Location Prediction Research Articles

A Multichannel MAC Protocol without Coordination or Prior Information for Directional Flying Ad hoc Networks

Achieving neighbor discovery for a directional flying ad hoc network (FANET) with multiple channels poses challenges for media access control (MAC) protocol design, as it requires simultaneous main lobe and channel rendezvous while dealing with the high UAV mobility. In order to achieve fast neighbor discovery for initial access without coordination or prior information, we first establish the theoretical supremum for the directional main lobe. Then, to achieve the supremum, we introduce the BR-DA and BR-DA-FANET algorithms to respectively establish the supremum on rendezvous between a pair of UAVs’ main lobes and rendezvous of main lobes for all UAVs in the FANET. To further accelerate the neighbor discovery process, we propose the neighbor discovery with location prediction protocol (ND-LP) and the avoiding communication interruption with location prediction (ACI-LP) protocol. ND-LP enables quick main lobe rendezvous and channel rendezvous, while ACI-LP enables beam tracking and channel rendezvous together with the avoidance of communication interruptions. The simulation results demonstrate that the proposed protocols outperform the state-of-the-art works in terms of neighbor discovery delay.

Months-long seismicity transients preceding the 2023 MW 7.8 Kahramanmaraş earthquake, Türkiye

Short term prediction of earthquake magnitude, time, and location is currently not possible. In some cases, however, documented observations have been retrospectively considered as precursory. Here we present seismicity transients starting approx. 8 months before the 2023 MW 7.8 Kahramanmaraş earthquake on the East Anatolian Fault Zone. Seismicity is composed of isolated spatio-temporal clusters within 65 km of future epicentre, displaying non-Poissonian inter-event time statistics, magnitude correlations and low Gutenberg-Richter b-values. Local comparable seismic transients have not been observed, at least since 2014. Close to epicentre and during the weeks prior to its rupture, only scarce seismic activity was observed. The trends of seismic preparatory attributes for this earthquake follow those previously documented in both laboratory stick-slip tests and numerical models of heterogeneous earthquake rupture affecting multiple fault segments. More comprehensive earthquake monitoring together with long-term seismic records may facilitate recognizing earthquake preparation processes from other regional deformation transients.

Leveraging Prompt and Top-K Predictions with ChatGPT Data Augmentation for Improved Relation Extraction

Relation extraction tasks aim to predict the type of relationship between two entities from a given text. However, many existing methods fail to fully utilize the semantic information and the probability distribution of the output of pre-trained language models, and existing data augmentation approaches for natural language processing (NLP) may introduce errors. To address this issue, we propose a method that introduces prompt information and Top-K prediction sets and utilizes ChatGPT for data augmentation to improve relational classification model performance. First, we add prompt information before each sample and encode the modified samples by pre-training the language model RoBERTa and using these feature vectors to obtain the Top-K prediction set. We add a multi-attention mechanism to link the Top-K prediction set with the prompt information. We then reduce the possibility of introducing noise by bootstrapping ChatGPT so that it can better perform the data augmentation task and reduce subsequent unnecessary operations. Finally, we investigate the predefined relationship categories in the SemEval 2010 Task 8 dataset and the prediction results of the model and propose an entity location prediction task designed to assist the model in accurately determining the relative locations between entities. Experimental results indicate that our model achieves high results on the SemEval 2010 Task 8 dataset.

Wind field and model resolution optimization and testing source terms for the prediction of cyclone induced waves in the Bay of Bengal

The present study examines the impact of wind data resolution on wave climate prediction using the wind-wave model, WAM using different spatial resolutions of wind data such as 1/5°, 1/10°, 1/20°, and 1/40°, obtained from the Weather Research and Forecasting (WRF) model. It focuses on the prediction of extreme waves caused by three cyclones (Phailin, Hudhud, and Vardah) that hit the Bay of Bengal region of the North Indian Ocean in 2013, 2014, and 2016, respectively. These cyclones had a significant impact on the east coast of India and have generated significant wave height (Hs) of more than 10 m. To accurately predict these high-impact waves, it is necessary to have suitable wind resolution and a stable wave model grid resolution. We also investigate the effect of different spatial resolutions (1/5°, 1/10°, 1/20°, and 1/40°) on wave prediction. There is no significant difference in Hs prediction accuracy for wind data and wave model resolutions better than 1/10°. Additionally, the WAM model is tested with Ardhuin's input and dissipation source terms included in the model which doubles the source terms computational time required by the model although providing better predictions in the nearshore locations. An optimal model resolution of 1/10°, along with Ardhuin's source terms, is recommended for better wave prediction.

