Future Step Research Articles

Deep Curious Feature Selection: A Recurrent, Intrinsic-Reward Reinforcement Learning Approach to Feature Selection

Feature selection (FS) is an important step in the process of building machine learning-based models. The goal of the feature selection step is to find a small subset of features that will provide good prediction results by removing noisy, irrelevant, or repetitive features. Commonly used Wrapper methods use a machine learning model as a black box and its performance as the goal function for evaluating different features' subsets and selecting the best one. To avoid examining all possible subsets (an NP-hard problem), search algorithms are used to find the subsets to be examined, in a heuristic way. As exhaustive search methods are computationally complicated, most methods use simple and greedy search methods that yield only locally optimal results and are not sensitive to possible features interactions, which means that a feature may be chosen at the expense of two others that are more informative together. We analyze the problem of searching the features subset space with reference to two dimensions, namely, memory of past selected features subset and future selected features. We propose a new wrapper feature selection method based on the deep artificial curiosity framework, which implements intrinsic reward reinforcement learning, with long short-term memory unit (LSTM). This novel algorithm integrates these two elements of memory and future step. We show that our method, called the Deep Curious Feature Selection (DCFS) algorithm, deals with feature interactions, and provides a feature subset which improves the accuracy of learning models on artificial and real datasets.

Set up a supply chain observatory through the comparison of multi-criteria parsimonious methods

PurposeThis study presents the development of a supply chain (SC) observatory, which is a benchmarking solution to support companies within the same industry in understanding their positioning in terms of SC performance.Design/methodology/approachA case study is used to demonstrate the set-up of the observatory. Twelve experts on automatic equipment for the wrapping and packaging industry were asked to select a set of performance criteria taken from the literature and evaluate their importance for the chosen industry using multi-criteria decision-making (MCDM) techniques. To handle the high number of criteria without requiring a high amount of time-consuming effort from decision-makers (DMs), five subjective, parsimonious methods for criteria weighting are applied and compared.FindingsA benchmarking methodology is presented and discussed, aimed at DMs in the considered industry. Ten companies were ranked with regard to SC performance. The ranking solution of the companies was on average robust since the general structure of the ranking was very similar for all five weighting methodologies, though simplified-analytic hierarchy process (AHP) was the method with the greatest ability to discriminate between the criteria of importance and was considered faster to carry out and more quickly understood by the decision-makers.Originality/valueDeveloping an SC observatory usually requires managing a large number of alternatives and criteria. The developed methodology uses parsimonious weighting methods, providing DMs with an easy-to-use and time-saving tool. A future research step will be to complete the methodology by defining the minimum variation required for one or more criteria to reach a specific position in the ranking through the implementation of a post-fact analysis.

The joint effects of mixture exposure to multiple meteorological factors on step count: A panel study in China

The public health burden of increasing extreme weather events has been well documented. However, the influence of meteorological factors on physical activity remains limited. Existing mixture effect methods cannot handle cumulative lag effects. Therefore, we developed quantile g-computation Distributed lag non-linear model (QG-DLNM) by embedding a DLNM into quantile g-computation to allow for the concurrent consideration of both cumulated lag effects and mixture effects. We gathered repeated measurement data from Henan Province in China to investigate both the individual impact of meteorological factor on step counts using a DLNM, and the joint effect using the QG-DLNM. We projected future step counts linked to changes in temperature and relative humidity driven by climate change under three scenarios from the sixth phase of the Coupled Model Intercomparison Project. Our findings indicate there are inversed U-shaped associations for temperature, wind speed, and mixture exposure with step counts, peaking at 11.6 °C in temperature, 2.7 m/s in wind speed, and 30th percentile in mixture exposure. However, there are negative associations between relative humidity and rainfall with step counts. Additionally, relative humidity possesses the highest weights in the joint effect (49% contribution). Compared to 2022s, future step counts are projected to decrease due to temperature changes, while increase due to relative humidity changes. However, when considering both future temperature and humidity changes driven by climate change, the projections indicate a decrease in step counts. Our findings may suggest Chinese physical activity will be negatively influenced by global warming.

