Data-driven Model Research Articles

A data-driven agent-based model of primary school segregation in Amsterdam

ABSTRACT Theoretical agent-based models of residential and school choice have shown that substantial segregation can emerge as an (unintended) consequence of interactions between individual households and feedback mechanisms, despite households being relatively tolerant. However, for school choice, existing models have mostly been highly stylized, leaving open whether they are relevant for understanding school segregation in concrete empirical settings. To bridge this gap, this study develops an empirically calibrated agent-based model focusing on primary school choice in Amsterdam. Consistent with existing models, results show that substantial school segregation emerges when schools are chosen based on a trade-off between composition and distance, and also when households are relatively tolerant. Additionally, findings of (hypothetical) policy simulations suggest that it is important to understand which preferences for school composition and distance households have and how these interact. We find that the effects of policies aiming to reduce school segregation through geographical restricting mechanisms are highly dependent on those interacting preferences. Also, we assessed the contribution of residential segregation to school segregation. Our findings may have implications for methodologies aiming to estimate school choice preferences, such as discrete choice models, as these methodologies do not explicitly control for implications of these interactions and feedback mechanisms, which might lead to incorrect inference.

Assessment of Roughness Characterization Methods for Data-Driven Predictions

AbstractA comparative analysis is undertaken to explore the impact of various roughness characterization methods as input variables on the performance of data-driven predictive models for estimating the roughness equivalent sand-grain size $$k_s$$ k s . The first type of model, denoted as $$\text {ENN}_\text {PS}$$ ENN PS , incorporates the roughness height probability density function (p.d.f.) and power spectrum (PS), while the second type of model, $$\text {ENN}_\text {PA}$$ ENN PA , utilizes a finite set of 17 roughness statistical parameters as input variables. Furthermore, a simplified parameter-based model, denoted as $$\text {ENN}_\text {PAM}$$ ENN PAM , is considered, which features only 6 input roughness parameters. The models are trained based on identical databases and evaluated using roughness samples similar to the training databases as well as an external testing database based on literature. While the predictions based on p.d.f. and PS achieves a stable error level of around 10% among all considered testing samples, a notable deterioration in performance is observed for the parameter-based models for the external testing database, indicating a lower extrapolating capability to diverse roughness types. Finally, the sensitivity analysis on different types of roughness confirms an effective identification of distinct roughness effects by $$\text {ENN}_\text {PAM}$$ ENN PAM , which is not observed for $$\text {ENN}_\text {PA}$$ ENN PA . We hypothesize that the successful training of $$\text {ENN}_\text {PAM}$$ ENN PAM is attributed to the enhanced training efficiency linked to the lower input dimensionality.

Robust maximum fairness consensus models with aggregation operator based on data-driven method

In group decision-making (GDM), when decision-makers (DMs) feel it is unfair, they may take uncooperative measures to disrupt the consensus-reaching process (CRP). On the other hand, it is difficult for the moderator to objectively determine each DM’s unit consensus cost and weight in CRP. Hence, this paper proposes data-driven robust maximum fairness consensus models (RMFCMs) to address these. First, this paper uses the robust optimization method to construct multiple uncertainty sets to describe the uncertainty of the DMs’ unit adjustment cost and proposes the RMFCMs. Subsequently, based on the DMs’ historical data, the DMs’ weights in the CRP are determined by a data-driven method based on the kernel density estimation (KDE) method. Finally, this paper also applies the proposed models to the carbon emission reduction negotiation process between governments and enterprises, and the experimental results verify the rationality and robustness of the proposed consensus model.

