Short-term train arrival delay prediction:a data-driven approach

Purpose-To optimize train operations,dispatchers currently rely on experience for quick adjustments when delays occur.However,delay predictions often involve imprecise shifts based on known delay times.Real-time and accurate train delay predictions,facilitated by data-driven neural network models,can significantly reduce dispatcher stress and improve adjustment plans.Leveraging current train operation data,these models enable swift and precise predictions,addressing challenges posed by train delays in high-speed rail networks during unforeseen events.Design/methodology/approach-This paper proposes CBLA-net,a neural network architecture for predicting late arrival times.It combines CNN,Bi-LSTM,and attention mechanisms to extract features,handle time series data,and enhance information utilization.Trained on operational data from the Beijing-Tianjin line,it predicts the late arrival time of a target train at the next station using multidimensional input data from the target and preceding trains.Findings-This study evaluates our model’s predictive performance using two data approaches:one considering full data and another focusing only on late arrivals.Results show precise and rapid predictions.Training with full data achieves aMAEof approximately 0.54 minutes and a RMSEof 0.65 minutes,surpassing the model trained solely on delay data(MAE:is about 1.02 min,RMSE:is about 1.52 min).Despite superior overall performance with full data,the model excels at predicting delays exceeding 15 minutes when trained exclusively on late arrivals.For enhanced adaptability to real-world train operations,training with full data is recommended.Originality/value-This paper introduces a novel neural network model,CBLA-net,for predicting train delay times.It innovatively compares and analyzes the model’s performance using both full data and delay data formats.Additionally,the evaluation of the network’s predictive capabilities considers different scenarios,providing a comprehensive demonstration of the model’s predictive performance.

Separation density prediction of geldart A^(-)dense medium in gas-solid fluidized bed coal beneficiators

Gas-solid Fluidized Bed Coal Beneficiator(GFBCB)process is a crucial dry coal beneficiation fluidization technology.The work employs the GFBCB process alongside a novel Geldart A^(-)dense medium,consisting of Geldart A magnetite particles and Geldart C ultrafine coal,to separate small-size separated objects in the GFBCB.The effects of various operational conditions,including the volume fraction of ultrafine coal,the gas velocity,the separated objects size,and the separation time,were investigated on the GFBCB's separation performance.The results indicated that the probable error for 6∼3 mm separated objects could be controlled within 0.10 g/cm^(3).Compared to the traditional Geldart B/D dense medium,the Geldart A/A^(-)dense medium exhibited better size-dependent separation performance with an overall probable error 0.04∼0.12 g/cm^(3).Moreover,it achieved a similar separation accuracy to the Geldart B/D dense medium fluidized bed with different external energy for the small-size object beneficiation.The work furthermore validated a separation density prediction model based on theoretical derivation,available for both Geldart B/D dense medium and Geldart A/A^(-)dense medium at different operational conditions.

Prediction of Outcomes in Mini-Basketball Training Program for Preschool Children with Autism Using Machine Learning Models

