans were shown to be robust to average geometrical uncertainties despite targets having no margins, however occurrence of GTV under-coverage increased under 95th percentile scenarios. The margin was proven to substantially improve the target dose coverage with limited change to local normal brain doses, although not all sources of geometrical uncertainty were considered.

Experimental investigation of replication accuracy of polymer-based microfluidic chip fabricated through induction-assisted hot embossing and parametric optimization through nature-inspired algorithms

Hot embossing (HE) creates micron-scaled patterns on a polymer substrate. Conventional-HE faces a long cycle-time issue and reduces productivity. An induction-assisted HE (IHE) setup was developed in-house to solve this issue. A microfluidic chip was made using an in-house developed setup on polymethyl methacrylate. Micro-features on aluminum-6061 mold were made using fiber laser machining instead of photolithography, followed-by reactive ion etching or electroplating to save both costs and time. Accurate micropattern replication is vital for HE. Embossing factors affect replication accuracy. Less deviation in micro-channel depth improves replication accuracy. The experimental data was gathered via L27 orthogonal array. Linear regression was used to connect process parameters and response variables. The embossing temperature affects micro-channel depth by 79.78%. Particle swarm optimization, artificial bee colony (ABC), firefly algorithm (FA), and grey wolf optimization (GWO) optimized IHE process parameters. Firefly algorithm outperformed. The FA converged in four iterations, whereas particle swarm optimization took 12 and ABC 38. All methods (excluding GWO) reached 6.5809 μm, the global best. A confirmation test shows optimal operating conditions reduce embossed micro-channel depth variation from 126.758 μm to 7.0327 μm and improve replication accuracy from 19.94% to 95.50%. The predicted and experimental embossed micro-channel depth deviation percentile error is 6.86%. An IHE setup was used to bond the microfluidic chip. The microchannel in the microfluidic chip was treated with vaporized bonding before thermal bonding to prevent deformation/blocking. No blue ink leaked during the testing of the manufactured microfluidic chip, showing acceptable quality.

A hybrid method of correcting CBCT for proton range estimation with deep learning and deformable image registration

Objective. This study aimed to develop a novel method for generating synthetic CT (sCT) from cone-beam CT (CBCT) of the abdomen/pelvis with bowel gas pockets to facilitate estimation of proton ranges. Approach. CBCT, the same-day repeat CT, and the planning CT (pCT) of 81 pediatric patients were used for training (n = 60), validation (n = 6), and testing (n = 15) of the method. The proposed method hybridizes unsupervised deep learning (CycleGAN) and deformable image registration (DIR) of the pCT to CBCT. The CycleGAN and DIR are respectively applied to generate the geometry-weighted (high spatial-frequency) and intensity-weighted (low spatial-frequency) components of the sCT, thereby each process deals with only the component weighted toward its strength. The resultant sCT is further improved in bowel gas regions and other tissues by iteratively feeding back the sCT to adjust incorrect DIR and by increasing the contribution of the deformed pCT in regions of accurate DIR. Main results. The hybrid sCT was more accurate than deformed pCT and CycleGAN-only sCT as indicated by the smaller mean absolute error in CT numbers (28.7 ± 7.1 HU versus 38.8 ± 19.9 HU/53.2 ± 5.5 HU; P ≤ 0.012) and higher Dice similarity of the internal gas regions (0.722 ± 0.088 versus 0.180 ± 0.098/0.659 ± 0.129; P ≤ 0.002). Accordingly, the hybrid method resulted in more accurate proton range for the beams intersecting gas pockets (11 fields in 6 patients) than the individual methods (the 90th percentile error in 80% distal fall-off, 1.8 ± 0.6 mm versus 6.5 ± 7.8 mm/3.7 ± 1.5 mm; P ≤ 0.013). The gamma passing rates also showed a significant dosimetric advantage by the hybrid method (99.7 ± 0.8% versus 98.4 ± 3.1%/98.3 ± 1.8%; P ≤ 0.007). Significance. The hybrid method significantly improved the accuracy of sCT and showed promises in CBCT-based proton range verification and adaptive replanning of abdominal/pelvic proton therapy even when gas pockets are present in the beam path.

Short-term load forecasting using neural networks and global climate models: An application to a large-scale electrical power system

This paper focuses on the development of shallow and deep neural networks in the form of multi-layer perceptron, long-short term memory, and gated recurrent unit to model the short-term load forecasting problem. Different model architectures are tested, and global climate model information is used as input to generate more accurate forecasts. A real study case is presented for the Brazilian interconnected power system and the results generated are compared with the forecasts from the Brazilian Independent System Operator model. In general terms, results show that the bidirectional versions of long-short term memory and gated recurrent unit produce better and more reliable predictions than the other models. From the obtained results, the recurrent neural networks reach Nash-Sutcliffe values up to 0.98, and mean absolute percentile error values of 1.18%, superior than the results obtained by the Independent System Operator models (0.94 and 2.01% respectively). The better performance of the neural network models is confirmed under the Diebold-Mariano pairwise comparison test.