Model For Correction Research Articles

Prediction of Traffic Volume Based on Deep Learning Model for AADT Correction

Prediction of Traffic Volume Based on Deep Learning Model for AADT Correction

An experimental study on three-port measurements for acoustic characterisation of the perforate reactance

The usage of perforated plates in passive noise-control systems in e.g. aircraft liners and mufflers involves the standard operating conditions of grazing flow and high-amplitude acoustic incidence. This study focuses on the usage of the three-port experiment technique for the experimental passive acoustic characterisation of a perforated plate. Expanding on the previous paper about the resistance of the perforate, the influence of operating conditions on the imaginary part of the transfer impedance, i.e., the reactance is discussed here. The relationship between the end correction in the linear range, without the presence of grazing flow, and the frequency of acoustic excitation is demonstrated. The reduction in the value of the mass end correction under the individual effects of grazing flow, and high-amplitude excitation is presented. Additionally, depending on the excitation wavelength, the partial recovery of the reduced value of the end correction under simultaneous exposure is also shown. Using the experimental results, models for the end correction, and by extension the reactance are proposed. These models take into account the geometrical parameters of the perforate, and the dimensionless Mach, Shear and Strouhal numbers.

Big Brother Is Watching You: Leveraging Artificial Intelligence for Automated Fuel Cell Monitoring (Technical Report)

In this technical report, we present our second-generation digital twin model for monitoring of fuel cells. This digital twin combines two data-driven machine learning models: a stationary model for stationary cell voltage prediction and a degradation model for degradation correction. This combination of both models results in a precise probabilistic prediction of the cell voltage over time.Furthermore, we present a web-based framework for automated fuel cell monitoring in real time. This framework allows training a digital twin on existing data and subsequently applying the twin to automatically check new data for anomalies.

Big Brother Is Watching You: Leveraging Artificial Intelligence for Automated Fuel Cell Monitoring (Technical Report)

In this technical report, we present our second-generation digital twin model for monitoring of fuel cells. This digital twin combines two data-driven machine learning models: a stationary model for stationary cell voltage prediction and a degradation model for degradation correction. This combination of both models results in a precise probabilistic prediction of the cell voltage over time.Furthermore, we present a web-based framework for automated fuel cell monitoring in real time. This framework allows training a digital twin on existing data and subsequently applying the twin to automatically check new data for anomalies.

Adaptive control strategy for microgrid inverters based on Narendra model

In recent years, microgrid technology has been widely studied and applied. However, with times developing, the installed capacity of distributed power generation devices has been improved, and work is being carried out in increasingly complex situations, resulting in a decline in the control performance of microgrids. In view of this, to effectively improve inverter’s control performance, research is conducted on the fusion of Narendra model and adaptive control strategies for real-time voltage correction and compensation in complex situations. Compared to traditional inverters, inverters under research methods have faster voltage recovery speed when encountering load switching, and can recover in about one cycle, with good control performance. In the comparison between the improved inverter adaptive control system and the inverter adaptive system, the improved inverter voltage recovery speed is faster, can be restored within one cycle, and the control effect of the inverter is better. The harmonic rate of the port voltage has decreased from 10.43 to 1.92%. The applicability of the research method was verified. It indicats that the research method can improve inverter’s control effect and solve problems such as voltage deviation, three-phase asymmetry, harmonic pollution, etc. that are easily generated by the output terminal voltage. Simultaneously, research has provided theoretical basis and data support for the research of microgrids.

Development and Validation of a Realistic Neonatal Intestinal Jejunoileal Atresia Simulator for the Training of Pediatric Surgeons.