In recent years evidence has emerged suggesting that Mini-basketball training program(MBTP)can be an effec-tive intervention method to improve social communication(SC)impairments and restricted and repetitive beha-viors(RRBs)in preschool children suffering from autism spectrum disorder(ASD).However,there is a considerable degree if interindividual variability concerning these social outcomes and thus not all preschool chil-dren with ASD profit from a MBTP intervention to the same extent.In order to make more accurate predictions which preschool children with ASD can benefit from an MBTP intervention or which preschool children with ASD need additional interventions to achieve behavioral improvements,further research is required.This study aimed to investigate which individual factors of preschool children with ASD can predict MBTP intervention out-comes concerning SC impairments and RRBs.Then,test the performance of machine learning models in predict-ing intervention outcomes based on these factors.Participants were 26 preschool children with ASD who enrolled in a quasi-experiment and received MBTP intervention.Baseline demographic variables(e.g.,age,body,mass index[BMI]),indicators of physicalfitness(e.g.,handgrip strength,balance performance),performance in execu-tive function,severity of ASD symptoms,level of SC impairments,and severity of RRBs were obtained to predict treatment outcomes after MBTP intervention.Machine learning models were established based on support vector machine algorithm were implemented.For comparison,we also employed multiple linear regression models in statistics.Ourfindings suggest that in preschool children with ASD symptomatic severity(r=0.712,p<0.001)and baseline SC impairments(r=0.713,p<0.001)are predictors for intervention outcomes of SC impair-ments.Furthermore,BMI(r=-0.430,p=0.028),symptomatic severity(r=0.656,p<0.001),baseline SC impair-ments(r=0.504,p=0.009)and baseline RRBs(r=0.647,p<0.001)can predict intervention outcomes of RRBs.Statistical models predicted 59.6%of variance in post-treatment SC impairments(MSE=0.455,RMSE=0.675,R2=0.596)and 58.9%of variance in post-treatment RRBs(MSE=0.464,RMSE=0.681,R2=0.589).Machine learning models predicted 83%of variance in post-treatment SC impairments(MSE=0.188,RMSE=0.434,R2=0.83)and 85.9%of variance in post-treatment RRBs(MSE=0.051,RMSE=0.226,R2=0.859),which were better than statistical models.Ourfindings suggest that baseline characteristics such as symptomatic severity of 144 IJMHP,2022,vol.24,no.2 ASD symptoms and SC impairments are important predictors determining MBTP intervention-induced improvements concerning SC impairments and RBBs.Furthermore,the current study revealed that machine learning models can successfully be applied to predict the MBTP intervention-related outcomes in preschool chil-dren with ASD,and performed better than statistical models.Ourfindings can help to inform which preschool children with ASD are most likely to benefit from an MBTP intervention,and they might provide a reference for the development of personalized intervention programs for preschool children with ASD.

Disturbance rejection tube model predictive levitation control of maglev trains

Magnetic levitation control technology plays a significant role in maglev trains.Designing a controller for the levitation system is challenging due to the strong nonlinearity,open-loop instability,and the need for fast response and security.In this paper,we propose a Disturbance-Observe-based Tube Model Predictive Levitation Control(DO-TMPLC)scheme combined with a feedback linearization strategy for the levitation system.The proposed strategy incorporates state constraints and control input constraints,i.e.,the air gap,the vertical velocity,and the current applied to the coil.A feedback linearization strategy is used to cancel the nonlinearity of the tracking error system.Then,a disturbance observer is implemented to actively compensate for disturbances while a TMPLC controller is employed to alleviate the remaining disturbances.Furthermore,we analyze the recursive feasibility and input-to-state stability of the closed-loop system.The simulation results indicate the efficacy of the proposed control strategy.

Step-by-Step Numerical Prediction of Aerodynamic Noise Generated by High Speed Trains

In this paper,the unsteady flow around a high-speed train is numerically simulated by detached eddy simulation method(DES),and the far-field noise is predicted using the Ffowcs Williams-Hawkings(FW-H)acoustic model.The reliability of the numerical calculation is verified by wind tunnel experiments.The superposition relationship between the far-field radiated noise of the local aerodynamic noise sources of the high-speed train and the whole noise source is analyzed.Since the aerodynamic noise of high-speed trains is derived from its different components,a stepwise calculation method is proposed to predict the aerodynamic noise of high-speed trains.The results show that the local noise sources of high-speed trains and the whole noise source conform to the principle of sound source energy superposition.Using the head,middle and tail cars of the high-speed train as noise sources,different numerical models are established to obtain the far-field radiated noise of each aerodynamic noise source.The far-field total noise of high-speed trains is predicted using sound source superposition.A step-by-step calculation of each local aerodynamic noise source is used to obtain the superimposed value of the far-field noise.This is consistent with the far-field noise of the whole train model’s aerodynamic noise.The averaged sound pressure level of the far-field longitudinal noise measurement points differs by 1.92 dBA.The step-by-step numerical prediction method of aerodynamic noise of high-speed trains can provide a reference for the numerical prediction of aerodynamic noise generated by long marshalling high-speed trains.