Optimal combined wake and active power control of large‐scale wind farm considering available power

AbstractThe high‐order nonlinear complex characteristics of traditional static wake model (SWM) restrict the fast optimization of wake effect. Therefore, this paper introduces a dynamic wake model (DWM) to describe the time‐varying wake characteristics of wind turbines (WTs) with low computational cost. The traditional SWM is linearized to derive the wake wind speed sensitivity coefficients, which represents the sensitivity of wake wind speed deficit with respect to the active power reference. Considering the natural propagation characteristics of wake effect, a wake delay function is added to realize the future wind speed prediction of different locations of the wind farm. And a joint control strategy of wake and active power based on model predictive control (MPC) is proposed to optimize the power contribution of each WT to minimize wake effects, and maximize total available power. Thus, the increased available power is able to satisfy the power demand from Transmission System Operator (TSO) for ancillary services and enhance power support capability. The control performance of the proposed control strategy is evaluated by simulations under constant and varying incoming wind speed.

Linking Arctic stratospheric polar vortex weakening to rising CO2-induced intensification of the Indo-Pacific warm pool during the past five decades

Accompanying the global rise in greenhouse gas emissions, a warming trend in the Indo-Pacific warm pool (IPWP) has exerted a discernible influence on tropical atmosphere–ocean interactions. However, the impact of this intensification of the IPWP on the Arctic stratospheric polar vortex (ASPV) remains unclear. In this study, we revealed a link between the changes in the IPWP and ASPV during the early winter months, with nearly half of the weakening in the ASPV attributable to the intensification of the IPWP from 1968 to 2020. Wave trains triggered by the elevated SST in the IPWP region lead to enhanced eastward-propagating flux convergence in the northern high-latitude stratosphere, ultimately resulting in a diminishing ASPV. With increasing atmospheric greenhouse gas, the ASPV is poised to further weaken in the future, particularly in the context of a more intense IPWP. Our finding has significant implications for early winter ASPV strength and location prediction and seasonal weather forecasting.

Abstract 17893: Deep Learning for the Prediction of Myocardial Scarring Through Electrocardiograms

Introduction: Cardiac Magnetic Resonance Imaging (MRI) with Late Gadolinium enhancement (LGE), the gold standard for myocardial scarring evaluation, is limited by cost, time, and expertise required. The Electrocardiogram (ECG), in contrast, is a widely available and affordable diagnostic tool. We applied deep learning (DL) models to ECGs to directly predict the presence of LGE on MRI. Methods: Patients who underwent a cardiac MRI and had at least one ECG collected from 2012 to 2022 were considered. Patients were classified into positive and negative for myocardial scarring via their MRI reports, labeled by experts. Patients with non-specific scarring were excluded from the dataset. Patient ECGs most recent to their MRI were utilized as input for the model. MRI reports and ECG waveforms were collected from NYU Langone Health. We implemented a 34-layer Neural Network architecture, adapted from a state-of-the-art arrhythmia detection model, to learn from a 10 second ECG sampled at 250hz. We evaluated the model using area under the receiver-operating curve (AUROC), specificity, sensitivity, and precision. Results: 8813 patients with MRI and ECG were included. 1559 were excluded due to non-specific scarring. 2739 patients were positive for scarring. Patients were split into a 70% train, 10% validation, and 20% test dataset. The deep learning model obtained a test set AUROC of 0.80 [95%CI, 0.77 - 0.81]. At 20% false positive rate, we obtained a sensitivity of 0.63, specificity of 0.80, and precision of 0.64. Conclusions: This study suggests that our DL model can predict for the presence of LGE from a surface ECG. A future direction of this work would be to evaluate the use of DL for the prediction of LGE location, quantity and specific phenotypes of myocardial disease.

