Artificial neural network-based temperature prediction in heat sinks with cross cuts fins. (2015)

• Record Type: Journal Article
• Authors: Kannan, V.; Arunkumar, S.; Venkatesan, Muniyandi
• Is Part Of: International journal of advanced intelligence paradigms. Volume 7:Number 3/4(2015)
• Page Start: 292
• Page End: 301
• Publication Date: 2015
• Subjects: artificial neural networks -- ANNs -- cross cut fins -- heat transfer -- numerical simulation -- temperature prediction -- heat sinks -- base plate temperature -- modelling; Artificial intelligence -- Periodicals; Machine theory -- Periodicals; Fuzzy logic -- Periodicals; 006.305
• Languages: English
• ISSNs: 1755-0386

Abstract

Soft computing techniques are predominantly used in varied sectors of research for accurate prediction and optimisation. Recent advancements in artificial neural network (ANN) and its capability to predict the data have increased its range of applications including heat transfer problems. The present paper aims to predict base plate temperature of cross cut fins of variable height (5-15 mm), fin spacing (5.85-32 mm), number of crosscuts (0-5), heat duty (20-100 W) for constant fin thickness (3 mm) and base plate dimensions (180 × 250 mm). A neural network model is designed for a rectangular heat sink with cross cut fins. The model is developed using the data from numerical heat transfer calculation carried out in ANSYS FLUENT©. The numerical model is validated with similar experimental results available in literature. The simulated data are used to train several neural-network configurations and the best neural-network model is used for prediction. The ANN model is used to predict the base plate temperatures of cross cut fins. The results reveal the ability of using ANN for accurate prediction of base plate temperature.