High-Speed Railway Train Timetable Conflict Prediction Based on Fuzzy Temporal Knowledge Reasoning

Trains are prone to delays and deviations from train operation plans during their operation because of internal or external disturbances. Delays may develop into operational conflicts between adjacent trains as a result of delay propagation, which may disturb the arrangement of the train operation plan and threaten the operational safety of trains. Therefore, reliable conflict prediction results can be valuable references for dispatchers in making more efficient train operation adjustments when conflicts occur. In contrast to the traditional approach to conflict prediction that involves introducing random disturbances, this study addresses the issue of the fuzzification of time intervals in a train timetable based on historical statistics and the modeling of a high-speed railway train timetable based on the concept of a timed Petri net. To measure conflict prediction results more comprehensively, we divided conflicts into potential conflicts and certain conflicts and defined the judgment conditions for both. Two evaluation indexes, one for the deviation of a single train and one for the possibility of conflicts between adjacent train operations, were developed using a formalized computation method. Based on the temporal fuzzy reasoning method, with some adjustment, a new conflict prediction method is proposed, and the results of a simulation example for two scenarios are presented. The results prove that conflict prediction after fuzzy processing of the time intervals of a train timetable is more reliable and practical and can provide helpful information for use in train operation adjustment, train timetable improvement, and other purposes.

Genomic prediction using composite training sets is an effective method for exploiting germplasm conserved in rice gene banks

Germplasm conserved in gene banks is underutilized,owing mainly to the cost of characterization.Genomic prediction can be applied to predict the genetic merit of germplasm.Germplasm utilization could be greatly accelerated if prediction accuracy were sufficiently high with a training population of practical size.Large-scale resequencing projects in rice have generated high quality genome-wide variation information for many diverse accessions,making it possible to investigate the potential of genomic prediction in rice germplasm management and exploitation.We phenotyped six traits in nearly 2000 indica(XI)and japonica(GJ)accessions from the Rice 3K project and investigated different scenarios for forming training populations.A composite core training set was considered in two levels which targets used for prediction of subpopulations within subspecies or prediction across subspecies.Composite training sets incorporating 400 or 200 accessions from either subpopulation of XI or GJ showed satisfactory prediction accuracy.A composite training set of 600 XI and GJ accessions showed sufficiently high prediction accuracy for both XI and GJ subspecies.Comparable or even higher prediction accuracy was observed for the composite training set than for the corresponding homogeneous training sets comprising accessions only of specific subpopulations of XI or GJ(within-subspecies level)or pure XI or GJ accessions(across-subspecies level)that were included in the composite training set.Validation using an independent population of 281 rice cultivars supported the predictive ability of the composite training set.Reliability,which reflects the robustness of a training set,was markedly higher for the composite training set than for the corresponding homogeneous training sets.A core training set formed from diverse accessions could accurately predict the genetic merit of rice germplasm.

Wear characteristics and prediction of wheel profiles in high-speed trains

Wheel/rail relationship is a fundamental problem of railway system. Wear of wheel profiles has great effect on vehicle performance. Thus, it is important not just for the analysis of wear characteristics but for its prediction. Actual wheel profiles of the high-speed trains on service were measured in the high-speed line and the wear characteristics were analyzed which came to the following results. The wear location was centralized from-15 mm to 25 mm. The maximum wear value appeared at the area of 5 mm from tread center far from wheel flange and it was less than 1.5 mm. Then, wheel wear was fitted to get the polynomial functions on different locations and operation mileages. A binary numerical prediction model was raised to predict wheel wear. The prediction model was proved by vehicle system dynamics and wheel/rail contact geometry. The results show that the prediction model can reflect wear characteristics of measured profiles and vehicle performances.

