Wyniki wyszukiwania - BazTech

1

Machine learning models for cracking torque and pre‑cracking stiffness of RC beams

Shenggang Chen, Quanquan Guo, Yingying Zhang, Hexiang Hu, Bei Shen

Archives of Civil and Mechanical Engineering

|

2023

|

Vol. 23, no. 1

art. no. e6, 2023

EN

The torsional behavior of RC beams is a complex work involving interactions of different design parameters and mechanisms. Considering the limitations and lower accuracy of traditional calculation theories, two machine learning models, including artificial neural network (ANN) model and random forest (RF) model, were applied for the frst time to predict the cracking torque and initial or pre-cracking torsional stiffness of RC beams. A comprehensive database consisting 159 experimental results of RC beams with solid or hollow sections was compiled, with input variables including dimension parameters of cross-section, compressive stress of concrete, elastic modulus and strength ratio of reinforcements. The performance of the models was appraised by various statistical estimators and safety ratio, and compared with different theories for cracking torque and initial stiffness. Among all the calculation models, RF model achieved the best overall prediction performance with the highest coeffcient of determination (R2=0.985 for cracking torque and R2=0.978 for initial stifness) and lowest root-mean-square error (RMSE=5.867 for cracking torque and RMSE=3.994 for initial stiffness). However, theories for cracking torque, i.e., plastic theory, Bredt thin-tube theory and skew-bending theory, gave huge underestimation, whereas greatly exaggerated initial stiffness was obtained by elastic theory and simplified soften membrane model for torsion theory. Besides, input variable importance analysis was conducted, revealing that dimension parameters of cross-section were the most critical features to decide prediction performance for pre-cracking torsional performance of RC beams. The achievements of this paper may provide references to the establishment of new predicting model for pre-cracking torsional response of RC beams.

2

Supporting of manufacturing system based on demand forecasting tool

Kmiecik Mariusz, Zangana Hawre

LogForum

|

2022

|

Vol. 18, no. 1

33--48

EN

Background: Enterprises’ decision-making could be facilitated by properly creating or choosing and implementing demand forecasting systems. Currently, there are more and more advanced forecasting algorithms based on sophisticated technologies such as artificial neural networks and machine learning. The following research paper focuses on a case study of an automotive manufacturer. The main research aim is to propose the proper demand forecasting tool and show the prospects for implementing the mentioned solution. Methods: The research paper contains the statistical analysis of a chosen time series referring to the demanded quantity of the manufactured products. To create forecasts, models based on the following forecasting algorithms were created: ARIMA, ELM (Extreme Learning Machine), and NNAR (Neural Network Autoregressive). All algorithms are based on the R programming language. All algorithms are run in the same time series where the training and testing periods were established. Results: According to the forecasts ex-post errors and FVA (forecasts value-added) analysis, the best fitting algorithm is the algorithm based on ELM. It yields the most accurate predictions. All other models fail to add value to the forecast. Specifically, the ARIMA models damage the forecast dramatically. Such significant magnitudes of negative FVA values indicate that choosing not to forecast and plan based on the sales of the same period of the previous year is a better choice. However, in the case of the ELM model, the forecasts can be worth the time, finance, and human resources put into preparing them. Conclusion: The increased accuracy of ELM forecasts can contribute to optimizing the process of reaching consensus forecasts. While unconstrained statistical forecasts tend to be overridden, not only to produce constrained forecasts incorporating various variables such as calendar events, promotional activities, supply capacity, and operational abilities, they are also overridden by planners to reflect their foreseeing of demand. The proposed solution could also be easily implemented in the resource planning process to improve it. The proposition of the resource planning process supported by the proposed forecasting system is also shown in the following paper using a BPMN 2.0 (Business Process Modelling Notation 2.0) map.

3

Climate Change and its Effect on the Energy Production from Renewable Sources – A Case Study in Mediterranean Region

Gjika Eralda, Basha Lule, Sokoli Arnisa

Journal of Ecological Engineering

|

2022

|

Vol. 23, nr 12

285--298

EN

In terms of climate forecasting, the Mediterranean region is among the most difficult. It is correlated with the five significant subtropical high pressure belts of the oceans and is symbolized by dry and hot summer and cold and rainy winter. Due to its location in the area, Albania is particularly susceptible to climatic changes. It has been noted that summertime sees the greatest temperature increases. More intense heat waves that stay longer and occur more frequently are anticipated in the eastern Mediterranean. The seasonal patterns of precipitation have not changed, but the amount of rain has become more intense. The effects of climate change have drawn attention to various renewable energy sources, including solar and wind power. In this study, the changes and prospective in average temperature, rainfall, humidity, CO2 emission and their impact in energy production were investigated. Several different models such as Auto Regressive Integrated Moving Average method; Prophet algorithm; Elastic-Net Regularized Generalized Linear Model; Random Forest Regression models; Prophet Boost algorithm; have been built for the study and prediction of each variable. The appropriate models are used to determine the anticipated values of the indicators for a period of four years. The prediction shows an increase in CO2 emission which leads to a decrease in energy production by hydropower. These findings suggest the use of other renewable sources for energy production in the country and the Mediterranean region.