Dual-Signal Feature Spaces Map Protein Subcellular Locations Based on Immunohistochemistry Image and Protein Sequence.

Protein is one of the primary biochemical macromolecular regulators in the compartmental cellular structure, and the subcellular locations of proteins can therefore provide information on the function of subcellular structures and physiological environments. Recently, data-driven systems have been developed to predict the subcellular location of proteins based on protein sequence, immunohistochemistry (IHC) images, or immunofluorescence (IF) images. However, the research on the fusion of multiple protein signals has received little attention. In this study, we developed a dual-signal computational protocol by incorporating IHC images into protein sequences to learn protein subcellular localization. Three major steps can be summarized as follows in this protocol: first, a benchmark database that includes 281 proteins sorted out from 4722 proteins of the Human Protein Atlas (HPA) and Swiss-Prot database, which is involved in the endoplasmic reticulum (ER), Golgi apparatus, cytosol, and nucleoplasm; second, discriminative feature operators were first employed to quantitate protein image-sequence samples that include IHC images and protein sequence; finally, the feature subspace of different protein signals is absorbed to construct multiple sub-classifiers via dimensionality reduction and binary relevance (BR), and multiple confidence derived from multiple sub-classifiers is adopted to decide subcellular location by the centralized voting mechanism at the decision layer. The experimental results indicated that the dual-signal model embedded IHC images and protein sequences outperformed the single-signal models with accuracy, precision, and recall of 75.41%, 80.38%, and 74.38%, respectively. It is enlightening for further research on protein subcellular location prediction under multi-signal fusion of protein.

Two-dimensional acoustic emission source localization on layered engineered wood by machine learning: a case study of laminated veneer lumber plate structure

Engineered wood or mass timber has gained increasing popularity in building construction, and layered engineered wood is a major category of mass timber design since it enables manufacturing structural members with a wide range of geometry. Thus, there is a potential rising demand for structural health monitoring on engineered wood-based structural members and buildings. This study investigates the feasibility of using an important and practical acoustic emission (AE) method for damage localization, specifically two-dimensional (2D) AE source localization, in a representative layered engineered wood sample, namely laminated veneer lumber (LVL) plate. While 2D AE source localization is generally straightforward in isotropic materials, the problem becomes challenging for anisotropic materials with angle-dependent wave velocities. It is even more complicated if heterogeneity involves, which turns out to be the case for layered engineered wood. In this study, we rely on the AE feature of difference in time of arrival (dTOA) and develop three methods to address the challenges of 2D AE source localization raised by anisotropy and heterogeneity in an LVL plate. The benchmark velocity profile method (VPM) is first implemented in an LVL plate. With knowledge of the angle-dependent velocity, the source location predictions by the VPM are generally erroneous even with predicted source location outside of the region of interest. Furthermore, the general regression neural network (GRNN) is developed using different combinations of dTOA components, resulting in improved prediction performance. Third, the Gaussian process regression (GPR) is developed by maximizing the marginal likelihood of the training dataset. Moreover, to lessen the computation burden, the lower bound of the logarithm likelihood of the whole models is derived and decomposed through Jensen’s inequality and Bayes’ theorem, providing the theoretical background for training models with different combinations of dTOAs individually.

Research and Application of Intelligent Weather Push Model Based on Travel Forecast and 5G Message