Research on Urban Architecture Design under the Concept of Smart City

After the digital city and smart city concepts, the latest proposed smart city concept is a general summary of the advanced form of urban information development, which plays an important role in the economic transformation of cities, social governance, improvement of residents' quality of life and energy conservation and environmental protection. The smart city puts forward new requirements for the architectural design of the city, and how to reasonably use the smart city concept in the architectural design is still a major issue for architects and professionals to think about. To this end, this paper proposes a state tracking based full working condition modeling method for nonlinear industrial systems. For the problem that the amount of historical data is too large and it is difficult to screen the modeling data, a sliding window is designed to screen the steady-state data, and a fast recursive algorithm of standard deviation in the window is derived; the influence mechanism of unknown disturbances on the system is analyzed, and the data modeling by dynamic regression to the steady state is selected, and a data-driven modeling algorithm that can eliminate the influence of disturbances is proposed; the model information contained in the process industrial big data is used to apply the higher order function to fit each A linear variable parameter transfer function model based on the characteristic parameters is proposed. The identification of an industrial process is shown to be effective. The main points of urban architectural design should be noted under the concept of smart city from four aspects: safety, diversity, environmental protection and emotionality.

Analysis of window opening in arctic community housing and development of data-driven models

Achieving low-energy, comfortable buildings in remote Arctic regions like Nunavik, Canada, presents multiple challenges due to high heating demand and limited energy supply. Understanding human behavior and the impact of practices on energy use is a key element in improving building performance. While previous research on building performance in Arctic regions has often adopted a building-centered approach, this study offers a complementary approach by investigating occupant behavior, focusing on window openings. The sample includes 10 monitored houses in the village of Quaqtaq, Nunavik. Data on weather conditions and window and door status were collected from September 2018 to August 2020. Behavior trends were identified using metrics such as the opening ratio, median duration of an opening, and number of openings. Correlations between opening/closing actions and the conditions of the outdoor environment were analyzed. Findings indicate that occupants interact more with common room windows and tend to leave bedroom windows open for longer periods. Time of the day, temperature, and relative humidity influence both opening and closing actions. Solar irradiance is also an important driver for open actions, while mean wind speed is significant in predicting closing actions. Data-driven models were developed using logistic regression. Comparison between predicted and real data demonstrated a good performance in estimating the opening ratio over the year and seasonal variations but a tendency to overestimate the number of openings and underestimate the duration of an opening.

Production Prediction Model of Tight Gas Well Based on Neural Network Driven by Decline Curve and Data

The accurate prediction of gas well production is one of the key factors affecting the economical and efficient development of tight gas wells. The traditional oil and gas well production prediction method assumes strict conditions and has a low prediction accuracy in actual field applications. At present, intelligent algorithms based on big data have been applied in oil and gas well production prediction, but there are still some limitations. Only learning from data leads to the poor generalization ability and anti-interference ability of prediction models. To solve this problem, a production prediction method of tight gas wells based on the decline curve and data-driven neural network is established in this paper. Based on the actual production data of fractured horizontal wells in three tight gas reservoirs in the Ordos Basin, the prediction effect of the Arps decline curve model, the SPED decline curve model, the MFF decline curve model, and the combination of the decline curve and data-driven neural network model is compared and analyzed. The results of the case analysis show that the MFF model and the combined data-driven model have the highest accuracy, the average absolute percentage error is 14.11%, and the root-mean-square error is 1.491, which provides a new method for the production prediction of tight gas wells in the Ordos Basin.

Intrinsic functional connectivity of motor and heteromodal association cortex predicts individual differences in regulatory focus.

Regulatory focus theory (RFT) describes two cognitive-motivational systems for goal pursuit-the promotion and prevention systems-important for self-regulation and previously implicated in vulnerability to psychopathology. According to RFT, the promotion system is engaged in attaining ideal goals (e.g. hopes and dreams), whereas the prevention system is associated with accomplishing ought goals (e.g. duties and obligations). Prior task-based functional magnetic resonance imaging (fMRI) studies have mostly explored the mapping of these two systems onto the activity of a priori brain regions supporting motivation and executive control in both healthy and depressed adults. However, complex behavioral processes such as those guided by individual differences in regulatory focus are likely supported by widely distributed patterns of intrinsic functional connectivity. We used data-driven connectome-based predictive modeling to identify patterns of distributed whole-brain intrinsic network connectivity associated with individual differences in promotion and prevention system orientation in 1,307 young university volunteers. Our analyses produced a network model predictive of prevention but not promotion orientation, specifically the subjective experience of successful goal pursuit using prevention strategies. The predictive model of prevention success was highlighted by decreased intrinsic functional connectivity of both heteromodal association cortices in the parietal and limbic networks and the primary motor cortex. We discuss these findings in the context of strategic inaction, which drives individuals with a strong dispositional prevention orientation to inhibit their behavioral tendencies in order to shield the self from potential losses, thus maintaining the safety of the status quo but also leading to trade-offs in goal pursuit success.