4

Prognozowanie i optymalizacja stężenia związków biogennych w osadzie czynnym z oczyszczalni ścieków przy wykorzystaniu wybranych modeli uczenia maszynowego

Wodołażski Artur

Przemysł Chemiczny

|

2021

|

T. 100, nr 6

565--568

PL

Proces utleniania osadu czynnego jest szeroko stosowany do usuwania zanieczyszczeń w oczyszczalniach ścieków. Rosnąca ilość przetwarzanych odpadów i marnotrawionej żywności prowadzi do wzrostu stężenia azotu całkowitego i fosforu całkowitego, co ma wpływ na pogorszenie jakości przetwarzanych ścieków. Do optymalizacji procesowej użyto algorytmu genetycznego. Do celów predykcji stężenia związków biogennych w osadzie czynnym zastosowano dwa różne modele uczenia maszynowego: sztuczną sieć neuronową (ANN-MLP) i regresyjną metodę wektorów nośnych (SVR). Wyniki pokazały, że oba modele można skutecznie zastosować do prognozowania stężenia związków biogennych, jednak model z użyciem ANN wykazał większą dokładność na etapie treningowym i walidacji niż SVR. Przewidywanie i optymalizacja składu związków biogennych w procesie tlenowego oczyszczania ścieków ma istotny wpływ na jakość uzyskiwanych oczyszczonych ścieków.

EN

A genetic algorithm was used to optimize the process. The concn. of nutrients in the activated sludge was predicted using 2 different machine learning models, artificial neural network (ANN-MLP) and support vector regression (SVR). Both models can be used effectively to predict nutrient concns., but the ANN model showed higher training and validation accuracy than the SVR.

5

Estimation of the depth of penetration in a plunging hollow jet using artificial intelligence techniques

Bodana D., Tiwari N. M., Ranjan S., Ghanekar U.

Archives of Materials Science and Engineering

|

2020

|

Vol. 103, nr 2

49--61

EN

Purpose: Experimental investigations assessment and comparison of different classical models and machine learning models employed with Gaussian process regression (GPR) and artificial neural network (ANN) in the estimation of the depth of penetration (Hp) of plunging hollow jets. Design/methodology/approach: In this analysis, a set of data of 72 observations is derived from laboratory tests of plunging hollow jets which impinges into the water pool of tank. The jets parameters like jet length, discharge per unit water depth and volumetric oxygen transfer coefficient (Kla20) are varied corresponding to the depth of penetration (Hp) are estimated. The digital image processing techniques is used to estimate the depth of penetration. The Multiple nonlinear regression is used to establish an empirical relation representing the depth of penetration in terms of jet parameters of the plunging hollow jets which is further compared with the classical equations used in the previous research. The efficiency of MNLR and classical models is compared with the machine learning models (ANN and GPR). Models generated from the training data set (48 observations) are validated on the testing data set (24 observations) for the efficiency comparison. Sensitivity assessment is carried out to evaluate the impact of jet variables on the depth of penetration of the plunging hollow jet. Findings: The experimental performance of machine learning models is far better than classical models however, MNLR for predicting the depth of penetration of the hollow jets. Jet length is the most influential jet variable which affects the Hp. Research limitations/implications: The outcomes of the models efficiency are based on actual laboratory conditions and the evaluation capability of the regression models may vary beyond the availability of the existing data range. Practical implications: The depth of penetration of plunging hollow jets can be used in the industries as well as in environmental situations like pouring and filling containers with liquids (e.g. molten glass, molten plastics, molten metals, paints etc.), chemical and floatation process, wastewater treatment processes and gas absorption in gas liquid reactors. Originality/value: The comprehensive analyses of the depth of penetration through the plunging hollow jet using machine learning and classical models is carried out in this study. In past research, researchers were used the predictive modelling techniques to simulate the depth of penetration for the plunging solid jets only whereas this research simulate the depth of penetration for the plunging hollow jets with different jet variables.

6

Przegląd metod wykorzystywanych do średnioterminowego prognozowania obciążeń elektroenergetycznych

Pełka Paweł

Przegląd Elektrotechniczny

|

2019

|

R. 95, nr 7

155--159

PL

W artykule dokonano przeglądu metod i modeli prognostycznych dedykowanych średnioterminowemu prognozowaniu obciążeń elektroenergetycznych. Opisano metody modelowania warunkowego i autonomicznego, modele klasyczne, modele inteligencji obliczeniowej i uczenia maszynowego oraz modele oparte na podobieństwie obrazów.

EN

The article reviews the methods and models of the medium-term load forecasting. Methods of conditional and autonomous modeling, classic models, computational intelligence and machine learning models are described, as well as pattern similarity-based models.

7

Information potential of the spectral response of Polish soils, in the NIR range, in the light of lucas database analyses. Soil properties vector model

Gruszczyński Stanisław

Infrastruktura i Ekologia Terenów Wiejskich

|

2019

|

nr II/1

95--104

EN

The paper presents simple machine learning models used for prediction of some soil properties based on the NIR spectral response. Data on mineral soils from Poland were taken from the LUCAS dataset. Machine learning model was used that is included in the category of so-called multilayer perceptron (MLP). The MLP model input was a vector of combined, transformed inputs made by means of the PLSR (partial last squares regression) algorithm (45 inputs in total). The output was a vector of properties, reduced to 9 components due to poor modelling effects of the P and K components. The estimation errors for texture, soil organic carbon (SOC) and carbonates can be considered acceptable from the point of view of their suitability in the development of cartographic documentation. It can be supposed that further regionalization will improve these results.