Vaccination Training for Pharmacy Undergraduates as a Compulsory Part of the Curriculum?-A Multicentric Observation.

In order to increase vaccination rates, the Government of Germany introduced vaccination against influenza and COVID-19 into the regular care administered by pharmacists. However, vaccination training is yet not integrated into the German pharmacy curriculum. Therefore, the Institute for Clinical Pharmacy and Pharmacotherapy in Duesseldorf had developed an innovative vaccination course using high-fidelity simulation for students. To investigate the acceptance further, the course was carried out at three different German universities (Bonn, Duesseldorf, Greifswald). Students were asked to give their self-assessment before and after and satisfaction only after the training course. Responses from 33 participants from the University of Bonn, 42 from the University of Duesseldorf and 49 from the University of Greifswald were analyzed. Every participant at the respective universities showed a significant increase in their self-assessment and indicated a high level of satisfaction with the course. The results also did not differ significantly between the respective universities. Consequently, the results lead to the hypothesis that the satisfaction of pharmacy students with this kind of training using high-fidelity simulation is very high and attractive, and can be recommended for other German universities. The integration of such vaccination training into the German pharmacy curriculum might be a future step.

Application of the Hydrodeoxygenation of Black Soldier Fly Larvae Lipids in Green Diesel Production

To produce green diesel from black soldier fly larvae (BSFL; Hermetia illucens), the maximization of lipids in production and hydrodeoxygenation (HDO) reactions was investigated. In this study, BSFL were fed 12 diets based on three different substrates (ground corn, food waste, and meat by-product). The proximate compositions of larvae were analyzed, and rearing time, production rate, and feeding mixture prices were also recorded. To maximize the lipid yield, the effects of growing temperature, drying method, and extraction temperature were investigated. The HDO reaction of BSFL oil with 1 wt % Pt/Al2O3 catalyst was carried out in a trickle bed reactor. The components of the lipids produced under optimal conditions and the components of lipids produced through the HDO reaction were compositionally analyzed. As a result of being fed ground corn, food waste, and meat by-products, it was confirmed that the diet with 30% ground corn and 70% meat by-product led to the highest lipid content in the BSFL. After considering the prices of the diets, we found that the most ideal feeding conditions that could be applied to actual insect farming were 70% food waste and 30% meat by-products. From the perspective of the rearing period, the most appropriate BSFL-rearing temperature was a medium temperature of 38 °C. After harvesting the BSFL, it was confirmed that the lipid yield improved when extracted at a temperature of 65–75 °C after drying using a microwave. The analysis results showed that the carbon distribution in hydrodeoxygenated BSFL oil offered an advantage when used as drop-in fuel, and this represents a promising future step for the HDO of BSFL lipids.

Reachability-Based Confidence-Aware Probabilistic Collision Detection in Highway Driving

Risk assessment is a crucial component of collision warning and avoidance systems for intelligent vehicles. Reachability-based formal approaches have been developed to ensure driving safety to accurately detect potential vehicle collisions. However, they suffer from over-conservatism, potentially resulting in false–positive risk events in complicated real-world applications. In this paper, we combine two reachability analysis techniques, a backward reachable set (BRS) and a stochastic forward reachable set (FRS), and propose an integrated probabilistic collision–detection framework for highway driving. Within this framework, we can first use a BRS to formally check whether a two-vehicle interaction is safe; otherwise, a prediction-based stochastic FRS is employed to estimate the collision probability at each future time step. Thus, the framework can not only identify non-risky events with guaranteed safety but also provide accurate collision risk estimation in safety-critical events. To construct the stochastic FRS, we develop a neural network-based acceleration model for surrounding vehicles and further incorporate a confidence-aware dynamic belief to improve the prediction accuracy. Extensive experiments were conducted to validate the performance of the acceleration prediction model based on naturalistic highway driving data. The efficiency and effectiveness of the framework with infused confidence beliefs were tested in both naturalistic and simulated highway scenarios. The proposed risk assessment framework is promising for real-world applications.