A multidimensional AI-trained correction to the 1D approximate model for Airborne TDEM sensing

The computational resources required to solve the full 3D inversion of time-domain electromagnetic data are immense. To overcome the time-consuming 3D simulations, we construct a surrogate model, more precisely, a data-driven statistical model to replace the 3D simulations. It is trained on 3D data and predicts the approximate output much faster. We construct a surrogate model that predicts the discrepancy between a 1D subsurface model and a deviation of the 1D assumption. The latter response is fastly computable with a semi-analytical 1D forward model. We exemplify the approach on a two-layered case. The results are encouraging even with few training samples. Given the computational cost related to the 3D simulations, there are limitations in the number of training samples that can be generated. In addition, certain applications require a wide range of parameters to be sampled, such as the electrical conductivity parameters in a saltwater intrusion case. The challenge of this work is achieving the best possible accuracy with only a few thousand samples. We propose to view the performance in terms of learning gain, representing the gain from the surrogate model whilst still acknowledging a residual discrepancy. Our works open new avenues for effectively simulating 3D TDEM data.

Mathematical models and analysis tools for risk assessment of unnatural epidemics: a scoping review

Predicting, issuing early warnings, and assessing risks associated with unnatural epidemics (UEs) present significant challenges. These tasks also represent key areas of focus within the field of prevention and control research for UEs. A scoping review was conducted using databases such as PubMed, Web of Science, Scopus, and Embase, from inception to 31 December 2023. Sixty-six studies met the inclusion criteria. Two types of models (data-driven and mechanistic-based models) and a class of analysis tools for risk assessment of UEs were identified. The validation part of models involved calibration, improvement, and comparison. Three surveillance systems (event-based, indicator-based, and hybrid) were reported for monitoring UEs. In the current study, mathematical models and analysis tools suggest a distinction between natural epidemics and UEs in selecting model parameters and warning thresholds. Future research should consider combining a mechanistic-based model with a data-driven model and learning to pursue time-varying, high-precision risk assessment capabilities.

A robust optimization method for new distribution systems based on adaptive data-driven polyhedral sets

In order to better describe the uncertainty of renewable energy output, this paper proposed a novel robust optimization method for new distribution systems based on adaptive data-driven polyhedral sets. First, an ellipsoidal uncertainty set was established using historical data on renewable energy output, and a data-driven convex hull polyhedral set was established by connecting high-dimensional ellipsoidal vertices; on this basis, an adaptive data-driven polyhedral set model was established to address the problem of high conservatism in the scaling process of convex hull polyhedral sets. Furthermore, a novel adaptive data-driven robust scheduling model for new distribution systems was established, and a column-and-constraint generation (C&CG) algorithm was used to solve the robust scheduling model. Finally, the improved IEEE-33 bus system simulation verification shows that the robust scheduling model for new distribution systems based on adaptive data-driven polyhedral sets can reduce conservatism and improve the robustness of optimization results.