In the realm of daily activity planning, precise weather forecasting services hold paramount significance. However, the prevalent dissemination of weather forecasts through conventional channels like radio, television, and the internet often yields only generalized regional predictions. This limitation contributes to diminished forecast reach, inadequate accuracy, and a lack of individualization, thwarting the effective distribution of meteorological insights and inhibiting the fulfillment of personalized forecast demands. Addressing these concerns, our study proposes a personalized weather forecasting approach that harnesses machine learning techniques and leverages the 5G messaging platform. By amalgamating projected user travel data, we augment personalized weather reports and extend user coverage to achieve tailored, timely, and high-quality weather services. Concretely, our research commences with an extensive analysis of large-scale user travel behavior data to extract pertinent travel attributes. Subsequently, we construct a user’s future location prediction model—dubbed the Loc-PredModel—by employing the Extreme Gradient Boosting (XGBoost) algorithm to forecast users’ trip destinations and arrival times. Anchored in the anticipated outcomes of user travel behavior, personalized weather data reports are formulated. Experimental results underscore the Loc-PredModel’s remarkable predictive prowess, demonstrating a root mean squared error (RMSE) value of 0.208 and a coefficient of determination (R2) value of 0.935, affirming its efficacy in prognosticating users’ trip destinations and arrival times. Furthermore, our 5G message-driven platform, rooted in intelligent personalized meteorological services, underwent testing within Chengdu city and garnered positive user feedback. Our research effectively surmounts the limitations of conventional weather forecasting platforms by furnishing users with more precise and customized weather information predicated on behavioral analysis and the 5G information ecosystem. This study not only advances the theoretical groundwork of intelligent meteorology but also offers invaluable insights and guidance for future advancement. By providing users with a more personalized and timely intelligent meteorological service experience, our approach exhibits transferability, with the research methodology and model potentially extendable nationwide or even on a larger scale beyond the study’s Chengdu-based scope.

Resource Optimization for Task Offloading With Real-Time Location Prediction in Pedestrian-Vehicle Interaction Scenarios

With the development of autonomous driving, task offloading of Internet of vehicles has become a hot research issue. In pedestrian-vehicle interaction scenarios, characteristics of tasks are constantly changing due to the influence of pedestrians and road conditions. Real-time offloading optimization and signaling are time-consuming, which may not meet the low delay requirement of task offloading. Therefore, this paper proposes a location prediction-based resource optimization scheme for task offloading in these scenarios. Firstly, the locations of pedestrians are predicted by the social force model based on their movement rules, and the locations of vehicles are predicted by the car-following model on the basis of ensuring pedestrian safety. The characteristics of tasks are obtained based on the predicted locations of vehicles. Then a neural network trained beforehand based on deep Q-learning is used to obtain a task offloading strategy. Since the tasks are obtained by prediction in advance, this strategy decision can be processed before vehicles arriving the predicted locations, which saves the time consumption of optimization and signaling. Besides, simulation results show that the proposed scheme still guarantees an acceptably low task offloading delay compared with the other methods, especially in congested areas.

FOG/VISIBILITY FORECAST AND VERIFICATION AT IGI AIRPORT, NEW DELHI DURING THE WINTER SEASONS OF 2020-21 & 2021-22

The reduction in visibility due to fog leads to cancellation of flights, delays and diversions at IGI Airport, New Delhi during the winter season. Accurate prediction of fog/ visibility is required within sufficient lead time to make the flight operations safer, economical and more convenient. This study attempts to improve the fog/visibility forecast at IGI airport for the months of December and January during the winter seasons of 2020-21 and 2021-22. Various meteorological parameters required to predict the fog/visibility are obtained from IMD GFS forecasted meteorological information which are valid for next 24 and 48-hours. Forecast verification was performed using the METAR and synoptic observations. The results for 2020-21 fog/visibility forecast for 24-hours are promising with a POD of 0.92, critical success index as 0.68, and false alarm ratio as 0.28. Moreover, the 2021-22 results for 24 -hours fog /visibility forecast are also observed to be promising with a POD of 0.73, critical success index of 0.54 and false alarm ratio at 0.32. This method gives fairly accurate predictions in point locations and can also be used for a larger spatial area. However, the 48-hours forecast performance needs further improvement. The method also predicted the wind speed and relative humidity which were found to be in agreement with the observed data to a great extent. In the upcoming years, the fog/visibility forecast method will be more robust with better prediction accuracy of meteorological parameters from model outputs and observations.

