Estimation of concrete compressive strength using artificial neural network

Abstract

In present paper, concrete compressive strength is evaluated using back propagation feed-forward artificial neural network. Training of neural network is performed using Levenberg-Marquardt learning algorithm for four architectures of artificial neural networks, one, three, eight and twelve nodes in a hidden layer in order to avoid the occurrence of overfitting. Training, validation and testing of neural network is conducted for 75 concrete samples with distinct w/c ratio and amount of superplasticizer of melamine type. These specimens were exposed to different number of freeze/thaw cycles and their compressive strength was determined after 7, 20 and 32 days. The obtained results indicate that neural network with one hidden layer and twelve hidden nodes gives reasonable prediction accuracy in comparison to experimental results (R=0.965, MSE=0.005). These results of the performed analysis are further confirmed by calculating the standard statistical errors: the chosen architecture of neural network shows the smallest value of mean absolute percentage error (MAPE=, variance absolute relative error (VARE) and median absolute error (MEDAE), and the highest value of variance accounted for (VAF).

U radu se daje procena čvrstoće betona pri pritisku, primenom veštačkih neuronskih mreža s prostiranjem signala unapred i propagacijom greške unazad. Obučavanje mreže sprovodi se korišćenjem Levenberg-Markart algoritma obučavanja za četiri različite arhitekture neuronskih mreža, s jednom jedinicom, tri jedinice, te osam i dvanaest jedinica u skrivenom sloju, radi odbacivanja efekta ,,pretreniranja'. Treniranje, validacija i testiranje neuronskih mreža izvodi se na osnovu rezultata eksperimentalnog ispitivanja čvrstoće pri pritisku na 75 uzoraka betona, s različitim vodocementnim faktorom i količinom superplastifikatora tipa melamina. Ispitivani uzorci betona izlagani su različitim ciklusima zamrzavanja/ otkravljivanja, a njihova čvrstoća pri pritisku određivana je nakon 7, 20 i 32 dana. Dobijeni rezultati ukazuju na to da neuronska mreža s dvanaest jedinica u skrivenom sloju daje ocenu čvrstoće zadovoljavajuće tačnosti u poređenju sa eksperimentalno dobijenim podacima (R≈0,97, MSE=0,005). Rezultati izvedene analize dodatno su potvrđeni sračunavanjem vrednosti standardnih statističkih grešaka: najmanjom vrednošću srednje apsolutne greške (MAPE), varijanse relativne vrednosti apsolutne greške (VARE) i medijane (MEDAE), kao i najvećom vrednošću sračunate varijanse (VAF) za izabranu arhitekturu neuronske mreže.