Background: Neonatal surgical pathology presents highly technical complexity and few opportunities for training. Many of the neonatal surgical entities are not replicable in animal models. Realistic 3D models are a cost-effective and efficient alternative for training new generations of pediatric surgeons. Methods: We conceptualized, designed, and produced an anatomically realistic model for the open correction of jejunoileal atresia. We validated it with two groups of participants (experts and non-experts) through face, construct, and content validity questionnaires. Results: The model was validated by eleven experts and nine non-experts. The mean procedure time for the experts and non-experts groups was 41 and 42 min, respectively. Six non-experts and one expert did not complete the procedure by the designed time (45 min) (p = 0.02). The mean score of face validity was 3.1 out of 4. Regarding construct validity, we found statistically significant differences between groups for the correct calculation of the section length of the antimesenteric border (Nixon's technique) (p < 0.01). Concerning content validity, the mean score was 3.3 out of 4 in the experts group and 3.4 out of 4 in the non-experts group. Conclusions: The present model is a realistic and low-cost valid option for training for open correction of jejunoileal atresia. Before drawing definitive conclusions, future studies with larger sample sizes and blinded validators are needed.

Development and validation of a realistic type III esophageal atresia simulator for the training of pediatric surgeons

BackgroundThe technical complexity and limited casuistry of neonatal surgical pathology limit the possibilities of developing the necessary technical competencies by specialists in training. Esophageal atresia constitutes the paradigm of this problem. The use of synthetic 3D models for training is a promising line of research, although the literature is limited.MethodsWe conceptualized, designed, and produced an anatomically realistic model for the open correction of type III oesophageal atresia. We validated it with two groups of participants (experts and non-experts) through face, construct, and content-validity questionnaires.ResultsThe model was validated by nine experts and nine non-experts. The mean procedure time for the experts and non-experts groups was 34.0 and 38.4 min, respectively. Two non-experts did not complete the procedure at the designed time (45 min). Regarding the face validity questionnaire, the mean rating of the model was 3.2 out of 4. Regarding the construct validity, we found statistically significant differences between groups for the equidistance between sutures, 100% correct in the expert group vs. 42.9% correct in the non-expert group (p = 0.02), and for the item “Confirms that tracheoesophageal fistula closure is watertight before continuing the procedure”, correctly assessed by 66.7% of the experts vs. by 11.1% of non-experts (p = 0.05). Concerning content validity, the mean score was 3.3 out of 4 for the experts and 3.4 out of 4 for the non-experts.ConclusionsThe present model is a cost-effective, simple-to-produce, and validated option for training open correction of type III esophageal atresia. However, future studies with larger sample sizes and blinded validators are needed before drawing definitive conclusions.

BGAT-CCRF: A novel end-to-end model for knowledge graph noise correction

Knowledge graph (KG) noise correction aims to select suitable candidates to correct the noises in KGs. Most of the existing studies have limited performance in repairing the noisy triple that contains more than one incorrect entity or relation, which significantly constrains their implementation in real-world KGs. To overcome this challenge, we propose a novel end-to-end model (BGAT-CCRF) that achieves better noise correction results. Specifically, we construct a balanced-based graph attention model (BGAT) to learn the features of nodes in triples’ neighborhoods and capture the correlation between nodes based on their position and frequency. Additionally, we design a constrained conditional random field model (CCRF) to select suitable candidates guided by three constraints for correcting one or more noises in the triple. In this way, BGAT-CCRF can select multiple candidates from a smaller domain to repair multiple noises in triples simultaneously, rather than selecting candidates from the whole KG to repair noisy triples as traditional methods do, which can only repair one noise in the triple at a time. The effectiveness of BGAT-CCRF is validated by the KG noise correction experiment. Compared with the state-of-the-art models, BGAT-CCRF improves the fMRR metric by 3.58% on the FB15K dataset. Hence, it has the potential to facilitate the implementation of KGs in the real world.

Analysis of models of measurement and correction of volumetric error of three-axis metal-cutting machine tool

The paper considers the improvement of the quality of products of machined production in connection with the quality of machining of various products and, accordingly, with the accuracy of the equipment used. It is noted that geometrical errors, which account for up to 40 % of the total error of a multi-axis, especially three-axis, metal-cutting machine tool, have the greatest influence on the accuracy of product machining. In order to improve the quality of product machining, it is necessary to correct its parameters according to the measurement information received during machine tool operation, for which the methods of mathematical modelling are used. Models of the volumetric error of a three-coordinate metal-cutting machine tool are developed, taking into account the summands of the first (linear) and second (quadratic) order of smallness. A comparative analysis of the developed models is carried out. The results of experiments on measurement of volumetric errors of the STAN S500 complex and their correction on the basis of the nonlinear theoretical model are given. It is found that in the range of angular errors 0–2.5′ the consideration of quadratic terms of the model in addition to linear ones does not lead to a significant reduction of the volumetric error. It is shown that when processing measurement information of multi-axis metal-cutting machines it is sufficient to limit the consideration of components of the volumetric error not higher than the first order of smallness. The research results are useful for the acceptance and periodic control of the volumetric error of metal-cutting machines, as well as for the programme reduction of the volumetric error.