BHE‐YOLO: Effective Small Target Detector for Aluminum Surface Defect Detection

AbstractAluminum surface defect detection is a critical task to make location and classification predictions about defects in the industrial production process. However, defects complexity and the requirement of rapid production have led to a great challenge for existing detection algorithms. In this work, an effective small target detector named BHE‐YOLO is proposed for aluminum surface defect detection. First, the BiFPN is modified and integrated with YOLOv5s to achieve effective weighted feature fusion and cross‐scale connection. Second, the Hard Swish activation function is applied to better extract defect feature information. Third, the Equalized Focal Loss function is introduced to replace the cross‐entropy formula of the negative sample confidence part of the loss function. Finally, the experiments are carried out on the aluminum profile defect dataset of the Aliyun Tianchi Competition. The results demonstrate that BHE‐YOLO has excellent performance in terms of Precision, Recall, F1 score, and mAP, and its superiority over the typical detection algorithms, especially for small target defects of aluminum surface.

Burst Speed Prediction of Impellers With Critical Strain Criteria

Abstract Accurate prediction of burst speed and critical location of a rotating disk is a great concern for turbomachinery designers and enables the reduction of test numbers required to show rotor integrity during turbo-engine certification. Existing burst methodologies mainly focused on axial disks with well-defined bore, web, and rim regions and mostly depended on correction factors that require previous test data of rotors with similar sizes and features. Furthermore, variation of radial and hoop stresses in a radial disk is less predictable than in axial disks due to the asymmetric nature of impellers, which results in uneven distribution of centrifugal and bending loads. This paper aims to explain the experimental and numerical effort dedicated to building an accurate burst methodology for impellers. Elastoplastic finite element analyses are performed by incorporating geometric and material nonlinearities. Analysis results allowed the prediction of failure locations and observation of redistribution of stresses due to material plasticity as rotor speed increases. An extensive test campaign is carried out to validate the numerical analysis. Critical strain criteria that correspond to the plastic strain of true ultimate tensile strength (UTS), are used to determine burst speeds before tests. Experimental results are compared with Robinson's average hoop stress criteria, Hallinan's criteria, and critical strain criteria to show the strong and weak sides of these methodologies for an impeller application. The numerical burst speeds and crack initiation locations associated with critical strain criteria are found to be in good agreement with the spin test results.

A future location prediction method based on lightweight LSTM with hyperparamater optimization

In this study, we presented a method for future location prediction based on machine learning over geopositioning data sets. There are large amounts of geopositioning data sets collected by mobile devices mainly due to modern geopositioning systems such as GPS, GLONASS and Galileo. Based on these geopositioning data sets, it is possible to have a wide variety of location-based services. These data sets can be used for future location prediction of objects, especially humans. Additionally, they have a high possibility for further applications. The purpose of this research is to present a simple and lightweight method that can be applicable to devices with lower computing capability devices, such as AIoT (Artificial Intelligence of Things) or EdgeML (Edge Machine Learning) devices. We introduced a basic LSTM (Long Short Term Memory) model with hyperparameter optimization, especially on window size of continuous geopositioning data, using limited previous geopositioning data for location prediction purposes. We found that the results of using our method for future location prediction are considerably fast and accurate compared with existing neural network-model-based approaches. We also applied our method to non-continuous geopositioning data sets and found it to be equally effective.

Ensemble learning with linguistic, summary language and psychological features for location prediction

Ensemble learning with linguistic, summary language and psychological features for location prediction