Abstract PO5-24-09: Optimizing the Therapeutic Treatment of Breast Cancer Through a Dual Data-driven Theoretical Model

Abstract Despite rapid advancements in treatments and technology, breast cancer remains the second leading cause of cancer related morbidity in women in the US. As a heterogeneous disease, the unique genetic and phenotypic profile of each tumor complicates the prediction of treatment responses and optimized treatment strategies. One primary manifestation of phenotypic plasticity and concomitant intra-tumoral heterogeneity is the Epithelial-Mesenchymal Transition (EMT). The acquisition of migratory and invasive properties through EMT facilitates tumor metastasis and results in worse patient outcomes. As a result, the EMT and its reverse process (MET) often influence treatment response. However, the different mechanisms through which these therapeutic drugs affect the EMT-related phenotypic switching rates as well as each phenotype’s replication rates are unclear, and further investigation is required for better understanding their precise effect for directing optimal treatment strategies and improved patient outcomes. Here, we compile an atlas of publicly available metastatic and non-metastatic RNA sequencing data of breast cancer cell lines treated with therapeutic and chemotherapeutic drugs and propose a dual data-driven theoretical framework for understanding the effects of each drug on breast cancer treatment response. We infer the cell fractions, EMT phenotypic transition rates and phenotypic replication rates of each case through an extension of our previously developed hybrid data-driven and theoretical model, COMET. We infer single-cell RNA sequencing-derived EMT-related trajectories and apply a continuous time Markov chain model to infer inter-state transition rates. By analyzing the atlas of RNA sequencing data and integration with our framework, we gain novel insights into the context-specific mechanisms by which therapeutic drugs affect the phenotypic transition rates. This comprehensive understanding will contribute to optimizing breast cancer therapy and ultimately improving patient outcomes. Citation Format: A. Najafi. Optimizing the Therapeutic Treatment of Breast Cancer Through a Dual Data-driven Theoretical Model [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-24-09.

Abstract PO5-19-01: The HER2-RADiCAL study (Response ADaptive CAre pLan) – Tailoring treatment for HER2 positive early breast cancer

Abstract Background The presence or absence of residual disease following neoadjuvant systemic anti-cancer therapy (neoSACT) for HER2-positive early breast cancer (HER2+ EBC) provides powerful prognostic information. Pathological complete response (pCR) following neoSACT, particularly in patients with earlier clinical TNM stage at diagnosis, identifies a population with excellent survival outcomes in whom the balance of toxicity associated with the current treatment pathway may be disproportionate to absolute clinical benefit. Aims >HER2-RADiCAL seeks to reduce the treatment burden and healthcare costs of treating HER2+ EBC by testing the hypothesis that pCR can be used as a functional response biomarker to select patients who can safely receive less systemic therapy with minimal/no loss of efficacy. Trial design and eligibility criteria HER2-RADiCAL is a response-directed interventional cohort/threshold-crossing design study embedded within a real-world data-driven clinical pathway model. Patients with ER-positive or negative HER2+ EBC with cT1N1 or cT2N0-1 stage at diagnosis are eligible after completion of standard of care neoSACT and locally determined pCR (ypT0/Tis ypN0). Patients with pathological features consistent with previous malignant involvement in >4 nodes are not eligible. Adjuvant trastuzumab +/- pertuzumab may have continued prior to study entry provided no more than 9 cycles of trastuzumab are received. After registration all participants receive a total of 9 cycles of trastuzumab including those administered prior to study entry. Participants receive no further pertuzumab and no adjuvant chemotherapy. Adjuvant endocrine therapy and radiotherapy are given as per standard of care. Central pathology review of pCR status will be conducted in a subset of cases. A Study Within a Trial (SWAT) will explore identification of factors influencing patient decision-making for participation in studies of response-adapted treatment. Statistical methods The primary endpoint is relapse-free interval. Recruitment of 720 participants over 3.5 years will provide 90% power to exclude an event rate >6.5% at 3 years (expected event rate ≤4%). Secondary endpoints include relapse-free survival, invasive breast cancer-free survival, invasive disease-free survival, distant recurrence-free interval, breast cancer-free interval, overall survival, treatment pathway adherence and cost-effectiveness. An interventional cohort is preferred to a randomised non-inferiority trial based on the known event rate in this population which is sufficiently low such that any deviation from this expected event rate would indicate failure of the reduced therapy strategy. Real-world data driven clinical pathway model Health economic modelling will compare the protocol-driven study cohort with two comparator pathways: a non-response adapted maximum therapy pathway (the standard clinical pathway prior to the study) and a real-world representative pathway derived from anonymised data for all patients treated for HER2+ EBC within the UK NHS. Patient and public involvement Our PPI partners continue to shape this research, holding key roles as trial management group members, overseeing progress of the study, contributing to protocol amendment development, and design and implementation of the SWAT. Current status The first patient was enrolled in December 2021. At 10th July 2023, 25 research sites have opened with 39 participants recruited. Site opening continues with a minimum target of 40. Responding to continued use of anthracycline-based regimens in UK practice, eligibility criteria have recently been extended to permit entry of patients who have received anthracycline-containing neoSACT. For further information contact her2radical-icrctsu@icr.ac.uk. Citation Format: Iain Macpherson, Stuart McIntosh, Lucy Kilburn, Holly Tovey, Laura Moretti, Katie Goddard, Indrani Bhattacharya, Clinton Boyd, Cliona Kirwan, Ciara O'Brien, Duncan Wheatley, Jennifer Glendenning, Mairead MacKenzie, Hilary Stobart, Claire Snowdon, Andrew Wardley, Abeer Shaaban, Peter Hall, David Cameron, Judith Bliss. The HER2-RADiCAL study (Response ADaptive CAre pLan) – Tailoring treatment for HER2 positive early breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-19-01.