Study on Various Position Sensing Technologies in Tractor Hitch Application

Conventionally, In the tractors, hitch position is obtained using a contact-based position sensor, the sensor is attached to the rockshaft mechanically with gears or linkages. In this arrangement, there are incidences of failures when working with heavy implements, as rockshaft is under severe loading conditions which tends to induce cracks or failures in sensor mechanical linkages mounting area. This results in machine downtime and rockshaft repair costs to the customer. These mounting threads in hardened Rockshaft reduce material strength which impacts the lifting capacity. There is associated difficulty at assembly line and further during maintenance for linkage installation and adjustments. Also, over the period, linkages develop a play at joints due to the friction. In the present work, we are sensing the Hitch position with contactless sensor technology. A contactless Hall effect sensor is selected, and proto setup has been designed to measure the hitch position by integrating the sensor and target. The prototype consists of a fixed Sensor and a rotating metal (ferromagnetic) piece placed inside a 3D printed casing. The profile of metal piece is such that when it is rotated about the axis, the distance between Sensor and metal piece changes gradually. This causes different output voltage at different amount of rotation in metal piece. Results from manual bench testing show that the concept is working. The sensor voltage output varies according to the hitch position in the bench setup. A high voltage value when metal piece is nearer to magnet and low value when its farther. A future step is to automate this setup on a tractor to remove errors due to manual rotation. This would be an enabler to analyze the sensor behaviors like hysteresis, repeatability, linearity more accurately.

Development of a tool for assessing the performance of long-term care systems in relation to care transition: Transitional Care Assessment Tool in Long-Term Care (TCAT-LTC)

BackgroundImproving the quality and safety of care transitions is a priority in many countries. Carrying out performance measurements play a significant role in improving quality of decisions undertaken by different actors involved in reforms. Therefore, the main objective of this paper is to present the development of an evaluation tool for assessing the performance of long-term care systems in relation to care transition, namely the Transitional Care Assessment Tool in Long-Term Care (TCAT-LTC). This study is performed as part of a larger European TRANS-SENIOR project.MethodsThe development of the TCAT-LTC involved three steps. First, we developed a conceptual model based on Donabedian’s quality framework and literature review. Second, we carried out a thorough process of item pool generation using deductive (systematic literature review) and deductive-inductive methods (in-depth interviews) with experts in the field of long-term care. Third, we conducted preliminary validation of the tool by asking experts in research and practice to provide an opinion on a tool and to assess content validity. Future fourth step will involve a tool’s pilot with country experts from Germany, the Netherlands and Poland.ResultsBy applying methodological triangulation, we developed the TCAT-LTC, which consists of 2 themes, 12 categories and 63 items. Themes include organizational and financial aspects. Organizational aspects include categories such as communication, transfer of information, availability and coordination of resources, training and education of staff, education/support of the patient/informal caregiver, involvement of the patient/informal caregiver, telemedicine and e-Health, and social care. Financial aspects include categories such as primary care, hospital, and long-term care. We also present the instructions on the application of the TCAT-LTC.ConclusionsIn this paper, we presented the development of the TCAT-LTC evaluation tool for assessing the performance of long-term care systems in relation to care transition. The TCAT-LTC is the first tool to assess the performance of long-term care systems in relation to care transition. Assessments can be carried out at the national and international level and enable to monitor, evaluate, and compare performance of the long-term care systems in relation to care transition within and across countries.