Subsidence detection in southwest Guangdong–Hong Kong–Macao Greater Bay Area using InSAR with GNSS corrected tropospheric delays

As one of the biggest bay areas in the world, the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) is subject to subsidence due to thick sedimentary. In this study, we combine 154 Sentinel-1 A/B images acquired between 2017 and 2022 with Global Navigation Satellite System (GNSS) observations to characterize the subsidence in southwestern GBA. Specifically, we use GNSS to establish the Iterative Tropospheric Decomposition (ITD) model for zenith tropospheric delay correction and the model is improved using random forest algorithm to alleviate errors caused by unreasonable interpolation methods, showing better performance than Generic Tropospheric Correction Online Service (GACOS) product. At GNSS stations with complete data, the improved model utilizing GNSS data achieves a Pearson correlation coefficient of 0.659, in contrast to a coefficient of only 0.434 using GACOS. The maximum subsidence rate detected is up to 11.32 cm/year in Zhuhai, accompanied by noticeable seasonal variations correlated with precipitation. In addition to the sediment consolidation and compaction, human activities such as agriculture, construction, and land reclamation further exacerbate subsidence in the GBA. This study enhances the utility of GNSS in improving tropospheric delay correction performance and providing more accurate observations of subsidence.

A three-directional stress-strain model-based physics-embedded prediction framework for metal tube full-bent cross-sectional characteristics

A metal tube system is known as the industrial blood vessel, among which the bent section is the most vulnerable part. The cross-sectional defects (CSDs) of the bent tube cause the flow fluctuation of the fluid inside the tube. The existing defect characterization methods are roughly presented by describing CSDs in some specific cross-sections, which results in the lack of the tube full-bent section (FBS) characteristic information. To comprehensively describe and predict the tube FBS characteristics, an advanced physics-embedded CSDs prediction framework is proposed. This framework includes an FBS-neutral layer displacement angle (NLDA) prediction module and an FBS-CSDs prediction module, which uses the method that integrates the analytical model and BiLSTM-based deep learning (DL) models to predict the CSDs in the FBS of the tube. A novel analytical model of CSDs that considers both three-directional stresses and strains during tube bending is embedded in the FBS-CSDs prediction module. The analytical model provides the initial predicted values of CSDs through the NLDA sequence obtained from the FBS-NLDA module. The inaccurate CSDs are then treated as physical information to be fed into DL models for further correction and prediction. The prediction performance of this framework is validated through numerical simulations and experiments. The results prove that the framework can accurately predict the CSDs in the tube FBS. The integration of DL models with the analytical model not only overcomes the limitations of the analytical model, but also improves the prediction accuracy and convergence speed of DL models.

AutoCorNN: An Unsupervised Physics-Aware Deep Learning Model for Geometric Distortion Correction of Brain MRI Images Towards MR-Only Stereotactic Radiosurgery.