Abstract PO5-07-03: Integrating mechanism-based and data-driven modeling to predict the response of triple negative breast cancer to therapy

Abstract Introduction Neoadjuvant chemotherapy (NAC) is the standard-of-care (SOC) for patients with locally advanced triple-negative breast cancer (TNBC) [1]. Unfortunately, only about half of TNBC patients achieve a pathological complete response (pCR) at the completion of NAC [2]. With the recent FDA approval of immunotherapy, the response rate improved by about 8% [3]. To try to further improve NAC response rates, we build upon methods that predict treatment response to optimize interventions and outcomes. Our previous studies have demonstrated that excellent predictive accuracy of TNBC response to NAC was achieved through calibrating mechanism-based models to both pre- and on- treatment imaging data [4]. However, by requiring on-treatment imaging data, therapy optimization can only occur after a patient begins NAC. Here, our goal is to relax the requirement of on-treatment imaging data such that we can predict, and potentially optimize, a patient’s response to NAC before initiating it. We will accomplish this through combining data-driven approaches with mechanism-based modeling to obtain spatiotemporal predictions of tumor response prior to NAC. More specifically, we integrate a reaction-diffusion equation based mathematical model with a U-Net based convolutional neural network (CNN). Methods The reaction-diffusion equation models the development of tumor cellularity over time as the sum of tumor cell diffusion, controlled by a fixed diffusivity, and tumor cell proliferation, controlled by a calibrated net proliferation rate. The net proliferation rate implicitly accounts for both proliferative and drug induced death effects. We calibrated the net proliferation rates in this model on a patient-specific basis for 128 patients from the ARTEMIS trial (NCT02276443) using three imaging timepoints – pre- (V1), post two cycles (V2), and post four cycles (V3) of A/C. We then employed a CNN to characterize the relationship between the calibrated net proliferation rates and the pre-treatment imaging data. The CNN takes pretreatment imaging inputs of the apparent diffusion coefficient map, percent enhancement maps from the dynamic contrast enhanced MRI timecourse, and precontrast T1 map. The CNN outputs an estimate of the calibrated net proliferation rate which can be used to run the mathematical model forward in time to predict cellularity maps at V2/3. In training the CNN, our loss considers both the concordance correlation coefficient (CCC) and the normalized root mean square error (NRMSE). We calculated loss on the net proliferation rate predictions as NRMSE with a CCC penalty to aid in characterizing the non-Gaussian distribution of the net proliferation rate. We calculated loss on the cellularity predictions as the CCC between predicted and measured voxel-wise cellularity. The CNN was trained using the Adam optimizer and a fivefold cross validation with a withheld test set of 26 patients. Results Using the CNN predictions to forward run the mechanism-based model in time to V3 we calculated change in total tumor cellularity and volume from V1 to V3. In the test cohort, we obtained CCC values between the predicted and measured changes of 0.95 and 0.91 for cellularity and volume, respectively. Discussion and summary Through integrating mechanistic modeling and deep learning, we can accurately predict the spatiotemporal development of TNBC response to NAC using only pretreatment imaging data. We will build upon these results by expanding our model to include separate proliferation and drug induced death terms to allow for therapy intervention and optimization. Additionally, we will incorporate other pretreatment data types including genetic data. [1] Liedtke et al., J of Clin Oncol, 2008. [2] Spring et al., Clin Cancer Res, 2020. [3] Schmid et al., NEJM, 2020. [4] Wu et al., Cancer Res, 2022. Citation Format: Casey Stowers, Chengyue Wu, Sidharth Kumar, Zhan Xu, Clinton Yam, Jong Bum Son, Jingfei Ma, Jonathan Tamir, Gaiane Rauch, Thomas Yankeelov. Integrating mechanism-based and data-driven modeling to predict the response of triple negative breast cancer to therapy [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-07-03.