Optimized resource usage with hybrid auto-scaling system for knative serverless edge computing

In the most popular serverless platform - Knative, dynamic resource allocation is implemented using horizontal auto-scaling algorithms to create or delete service instances based on different monitored metrics. However, the assigned resources for each instance are fixed. Vertical scaling up or down assigned resources per instance is required to avoid over-provisioning resources which are limited at the edge. Hybrid (horizontal and vertical) auto-scaling solutions proposed by existing works have several limitations. These solutions are optimized for separated services and get degraded performance when applied in a normal environment with multiple concurrent services. Further, most methods make significant changes to the original Knative platform, and have not been considered to be adopted since then. In this article, instead of Knative modification, we developed separated Kubernetes operators and custom resources (CRs) that can assist the Knative auto-scaler with optimal hybrid auto-scaling configurations based on traffic prediction. First, we characterize each service with a profile of different assigned resource levels pairing with their optimal target Knative’s horizontal scaling request concurrency. Then, based on these profiles, we calculate the best-assigned resources level, target concurrency level, and the number of required instances corresponding to each future time step’s predicted traffic. Finally, these configurations are applied to Knative’s default auto-scaler and services’ CR. Experiments done on our testbed compared our solution with a Knative hybrid auto-scaler solution that does not consider the service’s target request concurrency, and the default Knative horizontal auto-scaler. The results show our solution improvements up to 14% and 20% in terms of resource usage, respectively.

Older Persons' and Health Care Professionals' Design Choices When Co-Designing a Medication Plan Aiming to Promote Patient Safety: Case Study.

Harm from medications is a major patient safety challenge among older persons. Adverse drug events tend to arise when prescribing or evaluating medications; therefore, interventions targeting these may promote patient safety. Guidelines highlight the value of a joint plan for continued treatment. If such a plan includes medications, a medication plan promoting patient safety is advised. There is growing evidence for the benefits of including patients and health care professionals in initiatives for improving health care products and services through co-design. This study aimed to identify participants' needs and requirements for a medication plan and explore their reasoning for different design choices. Using a case study design, we collected and analyzed qualitative and quantitative data and compared them side by side. We explored the needs and requirements for a medication plan expressed by 14 participants (older persons, nurses, and physicians) during a co-design initiative in a regional health system in Sweden. We performed a directed content analysis of qualitative data gathered from co-design sessions and interviews. Descriptive statistics were used to analyze the quantitative data from survey answers. A medication plan must provide an added everyday value related to safety, effort, and engagement. The physicians addressed challenges in setting aside time to apply a medication plan, whereas the older persons raised the potential for increased patient involvement. According to the participants, a medication plan needs to support communication, continuity, and interaction. The nurses specifically addressed the need for a plan that was easy to gain an overview of. Important function requirements included providing instant access, automation, and attention. Content requirements included providing detailed information about the medication treatment. Having the plan linked to the medication list and instantly obtainable information was also requested. After discussing the needs and requirements for a medication plan, the participants agreed on an iteratively developed medication plan prototype linked to the medication list within the existing electronic health record. According to the participants, the medication plan prototype may promote patient safety and enable patient engagement, but concerns were raised about its use in daily clinical practice. The last step in the co-design framework is testing the intervention to explore how it works and connects with users. Therefore, testing the medication plan prototype in clinical practice would be a future step.

Maritime logistics optimisation for predictive maintenance at offshore wind farms

For offshore wind farms, a move from a preventive and corrective maintenance regime to a predictive maintenance regime requires new methods for modelling approaches for maritime logistics planning. This paper presents an overview of the maritime logistics planning problem for a predictive maintenance regime and introduces the current state-of-the-art for operational research in the field of operation and maintenance at offshore wind farms. Findings are that a combination of the vessel resource scheduling problem for operation and maintenance at offshore wind farms with predictive analysis and digital twins is a promising future research step. A framework for a decision support tool is presented that will help bridge the gap, both with respect to the academic path, and the gap between academic research and industry.