Impact of Subsurface Entrainment on ENSO Prediction: 1997-98 El Nio

Twenty-one-year hindcasts of sea surface temperature (SST) anomalies in the tropical Pacific were performed to validate the influence of ocean subsurface entrainment on SST prediction.A new hybrid coupled model was used that considered the entrainment of subsurface temperature anomalies into the sea surface.The results showed that predictions were improved significantly in the new coupled model.The predictive correlation skill increased by about 0.2 at a lead time of 9 months,and the root-mean-square (RMS) errors were decreased by nearly 0.2°C in general.A detailed analysis of the 1997-98 El Nio hindcast showed that the new model was able to predict the onset,peak (both time and amplitude),and decay of the 1997-98 strong El Nio event up to a lead time of one year,factors that are not represented well by many other forecast systems.This implies,in terms of prediction,that subsurface anomalies and their impact on the SST are one of the controlling factors in ENSO cycles.Improving the presentation of such effects in models would increase the forecast skill.

Prediction of Protein OmpH in Structure of C47-8 Pasteurella multocida

[Objective] This study aimed to predict the structure of protein OmpH from Pasteurella multocida C47-8 （PmC47-8） strain of yak. [Method] Online BLAST, signal peptide prediction, secondary structure prediction and protein characteristics of sequencing result of gene OmpH from PmC47-8 strain were analyzed. [Result] The similarities of gene OmpH from PmC47-8 with the published 81 OmpH genes were between 84% and 99%; a signal peptide was found with the cleavage sites between 20 and 21 in the polypeptide; secondary structure prediction showed that folding structure accounted for 49.8% and loop structure for 50.2%; it predicted that there were 7 O-glycosylation sites in OmpH protein with the amino acid residual sites of 2, 45, 48, 330, 716, 721, 723, respectively, and 2 N-glycosylation sites with the amino acid residual sites of 15 and 35. [Conclusion] This study lays the foundation for the study on the immunity of OmpH gene from yak.

Remaining Useful Life Prediction With Partial Sensor Malfunctions Using Deep Adversarial Networks

In recent years,intelligent data-driven prognostic methods have been successfully developed,and good machinery health assessment performance has been achieved through explorations of data from multiple sensors.However,existing datafusion prognostic approaches generally rely on the data availability of all sensors,and are vulnerable to potential sensor malfunctions,which are likely to occur in real industries especially for machines in harsh operating environments.In this paper,a deep learning-based remaining useful life(RUL)prediction method is proposed to address the sensor malfunction problem.A global feature extraction scheme is adopted to fully exploit information of different sensors.Adversarial learning is further introduced to extract generalized sensor-invariant features.Through explorations of both global and shared features,promising and robust RUL prediction performance can be achieved by the proposed method in the testing scenarios with sensor malfunctions.The experimental results suggest the proposed approach is well suited for real industrial applications.

Models to predict injury, physical fitness failure and attrition in recruit training: a retrospective cohort study

Background:Attrition rate in new army recruits is higher than in incumbent troops.In the current study,we identified the risk factors for attrition due to injuries and physical fitness failure in recruit training.A variety of predictive models were attempted.Methods:This retrospective cohort included 19,769 Army soldiers of the Australian Defence Force receiving recruit training during a period from 2006 to 2011.Among them,7692 reserve soldiers received a 28-day training course,and the remaining 12,077 full-time soldiers received an 80-day training course.Retrieved data included anthropometric measures,course-specific variables,injury,and physical fitness failure.Multivariate regression was used to develop a variety of models to predict the rate of attrition due to injuries and physical fitness failure.The area under the receiver operating characteristic curve was used to compare the performance of the models.Results:In the overall analysis that included both the 28-day and 80-day courses,the incidence of injury of any type was 27.8%.The 80-day course had a higher rate of injury if calculated per course(34.3%vs.17.6%in the 28-day course),but lower number of injuries per person-year(1.56 vs.2.29).Fitness test failure rate was significantly higher in the 28-day course(30.0%vs.12.1%).The overall attrition rate was 5.2%and 5.0%in the 28-day and 80-day courses,respectively.Stress fracture was common in the 80-day course(n=44)and rare in the 28-day course(n=1).The areas under the receiver operating characteristic curves for the course-specific predictive models were relatively low(ranging from 0.51 to 0.69),consistent with"failed"to"poor"predictive accuracy.The course-combined models performed somewhat better than the course-specific models,with two models having AUC of 0.70 and 0.78,which are considered"fair"predictive accuracy.Conclusion:Attrition rate was similar between 28-day and 80-day courses.In comparison to the 80-day full course,the 28-day course had a lower rate of injury but a higher number of injuries per person-year and of fitness test failure.These findings suggest fitness level at the commencement of training is a critically important factor to consider when designing the course curriculum,particularly short courses.