Learning-based adaption of robotic friction models

In the Fourth Industrial Revolution, wherein artificial intelligence and the automation of machines occupy a central role, the deployment of robots is indispensable. However, the manufacturing process using robots, especially in collaboration with humans, is highly intricate. In particular, modeling the friction torque in robotic joints is a longstanding problem due to the lack of a good mathematical description. This motivates the usage of data-driven methods in recent works. However, model-based and data-driven models often exhibit limitations in their ability to generalize beyond the specific dynamics they were trained on, as we demonstrate in this paper. To address this challenge, we introduce a novel approach based on residual learning, which aims to adapt an existing friction model to new dynamics using as little data as possible. We validate our approach by training a base neural network on a symmetric friction data set to learn an accurate relation between the velocity and the friction torque. Subsequently, to adapt to more complex asymmetric settings, we train a second network on a small dataset, focusing on predicting the residual of the initial network’s output. By combining the output of both networks in a suitable manner, our proposed estimator outperforms the conventional model-based approach, an extended LuGre model, and the base neural network significantly. Furthermore, we evaluate our method on trajectories involving external loads and still observe a substantial improvement, approximately 60%–70%, over the conventional approach. Our method does not rely on data with external load during training, eliminating the need for external torque sensors. This demonstrates the generalization capability of our approach, even with a small amount of data – less than a minute – enabling adaptation to diverse scenarios based on prior knowledge about friction in different settings.