Geometric distortions in brain MRI images arising from susceptibility artifacts at air-tissue interfaces pose a significant challenge for high-precision radiation therapy modalities like stereotactic radiosurgery, necessitating sub-millimeter accuracy. To achieve this goal, we developed AutoCorNN, an unsupervised physics-aware deep-learning model for correcting geometric distortions. Two publicly available datasets, the MPI-Leipzig Mind-Brain-Body with 318 subjects, and the Vestibular Schwannoma-SEG dataset, encompassing 242 patients were utilized. AutoCorNN integrates two 2D convolutional encoder-decoder neural networks with the forward physical model of MRI signal generation to predict undistorted MR and field map images from distorted MR input. The network is trained in an unsupervised manner by minimizing the mean absolute error between the measured and estimated k-space data, without requiring ground truth images during training or deployment. The model was evaluated on vestibular schwannoma cases. AutoCorNN achieved a peak signal-to-noise ratio (PSNR) of 41.35 ± 0.02dB, a root mean square error (RMSE) of 0.02 ± 0.003, and a structural similarity index (SSIM) of 0.99 ± 0.02 outperforming uncorrected and B0-mapping correction methods. Geometric distortions of about 1.6mm were observed at the air-tissue interfaces at the air canal and nasal cavity borders. Geometrically, distortion correction increased the target volume from 3.12 ± 0.52cc to 3.84 ± 0.54cc. Dosimetrically, AutoCorNN improved target coverage (0.96 ± 0.01 to 0.97 ± 0.02), conformity index (0.92 ± 0.03 to 0.94 ± 0.03), and reduced dose gradients outside the target. AutoCorNN achieves accurate geometric distortion correction comparable to conventional iterative methods while offering substantial computational acceleration, enabling precise target delineation and conformal dose delivery for improved radiation therapy outcomes.

RuGECToR: Rule-Based Neural Network Model for Russian Language Grammatical Error Correction

RuGECToR: Rule-Based Neural Network Model for Russian Language Grammatical Error Correction

A new model for pressure drop correction for series-arranged misaligned monoliths

This study presents a new model to predict the additional pressure drop for series-arranged misaligned monoliths, relevant for high-flow industrial applications. Long reactor tubes often contain multiple short monoliths, leading to extra pressure drops as the flow enters and leaves multiple monoliths. Existing models address aligned monoliths, but this research considers misalignment, where extra pressure drop is higher. Channel Reynolds from 50 to 320, spacings from 0.1 to 20 channel diameters, and several void fractions are investigated using a discrete channel model. Results indicate a significant impact of misalignment on extra pressure drops in closely spaced monoliths due to flow collisions. A correction factor model that adapts to misalignment is proposed, starting from large values for closely spaced channels and decreasing as spacing increases. The proposed model aligns with discrete channel data and phenomenological behavior of aligned channels both in trend and magnitude, showing an R2 above 0.93.

Deep-learning-based Attenuation Correction for 68Ga-DOTATATE Whole-body PET Imaging: A Dual-center Clinical Study.

Attenuation correction is a critical phenomenon in quantitative positron emission tomography (PET) imaging with its own special challenges. However, computed tomography (CT) modality which is used for attenuation correction and anatomical localization increases patient radiation dose. This study was aimed to develop a deep learning model for attenuation correction of whole-body 68Ga-DOTATATE PET images. Non-attenuation-corrected and computed tomography-based attenuation-corrected (CTAC) whole-body 68Ga-DOTATATE PET images of 118 patients from two different imaging centers were used. We implemented a residual deep learning model using the NiftyNet framework. The model was trained four times and evaluated six times using the test data from the centers. The quality of the synthesized PET images was compared with the PET-CTAC images using different evaluation metrics, including the peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), mean square error (MSE), and root mean square error (RMSE). Quantitative analysis of four network training sessions and six evaluations revealed the highest and lowest PSNR values as (52.86±6.6) and (47.96±5.09), respectively. Similarly, the highest and lowest SSIM values were obtained (0.99±0.003) and (0.97±0.01), respectively. Additionally, the highest and lowest RMSE and MSE values fell within the ranges of (0.0117±0.003), (0.0015±0.000103), and (0.01072±0.002), (0.000121±5.07xe-5), respectively. The study found that using datasets from the same center resulted in the highest PSNR, while using datasets from different centers led to lower PSNR and SSIM values. In addition, scenarios involving datasets from both centers achieved the best SSIM and the lowest MSE and RMSE. Acceptable accuracy of attenuation correction on 68Ga-DOTATATE PET images using a deep learning model could potentially eliminate the need for additional X-ray imaging modalities, thereby imposing a high radiation dose on the patient.