Prediction of vibrations from underground trains on Beijing metro line 15

The impact of vibrations due to underground trains on Beijing metro line 15 on sensitive equipment in the Institute of Microelectronics of Tsinghua University was discussed to propose a viable solution to mitigate the vibrations.Using the state-of-the-art three-dimensional coupled periodic finite element-boundary element(FE-BE) method,the dynamic track-tunnel-soil interaction model for metro line 15 was used to predict vibrations in the free field at a train speed of 80 km/h.Three types of tracks(direct fixation fasteners,floating slab track and floating ladder track) on the Beijing metro network were considered in the model. For each track,the acceleration response in the free field was obtained.The numerical results show that the influence of vibrations from underground trains on sensitive equipment depends on the track types.At frequencies above 10 Hz,the floating slab track with a natural frequency of 7 Hz can be effective to attenuate the vibrations.

Uni-LSDPM:基于预训练的统一在线学习会话退出预测模型

为了辅助学习者维持在线学习的连贯性以引导最优学习路径的执行,智能辅导系统(intelligent tutoring system,ITS)需要及时发现学习者退出学习的倾向,在合适的时间采取相应的干预措施,因此,在线学习会话退出预测研究十分必要.然而,与传统的课程辍学相比,会话退出发生的频率更高,单次学习时长更短,故需要在有限的行为数据中对学习会话退出状态进行准确预测.因此,学习行为的碎片性和预测结果的即时性、准确性是学习会话退出预测任务的挑战和难点.针对会话退出预测任务,提出了一种基于预训练-微调的统一在线学习会话退出预测模型(unified online learning session dropout prediction model,Uni-LSDPM).该模型采用多层Transformer结构,分为预训练阶段和微调阶段.在预训练阶段,使用双向注意机制对学习者连续行为交互特征序列的特征表示进行学习.在微调阶段,应用序列到序列(sequenceto-sequence,Seq2Seq)的注意力机制对学习者连续行为交互特征序列与退出状态联合序列进行学习.基于EdNet公共数据集对模型进行预训练和微调,通过消融实验以获得最佳预测效果,并基于多个数据集进行了对比测试实验.实验结果表明,Uni-LSDPM在AUC和ACC方面优于现有的模型,并证明该模型具有一定的鲁棒性和扩展性.

Assessing the Efficacy of Improved Learning in Hourly Global Irradiance Prediction

Increasing global energy consumption has become an urgent problem as natural energy sources such as oil,gas,and uranium are rapidly running out.Research into renewable energy sources such as solar energy is being pursued to counter this.Solar energy is one of the most promising renewable energy sources,as it has the potential to meet the world’s energy needs indefinitely.This study aims to develop and evaluate artificial intelligence(AI)models for predicting hourly global irradiation.The hyperparameters were optimized using the Broyden-FletcherGoldfarb-Shanno(BFGS)quasi-Newton training algorithm and STATISTICA software.Data from two stations in Algeria with different climatic zones were used to develop the model.Various error measurements were used to determine the accuracy of the prediction models,including the correlation coefficient,the mean absolute error,and the root mean square error(RMSE).The optimal support vector machine(SVM)model showed exceptional efficiency during the training phase,with a high correlation coefficient(R=0.99)and a low mean absolute error(MAE=26.5741 Wh/m^(2)),as well as an RMSE of 38.7045 Wh/m^(2) across all phases.Overall,this study highlights the importance of accurate prediction models in the renewable energy,which can contribute to better energy management and planning.

