Short-Term Traffic Flow Forecasting Based on a Novel Combined Model

Sustainability, Год журнала: 2024, Номер 16(23), С. 10216 - 10216

Опубликована: Ноя. 22, 2024

To improve the forecasting accuracy of traffic flow, this paper proposes a flow algorithm based on Principal Component Analysis (PCA) and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) for data processing. The Grey Wolf Optimizer (GWO) is used to optimize weights combined model called GWO-PC-CGLX model, which consists Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), Extreme Gradient Boosting (XGBoost). Initially, PCA CEEMDAN are reduce dimensionality noise in air quality index (AQI) data. smoothed then input into CNN, GRU, LSTM, XGboost models forecasting. accuracy, GWO find optimal weight combination four single models. Taking from Jiayuguan Lanzhou Gansu Province as an example, compared actual data, values evaluation indicator R2 (Coefficient Determination) reached 0.9452 0.9769, respectively, superior those comparison research results not only but also provide effective support construction intelligent transportation systems sustainable management.

Язык: Английский

Evaluation and prediction of fracture energy of fiber reinforced geopolymer concrete

European Journal of Environmental and Civil engineering, Год журнала: 2024, Номер unknown, С. 1 - 23

Опубликована: Дек. 28, 2024

In recent years, Geopolymer concrete (GPC) stands as a promising alternative to traditional Portland cement-based in the quest for eco-friendly and sustainable construction materials. However, practical application of GPC remains limited due need controlled high-temperature curing environments. Furthermore, accurately predicting properties has become an urgent issue reduce time cost associated with laboratory experiments. This study aims investigate how different methods, both on-site precast application, impact fracture energy GPC. Three methods at varying temperatures, room temperature (25 °C), mobile dryer (50 heating cabinet (80 were considered. A comprehensive dataset comprising 194 test results was compiled propose diverse machine learning models capable forecasting energy, without various fibers (Steel, Polypropylene, Basalt, Polyvinyl alcohol…). The findings revealed that increased by 89% when rose from 25 50 °C. further increase 80 °C resulted additional 11% energy. Interestingly, exhibited slightly higher (up 8%) than cement their compressive strengths comparable. developed correlation coefficient values above 0.95 root mean squared error within 10% average actual values, indicating exceptional accuracy regardless fiber presence. Additionally, sensitivity analysis highlighted significance volume content primary factor influencing GPC, while fibers, notch depth ratio emerged determining factor, accounting 78% predictive importance.

Язык: Английский

Data-driven machine learning forecasting and design models for the tensile stress-strain response of UHPC

Structures, Год журнала: 2024, Номер 71, С. 107965 - 107965

Опубликована: Дек. 19, 2024

Язык: Английский