Hybrid Approaches: Combining Computational Mathematics and Artificial Intelligence for Enhanced Performance

The combination of computer mathematics and artificial intelligence (AI) looks like a good way to solve hard problems in many areas. In this outline, the possibility for these two fields to work together to improve problem-solving skills is looked at. Computational mathematics is a thorough way to understand and solve difficult mathematical problems. It is based on numerical analysis, optimization, and mathematical models. Artificial intelligence methods, on the other hand, like machine learning, deep learning, and neural networks, are very useful for making decisions and learning from data. Hybrid techniques can use the best parts of both computational mathematics and artificial intelligence to solve problems that standard approaches might not be able to solve. These methods make it possible to create complex algorithms that can quickly solve optimization problems, model complicated systems, and draw useful conclusions from big datasets. For example, hybrid methods can improve the accuracy and speed of computing in engineering and science exercises by mixing numerical methods with machine learning models for mistake correction and flexible improvement. Hybrid methods make it easier for people from different fields to work together, connecting academic theory to real-world uses. Researchers and professionals from different fields can come together to come up with new solutions that use the skills of both computational mathematics and artificial intelligence. This cross-disciplinary synergy can be seen in areas like banking, healthcare, and defense, where using both mathematical rigor and AI-driven predictive analytics can help with making difficult decisions. When you put together computer mathematics and AI, you can do new kinds of core study. AI can be used to study huge sets of solutions and find trends in data. This helps researchers learn more about mathematical events and make programs that work better. This looping process of learning and optimizing allows for constant improvement and revision, which finds new answers to math problems that have been around for a long time. Combining computational mathematics and artificial intelligence into hybrid methods has a huge amount of promise to speed up study and new ideas in many areas.

Near-surface velocity inversion and modeling method based on surface waves in petroleum exploration: a case study from Qaidam Basin, China

Surface waves are widely used in the study of underground structures at various scales because of their dispersion characteristics in layered media. Whether in natural seismology or engineering seismology, surface wave analysis methods have matured and developed for their respective fields. However, in oil and gas exploration, many data processors still tend to consider surface waves as noise that needs to be removed. To make more people pay attention to the application of surface waves and widely utilize surface waves carrying the near surface information in oil and gas exploration, this paper takes the data processing of LH site in Qinghai, China as an example to apply surface wave analysis methods to oil and gas exploration. We first preprocess and perform dispersion imaging method on the seismic record in the LH site to obtain frequency-phase velocity spectrum with good resolution and signal-to-noise ratio. Then, utilizing clustering algorithms, it automatically identifies and picks dispersion curves. Finally, through a simultaneous inversion algorithm of velocity and thickness, it inverts the dispersion curves and obtain S-wave velocity profiles in the depth range of 0–200 m. The near surface is divided into four zones based on velocity ranges and depth ranges. Additionally, we apply the surface waves inversion results as constraints to first-arrival tomography and obtain objectively accurate P-wave velocity profiles and Poisson’s ratio profiles. The results indicate that by applying surface wave analysis methods, the near surface velocity information carried by surface waves can be extracted, providing near surface velocity models for static correction and migration. At the same time, compared with the surface wave application in engineering seismology, the scale of oil and gas exploration is larger, so that the data processing of surface waves is particularly important, otherwise it will affect the picking of the dispersion curve and inversion.

Prediction Model for Lumbar Curve Correction After Selective Thoracic Fusion in Lenke 1 and 2 Adolescent Idiopathic Scoliosis.