Well-Candidate-Recognition Solution Offers Time Savings, Production Enhancement

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 213329, “Data-Driven Well-Candidate-Recognition Solution: Case Study in a Digital Oil Field in Abu Dhabi,” by Erismar Rubio, SPE, Nagaraju Reddicharla, SPE, and Mayada Ali Sultan Ali, ADNOC, et al. The paper has not been peer reviewed. _ A well-candidate-recognition (WCR) data-analytics solution was developed to expedite the process of identifying unhealthy wells that may require rig or rigless interventions based on data integration, automation, and advanced data-driven models. The solution expedites the well-performance-review process to pinpoint candidates for stimulation, nitrogen lift, gas lift conversion, and water or gas shutoff, providing a flexible visualization platform to highlight hidden well-performance insight. Introduction Mature oil fields will face challenges in terms of increasing water cut, lack of pressure support, and the requirement of artificial lift. The critical concern is prioritization of remedial actions that are economically attractive with high returns and low risk. In the past, the asset was prioritizing the rig and rigless intervention candidates on a manual basis with no technical framework, mostly relying on engineers’ backgrounds and experience. Lack of understanding and failure to apply lessons learned from previous jobs were frequent outcomes because no single, comprehensive database or documentation existed that effectively captured well-intervention history. Every year, the operator requested a well-performance-review report for every reservoir containing hundreds of diagnostic plots that required massive manual updates, although the subject asset is a digital oil field fully instrumented with real-time data streaming. However, no tools were available to integrate periodic and nonperiodic required data. The complete paper addresses these challenges of integrating huge amounts of data and model frameworks and developing a systematic workflow to identify opportunities for production enhancement. Data-driven models, combined with the integration of static and dynamic data, enable proactive surveillance routines that allow engineers to focus on problematic wells and opportunity generation in a timely manner. Objective The WCR data-analytics solution was developed as part of the operator’s digital transformation strategy to streamline the process of improving diagnostics, identifying unhealthy wells, and reviewing value gain. The solution aims to help achieve the following objectives: - Facilitate well-performance review to pinpoint candidates for rig and rigless intervention - Provide a flexible visualization platform to highlight hidden well performance insight - Integration of real-time data and official databases - Create a collaborative environment for improved decision-making - Improve rigless success factor for the most-expensive activities - Optimize and prioritize the reservoir-monitoring plan (RMP)

Utilizing sequential modeling in collaborative method for flood forecasting

Flood forecasting has been a major challenge in hydrology for decades. A variety of approaches have been developed, including numerical models and data-driven models, such as Deep Learning (DL), for disaster early-warning and monitoring systems. This study proposed two combination models, exploring different feature separation techniques to improve the 12-hour multi-step discharge prediction. Specifically, two different approaches were implemented, namely the separation of wet and dry periods (Combination Model No. 1; CM-1) and the separation of flood and non-flood events (Combination Model No. 2; CM-2) for comparison with the baseline model trained on an entire dataset. Multivariate time series discharge data were used, measured at upstream and downstream stations within the same river basins. Experiments were conducted in three case study areas in Thailand: the Lower Loei River Basin, the Upper Nan River Basin, and the Upper Chi River Basin.In the proposed model combination approach, three deep-learning models were investigated: vanilla recurrent neural networks (RNNs) and their variants, long short-term memory networks (LSTMs), and gated recurrent units (GRUs). In general, CM-2 exhibited superior performance compared to the other models in most of the experimented neural networks, while the CM-1 approach outperformed the baseline model. This emphasized the advantages of our proposed feature separation framework, particularly in improving model performance, with respect to flood events. However, some variations were observed in the Upper Chi River Basin because the more complex models, such as LSTMs and GRUs, possibly overfitted the training set with less fluctuating data. Our study highlighted the significant contribution of combining two models through feature separation, tailored to specific periods, thereby enhancing the accuracy of flood predictions.

A reduced model for pilot-scale vacuum-enhanced air gap membrane distillation (V-AGMD) modules: Experimental validation and paths for process improvement

A fast and robust computational model of a spiral-wound vacuum-enhanced air gap membrane distillation (V-AGMD) module at the pilot-scale is proposed and implemented. In contrast with data-driven models available in the literature, a physics-based approach is adopted for more reliable generalization beyond the validation dataset. A total of 86 experimental results, of which 41 are described in this work and 45 come from independent sources available in the literature, are used in the validation effort with quite favorable results. With the confidence on the robustness of the methodology due to the wide range of operational parameters and the use of spiral-wound modules of four different sizes included in the validation comparisons, a physical analysis is conducted varying the air gap pressure, the number of feedwater channels, the feedwater flow rate, and the membrane area. Improvements of up to 60% in both water productivity and energy efficiency can be achieved by intensifying the vacuum in the air gap or decreasing the number of feedwater channels. These parameters achieve performance gains due to less resistance in the air gap for vapor to migrate through it, in the former case, and a reduced temperature polarization effect, in the latter case. Smaller flow rates favor energy efficiency at the expense of water productivity by simultaneously decreasing transport-phenomena-related irreversibility and the partial pressure difference across the membrane and the air gap. In addition, this tradeoff between energy efficiency and driving force is shown to lead to an optimum value for the membrane area beyond which the permeate flow rate through the membrane starts to fall due to the small driving force. An illustrative case is predicted to achieve energy efficiency metrics, such as a gain-output ratio of 12.7, competitive with multi-effect distillation.

