A hybrid method based on a new clustering technique and multilayer perceptron neural networks for hourly solar radiation forecasting

Azimi, Rasool; Ghayekhloo, M.; Ghofrani, M.

doi:10.1016/j.enconman.2016.04.009

Cited by 125 publications

(48 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The objective of a clustering algorithm is to identify groups of similar objects, where objects in a cluster are more similar to each other than objects in different clusters (Halkidi et al, 2001). This study applied the well-known and commonly-used k-means clustering technique (MacQueen, 1967;Azimi et al, 2016;Bae et al, 2017). According to Hartigan and Wong (1979), the aim of the k-means algorithm is to divide m objects in n dimensions into k (where k ≤ n) partitions (or clusters), such that the within-cluster sum of squares is minimised.…”

Section: Methods 31 the K-means Clustering Methodsmentioning

confidence: 99%

“…According to Hartigan and Wong (1979), the aim of the k-means algorithm is to divide m objects in n dimensions into k (where k ≤ n) partitions (or clusters), such that the within-cluster sum of squares is minimised. The similarity between a pair of objects is defined by their distance, where the Euclidean distance is often used as a distance measure (Azimi et al, 2016). In the present study, each object is a daily profile of CC.…”

Section: Methods 31 the K-means Clustering Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Investigating diffuse irradiance variation under different cloud conditions in Durban, using k-means clustering

Govender

Sivakumar

2019

J. energy South. Afr.

View full text Add to dashboard Cite

Diffuse irradiance is important for the operation of solar-powered devices such as photovoltaics, so it is important to analyse its behaviour under different sky conditions. The primary cause of short-term irradiance variability is clouds. One approach to analyse the diffuse irradiance variation is to use cluster analysis to group together days experiencing similar cloud patterns. A study was carried out to examine the application of k-means clustering to daily cloud data in Durban, South Africa (29.87 °S; 30.98 °E), which revealed four distinct day-time cloud cover (CC) patterns classified as Class I, II, III and IV, corresponding to cloudy, sunny, or a combination of the two. Diffuse irradiance was then correlated with each of the classes to establish corresponding diurnal irradiance patterns and the associated temporal variation. Class I had highest diffuse irradiance variation, followed by Classes III, IV and II. To further investigate the local cloud dynamics, cloud types were also analysed for Classes I−IV. It was found that stratocumulus (low cloud category); altocumulus translucidus, castellanus and altocumulus (middle cloud category); and cirrus fibrates and spissatus (high cloud category), were the most frequently occurring cloud types within the different classes. This study contributes to the understanding of the diurnal diffuse irradiance patterns under the four most frequently occurring CC conditions in Durban. Overall, knowledge of these CC and associated diffuse irradiance patterns is useful for solar plant operators to manage plant output where, depending on the CC condition, the use of back-up devices may be increased or reduced accordingly.

show abstract

Section: Methods 31 the K-means Clustering Methodsmentioning

confidence: 99%

Section: Methods 31 the K-means Clustering Methodsmentioning

confidence: 99%

Investigating diffuse irradiance variation under different cloud conditions in Durban, using k-means clustering

Govender

Sivakumar

2019

J. energy South. Afr.

View full text Add to dashboard Cite

show abstract

“…raw data clustering and classification to create multiple homogeneous data groups with uniform patterns, for facilitating more efficient AI learning processes and leading to better prediction performances. Data classification methods observed in this review include principal component analysis [204], k-means clustering algorithm [205], k-fold cross validation [206], decision tree, SVM [207], and game theoretic self-organising map (GTSOM) [208]. Significant performance improvements were observed in these studies relative to model formulation without data classification.…”

Section: Referencesmentioning

confidence: 97%

Review of distributed generation (DG) system planning and optimisation techniques: Comparison of numerical and mathematical modelling methods

Theo

Lim

et al. 2017

Renewable and Sustainable Energy Reviews

242

105

View full text Add to dashboard Cite

“…The results revealed good concordance among measures, and estimates with RMSE ranged from 2.727 to 2.807 MJ m À2 . There are several ML models and the Artificial Neural Network (ANN) and Support Vector Machine (SVM) are the most widely used [12][13][14][15][16]. Yadav and Chandel [13] reviewed the main studies in the literature using ANN to estimate solar radiation.…”

Section: Introductionmentioning

confidence: 99%

Assessment of ANN and SVM models for estimating normal direct irradiation (Hb)

Santos

Escobedo

Teramoto

et al. 2016

Energy Conversion and Management

View full text Add to dashboard Cite

a b s t r a c tThis study evaluates the estimation of hourly and daily normal direct irradiation (H b ) using machine learning techniques (ML): Artificial Neural Network (ANN) and Support Vector Machine (SVM). Time series of different meteorological variables measured over thirteen years in Botucatu were used for training and validating ANN and SVM. Seven different sets of input variables were tested and evaluated, which were chosen based on statistical models reported in the literature. Relative Mean Bias Error (rMBE), Relative Root Mean Square Error (rRMSE), determination coefficient (R 2 ) and ''d" Willmott index were used to evaluate ANN and SVM models. When compared to statistical models which use the same set of input variables (R 2 between 0.22 and 0.78), ANN and SVM show higher values of R 2 (hourly models between 0.52 and 0.88; daily models between 0.42 and 0.91). Considering the input variables, atmospheric transmissivity of global radiation (kt), integrated solar constant (H sc ) and insolation ratio (n/N, n is sunshine duration and N is photoperiod) were the most relevant in ANN and SVM models. The rMBE and rRMSE values in the two time partitions of SVM models are lower than those obtained with ANN. Hourly ANN and SVM models have higher rRMSE values than daily models. Optimal performance with hourly models was obtained with ANN4 h (rMBE = 12.24%, rRMSE = 23.99% and ''d" = 0.96) and SVM4 h (rMBE = 1.75%, rRMSE = 20.10% and ''d" = 0.96). Optimal performance with daily models was obtained with ANN2 d (rMBE = À3.09%, rRMSE = 18.95% and ''d" = 0.97) and SVM2 d (rMBE = 0.60%, rRMSE = 19.39% and ''d" = 0.97). ANN and SVM models improved H b estimations as compared with other results from the literature. SVM has better performance than ANN to estimate H b , and it should be the first option of choice.

show abstract

A hybrid method based on a new clustering technique and multilayer perceptron neural networks for hourly solar radiation forecasting

Cited by 125 publications

References 23 publications

Investigating diffuse irradiance variation under different cloud conditions in Durban, using k-means clustering

Investigating diffuse irradiance variation under different cloud conditions in Durban, using k-means clustering

Review of distributed generation (DG) system planning and optimisation techniques: Comparison of numerical and mathematical modelling methods

Assessment of ANN and SVM models for estimating normal direct irradiation (Hb)

Contact Info

Product

Resources

About