A retrospective study. To identify independent risk factors and construct a prediction model for lumbar curve correction (LCC) after selective thoracic fusion (STF) in patients with Lenke 1 and 2 adolescent idiopathic scoliosis (AIS). STF has been widely applied to Lenke 1 and 2 AIS patients. However, LCC after STF is still controversial. One hundred twenty-eight patients undergoing STF with at least 2 years of follow-up were included. Cases were divided into a high-LCC group and a low-LCC group according to a rounded-up median of 65%. Forty-nine variables were taken into account. Logistic regression was applied to identify independent predictive factors. A prediction model was established by backward stepwise regression, and its evaluation was implemented on R. Five parameters showed independent predictive value for low LCC: right shoulder higher before surgery (right shoulder higher versus balanced: odds ratio [OR]=0.244, P =0.014), postoperative Cobb angle of lumbar curve (LC) (OR=1.415, P =0.001, cutoff value=11°), lowest instrumented vertebra (LIV) distal to end vertebra (no vs. yes: OR=4.587, P =0.013), postoperative LIV tilt (OR=0.686, P =0.010, cutoff value=6.85°) and postoperative LIV+1 tilt (OR=1.522, P =0.005, cutoff value=6.25°). The prediction model included 6 variables: lumbar modifier, preoperative shoulder balance, postoperative Cobb angle of LC, LIV position, postoperative LIV tilt, and postoperative LIV+1 tilt. The model evaluation demonstrated satisfactory capability and stability (area under curve=0.890, 10-fold cross-validation accuracy=0.782). Preoperative shoulder balance, Cobb angle of LC, LIV position, postoperative LIV and LIV+1 tilt could be used to prognosticate LCC after STF. A model with solid prediction ability was established, which could further our understanding of LCC and assist in making clinical decisions.

Battery lumped fractional‐order hysteresis thermoelectric coupling model for state of charge estimation adaptive to time‐varying core temperature conditions

AbstractAs electric vehicles become more common, there is increasing concern regarding their battery reliability and safety. The estimation accuracy is strongly correlated with the performance of the battery model. The lumped fractional‐order hysteresis thermoelectric coupling model (LFHTCM), considering temperature and hysteresis effects, is established in this paper. Firstly, the fractional‐order hysteresis sub‐model is established based on the Grunwald‐Letnikov (G‐L) fractional calculus principle and the recursive model of hysteresis voltage. Then, the thermal sub‐model is established by integrating the Bernardi heating mechanism and the heat transfer model. The fractional‐order sub‐model provides voltage to the thermal sub‐model for heat generation calculation. The thermal sub‐model provides temperature to the fractional‐order model for parameter correction. Finally, the fractional‐order unscented Kalman filtering (FOUKF) algorithm was developed to estimate the state of charge (SOC) of the battery. Experimental results confirmed the effectiveness of the model, with its estimation enhancing the utilization, operational optimization, and calibration of batteries in practical applications.

A hybrid model coupling process-driven and data-driven models for improved real-time flood forecasting

Accurate and reliable incoming flood forecasting is an important prerequisite for flood warning, flood risk analysis and reservoir flood control operation. This paper proposes a hybrid model for real-time flood forecasting that couples process-driven hydrological models (HMs) with data-driven models (DDMs). The generic hybrid model framework adds DDMs as the post-processing procedure for residual correction to the original results of HMs, and considers multiple uncertainties in input data, parameter and model structure simultaneously. The performance of the hybrid model is evaluated comprehensively in terms of deterministic forecast accuracy, interval forecast reliability, and the reliability and sharpness of probabilistic forecast. Taking the multireservoir system at the east Pi River as a study case, the results indicate that: (1) Compared to the benchmark model (ensemble XAJ model), the hybrid model with additional residual analysis show a significant improvement. The average continuous ranked probability score (CRPS) metric values calculated by the Stacking-Hybrid model improved by 71.5 %, 67.0 % and 38.1 % in the three data samples. Furthermore, the adaptability of the Stacking-Hybrid model for residual correction during short-duration intense rainfall events has been validated, with the relative error of the peak discharge improved to within ± 10 %. (2) The Stacking-Hybrid model, which also takes into account structure uncertainty, is able to better exploit the combined advantages and improve the stability of the model performance compared to those that only apply a single DDM. (3) When the number of iterations within the BOA reaches 300, the parameter optimization process is capable to search for the hyperparameters that bring out the best performance of the DDMs. (4) When the ensemble size reaches 200, the uncertainty of HM parameters can be fully defined, and the consumed computational resources can be controlled within an acceptable bound while ensuring stable model performance. Overall, the hybrid model that takes into account multiple sources of uncertainty generates both interval and probabilistic forecast in addition to deterministic forecast, which can provide richer risk information for subsequent flood warning and reservoir operation, making the flood prevention decisions more reliable and scientific.