An information entropy-based fuzzy stochastic configuration network for robust data modeling

In data-driven modeling, the generalization performance of a neural network based on the minimum mean square error criterion may be affected when the samples contain noise. To handle uncertain data modeling problems, this study contributes a novel robust data modeling method called information entropy-based fuzzy stochastic configuration networks (IEFSCNs), which incorporates the maximum correntropy criterion (MCC) to reflect the squared error of the training sample set and uses the quantized minimum error entropy (QMEE) criterion to mine the distribution structure of the modeling data and suppress outlier data. Fuzzy systems are designed to generate nodes in the fuzzy transformation layer. A nonlinear enhancement layer is established as a hidden layer to enhance the representational capacity of the target system. The fuzzy transformation and nonlinear enhancement layers are fully connected to determine the outputs of the IEFSCNs. In addition, an incremental parameter learning algorithm is developed based on a supervision mechanism for optimizing the cost function, and its convergence property is demonstrated. Owing to the superiorities of the cost function by synthesizing MCC and QMEE, IEFSCNs can improve the modeling accuracy and reduce noise suppression. The performances are evaluated using various regression and classification tasks. Compared with existing methods, including SCNs, RVFL, F-SCNs, RSC-KDE, RSC-MCC, and SM-RSC, the proposed method demonstrates an effective robust modeling performance by reducing network structure redundancy while enhancing its ability to suppress noisy data.

Data-driven modeling of the aerodynamic deformation and drag for a freely moving drop in the sub-critical Weber number regime

Accurate prediction of the dynamics and deformation of freely moving drops is crucial for numerous droplet applications. When the Weber number is finite but below a critical value, the drop deviates from its spherical shape and deforms as it is accelerated by the gas stream. Since aerodynamic drag on the drop depends on its shape oscillation, accurately modeling the drop shape evolution is essential for predicting the drop’s velocity and position. In this study, 2D axisymmetric interface-resolved simulations were performed to provide a comprehensive dataset for developing a data-driven model. Parametric simulations were conducted by systematically varying the drop diameter and free-stream velocity, achieving wide ranges of Weber and Reynolds numbers. The instantaneous drop shapes obtained in simulations are characterized by spherical harmonics. Temporal data of the drag and modal coefficients are collected from the simulation data to train a Nonlinear Auto-Regressive models with eXogenous inputs (NARX) neural network model. The overall model consists of two multi-layer perceptron networks, which predict the modal coefficients and the drop drag, respectively. The drop shape can be reconstructed with the predicted modal coefficients. The model predictions are validated against the simulation data in the testing set, showing excellent agreement for the evolutions of both the drop shape and drag.

Global evidence for joint effects of multiple natural and anthropogenic drivers on soil nitrogen cycling.

Global soil nitrogen (N) cycling remains poorly understood due to its complex driving mechanisms. Here, we present a comprehensive analysis of global soil δ15N, a stable isotopic signature indicative of the N input-output balance, using a machine-learning approach on 10,676 observations from 2670 sites. Our findings reveal prevalent joint effects of climatic conditions, plant N-use strategies, soil properties, and other natural and anthropogenic forcings on global soil δ15N. The joint effects of multiple drivers govern the latitudinal distribution of soil δ15N, with more rapid N cycling at lower latitudes than at higher latitudes. In contrast to previous climate-focused models, our data-driven model more accurately simulates spatial changes in global soil δ15N, highlighting the need to consider the joint effects of multiple drivers to estimate the Earth's N budget. These insights contribute to the reconciliation of discordances among empirical, theoretical, and modeling studies on soil N cycling, as well as sustainable N management.