Geometric inverse estimation for the inner wall of a furnace under transient heat conduction based on dual reciprocity boundary element method

A boundary shape estimation problem for the furnace’s inner surface is solved using the dual reciprocity boundary element method (DRBEM) and the conjugate gradient method (CGM) under transient heat conduction. The DRBEM is utilized to eliminate the drawbacks of traditional numerical methods and classical boundary element method of discretizing the entire computational domain, with only boundary discretization. The inversion results are obtained by applying the CGM to minimize the objective function, in which the sensitivity coefficients are calculated with the complex variable derivation method (CVDM), making the calculation precise and independent of step size. To verify the accuracy of the DRBEM in solving the transient heat conduction problem, the influencing factors including radial-basis function, the number of internal collocation points, and time step size are investigated. The influences of measurement time interval, future time step, initial guess, measurement error, and the number and position of measurement points on the inversion results are analyzed. Meanwhile, the effectiveness of the proposed approach is tested by numerical examples, and the inversion results show that it is stable, accurate, and efficient for identifying different and complicated unknown boundary shapes of the furnace.

Deep Probabilistic Koopman: Long-term time-series forecasting under periodic uncertainties

This paper introduces general mathematical techniques for stable long-term forecasts with calibrated uncertainty measures. For most time series models, the difficulty of obtaining accurate probabilistic future time step predictions increases with the prediction horizon. We propose a surprisingly simple class of models that characterizes time-varying distributions and enables reasonably accurate predictions thousands of time steps into the future. This technique, called Deep Probabilistic Koopman (DPK), is based on recent advances in linear Koopman operator theory and does not require time stepping for future time predictions. We demonstrate the long-term forecasting performance of these models on a diversity of domains, including electricity demand forecasting, atmospheric chemistry, and neuroscience. Our domain-agnostic technique outperforms all 177 domain-specific competitors in the most recent Global Energy Forecasting Competition for electricity demand modelling.

Deep learning-based pedestrian trajectory prediction and risk assessment at signalized intersections using trajectory data captured through roadside LiDAR

In recent years, rapid advancements in the Autonomous Vehicles (AVs) industry have greatly motivated the research and development in pedestrian trajectory prediction and risk assessment. One of the critical requirements for AVs is to predict the future trajectories of pedestrians and provide collision warnings in an accurate and prompt manner. Nevertheless, accurate prediction of pedestrian trajectories remains a technical challenge, mainly caused by the heterogeneity of pedestrian crossing behavior and uncertainties in vehicle-pedestrian interactions. This paper proposes a deep learning-based method for pedestrian trajectory prediction and risk assessment, using trajectory data extracted from roadside LiDAR data and corresponding signal phasing information at MLK and Georgia Avenue in Chattanooga, TN. Meanwhile, a set of criteria referred to as the risk factor is established to quantitatively evaluate the risk of the pedestrian crossing behavior, which also serves as a learnable feature. A Long Short-Term Memory (LSTM) network is proposed, which takes the following data as the input: the pedestrian trajectory data, signal phasing data, and risk factors from the past 10 steps. Meanwhile, the network predicts the pedestrian trajectory and risk factor at the future time step. In the experimental study, the root-mean-square errors between the predicted and ground truth x and y coordinates are 0.225 meters and 0.377 meters, respectively, and the F1 score value for the risk factor is 99.6%, demonstrating the efficacy of the proposed LSTM-based methodology on pedestrian trajectory prediction and risk assessment.

Fractionating distraction: How past- and future-relevant distractors influence integrated decisions.

Many everyday tasks require us to integrate information from multiple steps to make a decision. Dominant accounts of flexible cognition suggest that we are able to navigate such complex tasks by attending to each step in turn, yet few studies measure how we direct our attention to immediate and future task steps. Here, we used a two-step task to test whether participants are sensitive to information that is currently irrelevant but will be relevant in a future task step. Participants viewed two displays in sequence, each containing two superimposed moving dot clouds of different colors. Participants attended to one cued target color in each display and reported the average direction of the two target dot clouds. In a subset of trials, we presented a "decoy" distractor: the second target color appeared as the distractor in the first display. We regressed behavioral responses on the dot clouds' motion directions to track how this future-relevant "decoy" distractor influenced participants' reporting of the average target direction. We compared the influence of decoy distractors to never-relevant, recently relevant, and globally relevant distractor baselines. Across four experiments, we found that responses reflected what was immediately relevant, as well as the broader historical relevance of the distractors. However, relevance for a future task step did not reliably influence attention. We propose that attention in multistep tasks is shaped by what has been relevant in the current setting, and by the immediate demands of each task step. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Abstract 3661: Extracellular vesicle derived from neural stem cells as potential vehicles of genetic information to modulate glioblastoma biology

