Predictive modelling for solar thermal energy systems: A comparison of support vector regression, random forest, extra trees and regression trees

Ahmad, Muhammad Waseem; Reynolds, Jonathan; Rezgui, Yacine

doi:10.1016/j.jclepro.2018.08.207

Cited by 374 publications

(151 citation statements)

References 39 publications

Supporting

Mentioning

149

Contrasting

Unclassified

Order By: Relevance

“…The algorithm approximates the relationship between the outputs and inputs, while projecting input space in a higher dimensional space. In the present work, we make use of the framework defined in [44,45]:…”

Section: Support Vector Regressionmentioning

confidence: 99%

“…The coefficients W and b are estimated by using Equation 2i.e., by minimising a regularised risk function [44,45].…”

Section: Support Vector Regressionmentioning

confidence: 99%

“…On the other hand, the loss is the difference between the radius ε of the tube and predicted error, if the predicted value is outside the tube [36]. In order to relax constraints in the estimation of W and b, slack variables ζ 1 and ζ * 1 are introduced leading the above equation to become the primal objective function given by Equation (4) [44,45]:…”

Section: Support Vector Regressionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Highway Networks and Tree-Based Ensemble for Predicting Short-Term Building Energy Consumption

et al. 2018

Self Cite

View full text Add to dashboard Cite

Predictive analytics play a significant role in ensuring optimal and secure operation of power systems, reducing energy consumption, detecting fault and diagnosis, and improving grid resilience. However, due to system nonlinearities, delay, and complexity of the problem because of many influencing factors (e.g., climate, occupants’ behaviour, occupancy pattern, building type), it is a challenging task to get accurate energy consumption prediction. This paper investigates the accuracy and generalisation capabilities of deep highway networks (DHN) and extremely randomized trees (ET) for predicting hourly heating, ventilation and air conditioning (HVAC) energy consumption of a hotel building. Their performance was compared with support vector regression (SVR), a most widely used supervised machine learning algorithm. Results showed that both ET and DHN models marginally outperform the SVR algorithm. The paper also details the impact of increasing the deep highway network’s complexity on its performance. The paper concludes that all developed models are equally applicable for predicting hourly HVAC energy consumption. Possible reasons for the minimum impact of DHN complexity and future research work are also highlighted in the paper.

show abstract

Section: Support Vector Regressionmentioning

confidence: 99%

“…The coefficients W and b are estimated by using Equation 2i.e., by minimising a regularised risk function [44,45].…”

Section: Support Vector Regressionmentioning

confidence: 99%

Section: Support Vector Regressionmentioning

confidence: 99%

See 1 more Smart Citation

Deep Highway Networks and Tree-Based Ensemble for Predicting Short-Term Building Energy Consumption

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…These algorithms are popular among the scientific community due to its ease of use and being able to be applied in multiple fields as in the case of [15], in which they were used to obtain a model of the demand of electricity. Another transversal example is the application to the solar energy field [16], in which authors employed classifiers such as random forest, extra trees, and support vector regression to carry out a comparison among them and to evaluate which hour range is more useful for a solar thermal system.The present article proposes a combination of geomatic techniques along with different free/open sources to delimit potential geothermal areas potential in a simple and non-invasive way. The rest of the paper is organized as follows: Section 2 describes the methodology of the developed approach and the materials employed.…”

mentioning

confidence: 99%

mentioning

confidence: 99%

Detection of Geothermal Potential Zones Using Remote Sensing Techniques

González¹,

Rodríguez‐Gonzálvez²

2019

Remote Sensing

View full text Add to dashboard Cite

The transition towards a new sustainable energy model-replacing fossil fuels with renewable sources-presents a multidisciplinary challenge. One of the major decarbonization issues is the question of to optimize energy transport networks for renewable energy sources. Within the range of renewable energies, the location and evaluation of geothermal energy is associated with costly processes, such as drilling, which limit its use. Therefore, the present research is aimed at applying different geomatic techniques for the detection of geothermal resources. The workflow is based on free/open access geospatial data. More specifically, remote sensing information (Sentinel-2A and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)), geological information, distribution of gravimetric anomalies, and geographic information systems have been used to detect areas of shallow geothermal potential in the northwest of the province of Orense, Spain. Due to the variety of parameters involved, and the complexity of the classification, a random forest classifier was employed, since this algorithm works well with large sets of data and can be used with categorical and numerical data. The results obtained allowed identifying a susceptible area to be operated on with a geothermal potential of 80 W·m −1 or higher.2 of 20 drilling operations. For this reason, different researchers have tried to develop methodologies to reduce costs in the search for these areas, taking into great consideration the use of remote sensing as a means to prioritize zones of potential geothermal energy [8,9], due to the availability of free satellite imagery and its cost-effective ratio for the geothermal exploration of large areas. One approach that involves remote sensing is the search for areas of rocks altered by hydrological action using images from one or more sensors, since they are indicative of thermal fluids being discharged along faults [10]. Surface thermal properties are an indicator of geothermal resources, which can be measured by the use of images from thermal sensors [11,12]. Additionally, temperatures can also be obtained using multispectral imagery [13] acquiring land surface temperature (LST) values.Among the geomatic techniques, geographic information systems (GIS) offer a whole new dimensionality by combining the remote sensing data with other geographic information sources as in [14], in which gravimetric anomalies, thermal images, and the location of geological faults were employed. Among the available GIS techniques, the use of random forest classifiers stands out, since they allow the ability to extract information from categorical and numerical variables. As a result, they support the decision-taking process without significant computational requirements. These algorithms are popular among the scientific community due to its ease of use and being able to be applied in multiple fields as in the case of [15], in which they were used to obtain a model of the demand of electricity. Another transversal example is the app...

show abstract

A machine learning approach for electrical capacitance tomography measurement of gas–solid fluidized beds

Guo

Yang

et al. 2019

AIChE Journal

View full text Add to dashboard Cite

Electrical capacitance tomography has been widely used to obtain key hydrodynamic parameters of gas-solid fluidized beds, which is normally realized by first reconstructing images and then by analyzing these images. This indirect approach is time-consuming and hence difficult for on-line monitoring. Meanwhile, considering recurrence of similar flow patterns in fluidized beds, most of these calculations are repetitive and should be avoided. Here, we develop a machine learning approach to address these problems. First, superficial gas velocity linear-increasing strategy is used to perform highthroughput experiments to collect a large amount of training samples. These samples are used to train the map from normalized capacitance measurements to key parameters that obtained by an iterative image reconstruction algorithm off-line. The trained model can then be used for on-line monitoring. Preliminary tests revealed that the trained models show good prediction and generality for the estimation of the overall solid concentration and the equivalent bubble diameter.

show abstract

Predictive modelling for solar thermal energy systems: A comparison of support vector regression, random forest, extra trees and regression trees

Cited by 374 publications

References 39 publications

Deep Highway Networks and Tree-Based Ensemble for Predicting Short-Term Building Energy Consumption

Deep Highway Networks and Tree-Based Ensemble for Predicting Short-Term Building Energy Consumption

Detection of Geothermal Potential Zones Using Remote Sensing Techniques

A machine learning approach for electrical capacitance tomography measurement of gas–solid fluidized beds

Contact Info

Product

Resources

About