基于ATT-CNN-BiLSTM的虚拟编组列车时空轨迹预测

保障虚拟编组平稳追踪运行的关键问题是实现对列车运行状态的精准预测。针对列车运行过程多变的特点,提出基于融合注意力机制的卷积双向长短期记忆神经网络(ATT-CNN-BiLSTM)的时空轨迹预测方法。针对列车历史运行数据中非正常运行场景稀少产生的数据非均衡问题,利用卷积神经网络和双向长短期记忆网络提取列车运行数据维度之间的特征关联,并增加注意力机制提升泛化能力。同时引入运行时验证方法在线监控预测结果,降低由预测错误造成的行车风险。以成都地铁8号线真实数据为例进行实验,设计5种评价指标,通过基线模型与消融实验对所提ATT-CNN-BiLSTM进行评价,该模型对于异常场景的预测误差至少减小9.626%。

Real-time crash prediction on freeways using data mining and emerging techniques

Recent advances in intelligent transportation system allow traffic safety studies to extend from historic data-based analyses to real-time applications. The study presents a new method to predict crash likelihood with traffic data collected by discrete loop detectors as well as the web-crawl weather data. Matched case-control method and support vector machines （SVMs） technique were employed to identify the risk status. The adaptive synthetic over-sampling technique was applied to solve the imbalanced dataset issues. Random forest technique was applied to select the contributing factors and avoid the over-fitting issues. The results indicate that the SVMs classifier could successfully classify 76.32% of the crashes on the test dataset and 87.52% of the crashes on the overall dataset, which were relatively satisfactory compared with the results of the previous studies. Compared with the SVMs classifier without the data, the SVMs classifier with the web-crawl weather data increased the crash prediction accuracy by 1.32% and decreased the false alarm rate by 1.72%, showing the potential value of the massive web weather data. Mean impact value method was employed to evaluate the variable effects, and the results are identical with the results of most of previous studies. The emerging technique based on the discrete traffic data and web weather data proves to be more applicable on real- time safety management on freeways.

Yield strength prediction of rolled Al-(1.44-12.40)Si-0.7Mg alloy sheets under T4 condition

The effects of Si content on the microstructure and yield strength of Al-(1.44-12.40)Si-0.7 Mg(wt.%)alloy sheets under the T4 condition were systematically studied via laser scanning confocal microscopy(LSCM),DSC,TEM and tensile tests.The results show that the recrystallization grain of the alloy sheets becomes more refined with an increase in Si content.When the Si content increases from 1.44 to 12.4 wt.%,the grain size of the alloy sheets decreases from approximately 47 to 10μm.Further,with an increase in Si content,the volume fraction of the GP zones in the matrix increases slightly.Based on the existing model,a yield strength model for alloy sheets was proposed.The predicted results are in good agreement with the actual experimental results and reveal the strengthening mechanisms of the Al-(1.44-12.40)Si-0.7 Mg alloy sheets under the T4 condition and how they are influenced by the Si content.

Bi-variable damage model for fatigue life prediction of metal components

Based on the theory of continuum damage mechanics,a bi-variable damage mechanics model is developed,which,according to thermodynamics,is accessible to derivation of damage driving force,damage evolution equation and damage evolution criteria. Furthermore,damage evolution equations of time rate are established by the generalized Drucker＇s postulate. The damage evolution equation of cycle rate is obtained by integrating the time damage evolution equations,and the fatigue life prediction method for smooth specimens under repeated loading with constant strain amplitude is constructed. Likewise,for notched specimens under the repeated loading with constant strain amplitude,the fatigue life prediction method is obtained on the ground of the theory of conservative integral in damage mechanics. Thus,the material parameters in the damage evolution equation can be obtained by reference to the fatigue test results of standard specimens with stress concentration factor equal to 1,2 and 3.

基于PSO-ELM组合算法的热力站负荷预测研究

提出了一种粒子群优化极限学习机(PSO-ELM)算法用于热力站负荷预测,应用粒子群(PSO)算法优化极限学习机(ELM)的输入权值和隐含层阈值。将提出的组合算法应用于天津市某小区热力站的负荷预测中,并与ELM、支持向量回归(SVR)和粒子群优化支持向量回归(PSO-SVR)算法在同等条件下进行比较。结果表明,PSO-ELM在预测精度上优于其他算法;在热负荷波动较大时,表现优于PSO-SVR;在一定范围内样本容量对预测结果影响不大,PSO-ELM可遗忘更多的数据。