Concrete Strength Prediction

clangedbivalveΤεχνίτη Νοημοσύνη και Ρομποτική

19 Οκτ 2013 (πριν από 4 χρόνια και 19 μέρες)

110 εμφανίσεις

Concrete Strength Prediction
Using a Neural Network

Siripong Malasri, Ella Thorsteinsdottir, and
Jittapong Malasri

Christian Brothers University

650 East Parkway South

Memphis, TN 38104

Contact: pong@cbu.edu

Presented at MAESC 2006 Conference


3/31/06

© 2006 by S. Malasri

ASCE Concrete Canoe

Canoe Mix Ingredients

Cementitious

Materials:

cement, silica fume, fly ash

Aggregates:

perlite, glass bubble, microsphere

Water:


Admixtures:

air
-
entraining agent, superplasticizer

Concrete Mix Design


ACI method has been widely used for regular
weight concrete


No method has been established for lightweight
canoe concrete

Neural Networks


Neural networks are
capable of learning
from examples


Test neural network
concept with regular
weight concrete

Methodology


Generate 93 concrete mixes
(cases) from a computer
program based on ACI
methoc


74 cases were shown to a
backpropagation network


All 93 cases were used to
validate the network
performance

Input/Output

8 Input Factors

1 Output

Network Configuration

8 input cells, 14 hidden cells, 1 output cell

Learning

Results

All 93 cases (seen and unseen) are within 2.5% error!

Full Paper

Challenges for Canoe Mix


More components


Time consuming in
generating cases
from actual mixes
in lab


Inconsistency of
data from lab
mixes

Questions?

Canoe Design & Construction

Design Phase


Mix design


Hull design


Reinforcement

Construction Phase


Form construction


Reinforcement layout


Concrete pouring


Curing


Touch
-
up

Form: Profile Layer

Form: Styrofoam Layer

Form: Plaster Layer

Form: Plastic Layer

Form: Rib Construction

Reinforcement: Frame

Reinforcement: Wall

Reinforcement: Overall

Concrete: Mixing

Concrete: Pouring

Concrete: Curing

After all this hard work!