Abstract Glioblastoma (GBM) is a deadly brain tumor and its high aggressiveness is partly due to a subpopulation of tumor stem cells known as glioblastoma stem-like cells (GSCs). GBM establishes crosstalk with non-tumoral cells, mediated mainly by extracellular vesicles (EVs) and the interaction between GSCs and neural stem cells (NSCs) from the subventricular zone may influence GBM biology. Literature data suggests that cellular communication is an important process for GSCs stemness maintenance, and some authors consider NSCs as a possible cell of origin for GSCs. However, the specific effects that neural stem cells-derived extracellular vesicles (NSCs-EVs) promote in GSCs are still uncertain. Previous data from our group demonstrated specific groups of microRNAs (hsa-miR -137, -216a-5p, -216b-5p, 217), enriched in NSCs-EVs in contrast to intracellular compartments, suggesting these molecules are being produced for secretion. Our study aims to evaluate the role of the miRs contained in NSCs-EVs in GSCs biology. Human induced pluripotent stem cells (hiPSC) were differentiated into NSCs. NSCs were characterized by western blotting and immunofluorescence by the expression of neural commitment markers, such as nestin and β-III-tubulin. NSCs-EVs were isolated from NSCs’ culture media with a commercial kit and analyzed by NTA (nanosight) for EVs size profiling and western blotting to detect specific markers, such as Alix, CD63, and HSP90. In a first approach, we analyzed the effects played by those miRs through predictive online tools. We identified target genes related to stemness (NANOG, SOX2, KLF4), oncogenesis (KRAS, CDC42, CDK6) and metabolism (CAMK2A), besides the Wnt and Notch signaling pathways. The online predictions also identified transcriptional factors related to vasculogenesis and self-renewal, key processes for GSCs activity. We will expose GSCs to NSCs-EVs and evaluate the expression of neural and stemness markers. In a future step, the GSCs previously exposed to NSCs-EVs will be treated with temozolomide, and viability assays will be performed. Additionally, we intend to modulate the four miRs identified and evaluate if they act as effector molecules of NSCs-EVs. Our data corroborate with our initial hypothesis that NSCs-EVs may act as important signalling modes in GSCs activity and consequently in GBM biology. Supported by FAPESP:22/08106-1. Citation Format: João Pedro Alves de Araújo, Lilian Cruz, Maria Clara da Silva Souza, Mariana Brandão Prado, Rodrigo Nunes Alves, Bárbara Paranhos Coelho, Samuel Ribeiro Soares, Camila Félix Fernandes, Marilene Hohmuth Lopes. Extracellular vesicle derived from neural stem cells as potential vehicles of genetic information to modulate glioblastoma biology. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3661.

Simultaneous estimation of time-varying boundary heat flux and temperature-dependent thermophysical and optical parameters of grey translucent medium

A hybrid optimization method, which combines unscented Kalman smoothing (UKS) technique and sequential quadratic programming (SQP) algorithm, is proposed to estimate the time-dependent boundary heat flux and temperature-dependent optical and thermophysical parameters (such as absorption coefficient, refractive index and thermal conductivity) of grey translucent medium simultaneously. The transient coupled radiative and conductive problem in grey translucent medium is solved by finite volume method (FVM) in forward model. In inverse problem, the measured signals consist of the temperature and radiative intensity on the boundaries of the medium. The effects of model parameters (such as the number of future time step and process noise covariance) and different incident heat fluxes on the reconstructed results are studied by the UKS-SQP technique. The results indicate that the proposed hybrid optimization technique is accurate and effective for determining the time-varying boundary heat flux, temperature-dependent optical and thermophysical parameters of grey translucent medium even with the measurement error.

