D3AT-LSTM: An Efficient Model for Spatiotemporal Temperature Prediction Based on Attention Mechanisms DOI Open Access
Ting Tian,

Huijing Wu,

Xianhua Liu

и другие.

Electronics, Год журнала: 2024, Номер 13(20), С. 4089 - 4089

Опубликована: Окт. 17, 2024

Accurate temperature prediction is essential for economic production and human society’s daily life. However, most current methods only focus on time-series modeling prediction, ignoring the complex interplay of meteorological variables in spatial domain. In this paper, a novel model (D3AT-LSTM) proposed by combining three-dimensional convolutional neural network (3DCNN) attention-based gated cyclic network. Firstly, historical series eight surrounding pixels are combined to construct multi-dimensional feature tensor that integrates from temporal domain as input data. Convolutional units used analyze spatiotemporal patterns local sequence CNN modules them with parallel attention mechanisms. The fully connected layer finally makes final prediction. This method subsequently compared both classical state-of-art models such ARIMA (AR), long short-term memory (LSTM), Transformer using three indices: root mean square error (RMSE), absolute (MAE), coefficient determination (R2). results indicate D3AT-LSTM can achieve good accuracy AR, LSTMs, Transformer.

Язык: Английский

Optimization of enzymatic hydrolysis of corn starch to obtain glucose syrups by genetic algorithm DOI Creative Commons
Jonathan Serrano, Jesús Luis-Orozco, Héctor L. Ramírez

и другие.

DYNA, Год журнала: 2025, Номер 92(235), С. 83 - 91

Опубликована: Март 19, 2025

Este trabajo corresponde a la optimización de las variables operación hidrólisis enzimática almidón maíz para obtención jarabes glucosa utilizando el algoritmo genético Matlab (2020a). Para ello, proceso se modeló matemáticamente mediante metodología superficie respuesta. El diagrama Pareto indicó que sacarificación ejercen mayor influencia en conversión del almidón. modelo matemático es gran utilidad una mejor comprensión y control operacional nivel industrial. La solución problema muestra puede alcanzarse un equivalente máximo dextrosa 98,13% si realiza condiciones operacionales óptimas, cuales comprobaron experimentalmente. Los resultados muestran que, alcanzar rendimiento, licuefacción debe llevarse cabo temperatura 92oC, pH 6,3, dosis α-amilasa 1,5 mg enzima/g tiempo 1 hora; mientras realizarse 57oC, 4,9, glucoamilasa 1,15 34 horas. fenómeno reversión detectó cuando superó 35 horas, con incidencia negativa sobre dextrosa.

Процитировано

0

Influence of Dissolved Solids Composition on the Impurities Sedimentation of Alkalinized Sugarcane Juice DOI
Jonathan Serrano, Jesús Luis-Orozco, Osney Pérez Ones

и другие.

Sugar Tech, Год журнала: 2025, Номер unknown

Опубликована: Май 22, 2025

Язык: Английский

Процитировано

0

Sustainable Construction in post-industrial Ursus district in Warsaw: Biologically Active Areas on Roofs and Underground Garages DOI Open Access
Magdalena Daria Vaverková, Michał Kosakiewicz, Karolina Krysińska

и другие.

Inżynieria Bezpieczeństwa Obiektów Antropogenicznych, Год журнала: 2024, Номер 3, С. 10 - 35

Опубликована: Сен. 3, 2024

Urbanization leads to the expansion of urban areas and increased population density, which has a negative impact on natural resources, including green spaces. To address these environmental challenges, various measures sustainable development practices offer environmental, economic societal benefits. This article provides an overview different types roofs, discusses their advantages disadvantages, outlines current legislation biologically active in new residential developments. A project Ursus district Warsaw was used as case study compare intensive extensive roofs. The involved installation vegetation roof underground garage later building, with detailed descriptions layers used. One experiment examined seasonal variations water runoff from Three test sites simulated seasons, substrates containers watered twice daily. Results showed that spring, temperatures around 15°C, provided optimal conditions for establishment, while winter posed challenges due frost. Both roofs have positive social impacts, supporting pillars development. Examples global infrastructure further illustrate

Язык: Английский

Процитировано

0

D3AT-LSTM: An Efficient Model for Spatiotemporal Temperature Prediction Based on Attention Mechanisms DOI Open Access
Ting Tian,

Huijing Wu,

Xianhua Liu

и другие.

Electronics, Год журнала: 2024, Номер 13(20), С. 4089 - 4089

Опубликована: Окт. 17, 2024

Accurate temperature prediction is essential for economic production and human society’s daily life. However, most current methods only focus on time-series modeling prediction, ignoring the complex interplay of meteorological variables in spatial domain. In this paper, a novel model (D3AT-LSTM) proposed by combining three-dimensional convolutional neural network (3DCNN) attention-based gated cyclic network. Firstly, historical series eight surrounding pixels are combined to construct multi-dimensional feature tensor that integrates from temporal domain as input data. Convolutional units used analyze spatiotemporal patterns local sequence CNN modules them with parallel attention mechanisms. The fully connected layer finally makes final prediction. This method subsequently compared both classical state-of-art models such ARIMA (AR), long short-term memory (LSTM), Transformer using three indices: root mean square error (RMSE), absolute (MAE), coefficient determination (R2). results indicate D3AT-LSTM can achieve good accuracy AR, LSTMs, Transformer.

Язык: Английский

Процитировано

0