A Self-Attention Model for Next Location Prediction Based on Semantic Mining

With the rise in the Internet of Things (IOT), mobile devices and Location-Based Social Network (LBSN), abundant trajectory data have made research on location prediction more popular. The check-in data shared through LBSN hide information related to life patterns, and obtaining this information is helpful for location prediction. However, the trajectory data recorded by mobile devices are different from check-in data that have semantic information. In order to obtain the user’s semantic, relevant studies match the stay point to the nearest Point of Interest (POI), but location error may lead to wrong semantic matching. Therefore, we propose a Self-Attention model for next location prediction based on semantic mining to predict the next location. When calculating the semantic feature of a stay point, the first step is to search for the k-nearest POI, and then use the reciprocal of the distance from the stay point to the k-nearest POI and the number of categories as weights. Finally, we use the probability to express the semantic without losing other important semantic information. Furthermore, this research, combined with sequential pattern mining, can result in richer semantic features. In order to better perceive the trajectory, temporal features learn the periodicity of time series by the sine function. In terms of location features, we build a directed weighted graph and regard the frequency of users visiting locations as the weight, so the location features are rich in contextual information. We then adopt the Self-Attention model to capture long-term dependencies in long trajectory sequences. Experiments in Geolife show that the semantic matching of this study improved by 45.78% in TOP@1 compared with the closest distance search for POI. Compared with the baseline, the model proposed in this study improved by 2.5% in TOP@1.

Hydrogen leakage location prediction at hydrogen refueling stations based on deep learning

Accurate and efficient localization of hydrogen leakage is crucial for ensuring the safe and stable operation of hydrogen refueling stations. In this paper, a hybrid model (CEEMDAN–CNN–LSTM) based on data noise reduction and deep learning is proposed to predict the location of hydrogen leakage in hydrogen refueling stations. The model employs the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm to effectively reduce noise and process 720 hydrogen leakage condition datasets. The CNN-LSTM model is constructed by synergistically combining the spatiotemporal feature extraction capabilities of the convolutional neural network (CNN) with the temporal processing capabilities of the long-short-term memory neural network (LSTM) while optimizing hyperparameter combinations through GridSearch. Compared with other models, the proposed CEEMDAN–CNN–LSTM model demonstrates superior prediction performance, achieving an accuracy of 99.54 %, precision of 99.42 %, recall of 99.54 %, and F1-score of 99.46 %. Moreover, when utilizing a 5-s duration of leakage data as input, the model attains an impressive accuracy of 97.64 %. These results underscore the favorable application prospects of the CEEMDAN–CNN–LSTM model in the field of gas leak localization.

Efficient human 3D localization and free space segmentation for human-aware mobile robots in warehouse facilities.

Real-time prediction of human location combined with the capability to perceive obstacles is crucial for socially-aware navigation in robotics. Our work focuses on localizing humans in the world and predicting the free space around them by incorporating other static and dynamic obstacles. We propose a multi-task learning strategy to handle both tasks, achieving this goal with minimal computational demands. We use a dataset captured in a typical warehouse environment by mounting a perception module consisting of a Jetson Xavier AGX and an Intel L515 LiDAR camera on a MiR100 mobile robot. Our method, which is built upon prior works in the field of human detection and localization demonstrates improved results in difficult cases that are not tackled in other works, such as human instances at a close distance or at the limits of the field of view of the capturing sensor. We further extend this work by using a lightweight network structure and integrating a free space segmentation branch that can independently segment the floor space without any prior maps or 3D data, relying instead on the characteristics of the floor. In conclusion, our method presents a lightweight and efficient solution for predicting human 3D location and segmenting the floor space for low-energy consumption platforms, tested in an industrial environment.

Divide-and-conquer DNN approach for the inverse point source problem using a few single frequency measurements

We consider the inverse problem to determine the number and locations of acoustic point sources from single low-frequency partial data. The problem is particularly challenging in the sense that the data is available only at a few locations which span a small aperture. Integrating the deep neural networks (DNNs) and Bayesian inversion, we propose a divide-and-conquer approach by dividing the inverse problem into three subproblems. The first subproblem is to determine the number of point sources, which is formulated as a common machine learning task—classification. A simple DNN is proposed and trained to predict the numbers of the point sources. The second subproblem is to reconstruct the (approximate) locations of the point sources. We formulate the problem as a nonlinear function with the input being the measured data and the output being a carefully elaborated location vector. Then a second DNN is proposed to learn the mapping and predict the location vector effectively. The location vector is post-processed to provide an indicator (image) function for the (approximate) locations of the point sources. The third subproblem is to improve the accuracy of the location prediction, for which we employ a Bayesian inversion algorithm. This divide-and-conquer approach can effectively treat both phase and phaseless data as demonstrated by various examples.