Emerging trends and focus on the link between gut microbiota and type 1 diabetes: A bibliometric and visualization analysis.

To conduct the first thorough bibliometric analysis to evaluate and quantify global research regarding to the gut microbiota and type 1 diabetes (T1D). A literature search for research studies on gut microbiota and T1D was conducted using the Web of Science Core Collection (WoSCC) database on 24 September 2022. VOSviewer software and the packages Bibliometrix R and ggplot used in RStudio were applied to perform the bibliometric and visualization analysis. A total of 639 publications was extracted using the terms "gut microbiota" and "type 1 diabetes" (and their synonyms in MeSH). Ultimately, 324 articles were included in the bibliometric analysis. The United States and European countries are the main contributors to this field, and the top 10 most influential institutions are all based in the United States, Finland and Denmark. The three most influential researchers in this field are Li Wen, Jorma Ilonen and Mikael Knip. Historical direct citation analysis showed the evolution of the most cited papers in the field of T1D and gut microbiota. Clustering analysis defined seven clusters, covering the current main topics in both basic and clinical research on T1D and gut microbiota. The most commonly found high-frequency keywords in the period from 2018 to 2021 were "metagenomics," "neutrophils" and "machine learning." The application of multi-omics and machine learning approaches will be a necessary future step for better understanding gut microbiota in T1D. Finally, the future outlook for customized therapy toward reshaping gut microbiota of T1D patients remains promising.

Inverse Estimation of Transient Heat Flux Using Sequential Function Specification Method

Knowledge of surface heat flux is of great importance to numerous engineering fields. In this study, the sequential function specification method (SFSM) was employed to estimate the transient heat flux on the surface of semitransparent medium. The coupled radiation and conduction heat transfer was solved by discrete ordinate method combined with finite volume method. The temperature was used as input in estimation task which was solved by SFSM. Different types of heat fluxes were taken into account to test performance of the present technique. Retrieval results illustrated that the transient heat flux can be accurately determined using the SFSM. It was proved to be effective for different semitransparent media. For the heat flux with drastic changes, the SFSM showed some limitations. Moreover, the influences of future time step and measurement error on determination results were investigated. The SFSM was applicable even when the random deviation was increased to 1.5. The future time step of SFSM is suggested to be 3 for the estimation of transient heat flux on semitransparent medium surface.

Spatiotemporal Prediction Method for Vorticity of Three-dimensional Near-wall Cylindrical Flow Based on Convolutional Recurrent Neural Network

Due to the existence of a wall surface, the flow around the near-wall cylinder is a very complex unsteady dynamic behavior. However, this kind of feature extraction and fast prediction method of complex flow is difficult to solve in engineering. Therefore, a data-driven model based on Convolutional Recurrent Neural Network (CRNN) is proposed to solve the feature extraction and fast prediction of complex flow, and to predict the vorticity of the near wake region of a three-dimensional near-wall cylinder with G/D=1. CRNN model can analyze the relationship between Spatiotemporal evolution. The three-dimensional flow around the near-wall cylinder with Reynolds number 1500 meshed, and the vorticity was calculated after grid calculation to establish the Spatiotemporal evolution data set. For machine learning problems with high dimensions and large amounts of data, the Adaptive Moment (Adam) estimation method served as an effectively optimize. According to the Spatiotemporal prediction results of vorticity, the prediction results of the model fit well with the numerical simulation results, and the vorticity of the future time step is successfully predicted on the space and time series, that is, the evolutionary relationship of the Spatiotemporal series of vorticity near the wake of the near-wall cylinder is established. After the above model is established, only the initial multi-time vorticity data need to be obtained. The model can serve as predicting the vorticity field in the near wake region of the near-wall cylinder in the future.