Sie sind auf Seite 1von 11

Prediction of Compressive Strength

of Concrete: Machine Learning


Approaches

Dipro Dutta and Sudhirkumar V. Barai

Abstract Abrams’ law is commonly used to predict the compressive strength of


concrete with respect to the water content of the mix, but it is largely inaccurate.
High-performance concrete, with its complex additional ingredients, makes the
prediction more difficult. The goal of the paper is to find the most accurate model
for prediction of the compressive strength of a given concrete mix using machine
learning (ML). First, the various ML models are explained along with their working
principles. Second, the evaluation methods used for the error analysis in the study
are discussed. Third, the findings of the study are displayed and inferences are
drawn from them. It is found that the 2-nearest-neighbour performs the best with an
error of 8.5% and a standard deviation of 1.55.

Keywords Machine learning  Prediction  Strength of concrete

1 Introduction

The compressive strength is the most important property of a concrete mix.


Abrams’ law of 1919 gives an inverse proportionality between the compressive
strength of hardened concrete and the water-to-cement ratio (w/c) used in the
concrete mix measured by volume [1]. It, hence, implies that the strengths of
distinct but comparable concrete are identical as long as their w/c ratios remain the
same, regardless of the details of their composition. The law is acceptable within
wide limits but is not completely accurate [2]. Abrams assumed that the concrete is
completely compacted. For this, the air gaps present between the cement and the

D. Dutta
Civil Engineering, Manipal Institute of Technology, Manipal 576104, India
e-mail: diprodutta96@gmail.com
S. V. Barai (&)
Department of Civil Engineering, Indian Institute of Technology Kharagpur,
Kharagpur 721302, West Bengal, India
e-mail: skbarai@civil.iitkgp.ernet.in

© Springer Nature Singapore Pte Ltd. 2019 503


A. Rama Mohan Rao and K. Ramanjaneyulu (eds.), Recent Advances in Structural
Engineering, Volume 1, Lecture Notes in Civil Engineering 11,
https://doi.org/10.1007/978-981-13-0362-3_40
504 D. Dutta and S. V. Barai

dry particles of the aggregates need to be removed completely after the addition of
water-to-concrete mix. However, in practice, fully compacted concrete still contains
1% of air gaps [3].
Along with the basic components of conventional concrete (Portland cement,
fine and coarse aggregates and water), HPC incorporates supplementary cementi-
tious ingredients like fly ash, silica fume, blast furnace slag and usually a chemical
admixture, such as superplasticizer. These additional ingredients make HPC a
complex conglomerate, and predicting its strength becomes difficult.
Compressive strength can also be predicted using regression analyses and sev-
eral methods which can be collectively classified as machine learning (ML). One of
the popular machine learning approaches is Artificial Neural Networks (ANN).
The ANN model is inspired by the biological neural network. It utilises several
computing elements to parallel process the input data. These elements, called
artificial neurons, are arranged in layers and interact with each other via weighted
connections. Backpropagation algorithms are popularly used to solve problems of
ANN. Backpropagation minimises the error using a gradient descent method. Yeh
[5] compared regression and ANN for predicting compressive strength of the
concrete. He found that the ANN model performs better than regression.
There is a need to search and develop better machine learning models apart from
well-reported ANN models. This study explores ANN further, along with another
model of ML called the Instance-Based Learning (IBL) or k-Nearest Neighbours
(k-NN).
In the present study, the main objective is to compare the machine learning
models—ANN and IBL—for predicting compressive strength of the concrete. This
study presents—ANN based two back propagation algorithms and the IBL model
for integral values of k ranging from 1 to 6.

2 Background of Tools Used

Machine learning (ML), a subset of artificial intelligence, has evolved from the
study of pattern recognition. Two ML subfields have been studied. Yeh [5] com-
pared regression and ANN for predicting compressive strength of the concrete. He
found that the ANN model performs better than regression.
Instance-based learning (IBL) is a learning method which compares new
problem instances with the instances that the programme has been trained. Its k-
nearest training examples (k-nearest neighbours or k-NN) are utilised to deliver the
output, and the value is determined as the mean of the output values of its k-nearest
neighbours.
The contribution of the neighbours has weights assigned to them. This allows the
output to be more similar to the nearer neighbours than the distant ones.
Artificial neural networks (ANN) is learning method used to solve problems
requiring nonlinear and parallel processing through a mesh of simple computing
elements called artificial neurons. Each neuron can have R inputs (x) and each input is
Prediction of Compressive Strength of Concrete: Machine … 505

weighted (w). There is a bias with a constant input of 1. The sum of the weighted inputs
and the bias passes through the transfer function (f) and generates an output (a).
!
X
R
a ¼ f bþ wi xi
1

Backpropagation algorithm, commonly used in ANN, minimises the error by the


gradient descent method. The neurons are arranged in two or more layers and
interact with each other via weighted connections. The first layer is an input layer
where the data is fed into the network, and the last layer is an output layer which
contains the response of the network to the given input. The intermediate layers
(hidden layers) facilitate the interactions between the elements of the network. The
learning process goes through several iterations (epochs) where the weights are
modified till the network can properly associate between the inputs and their
respective outputs. Once trained, the network can rapidly simulate the desired
output to a given input of similar nature.
Backpropagation further involves several algorithms such as the Levenberg–
Marquardt algorithm (LM) and a variable-learning-rate-with-momentum gradient
descent algorithm (GDX).
LM is an iterative technique that locates the minimum of a function as the sum of
squares of nonlinear functions. It is used to solve nonlinear least-squares problems [4].
In the standard steepest descent algorithm, the training rate is a constant value.
An optimum learning rate is important for the algorithm’s performance. A high rate
makes the algorithm oscillate unstably, whereas a low rate makes the algorithm take
too long to converge. In fact, the ideal rate keeps changing during the training of the
network. The algorithm performs better if the learning rate is allowed to change
during training. GDX implements a momentum factor to determine the varying rate.

3 Data Collection and Analysis

The data was collected from the machine learning repository of the University of
California, Irvine [5]. It provided 1030 samples. The output variable is the com-
pressive concrete strength (MPa) which depends on eight input variables:
1. Cement (kg/m3),
2. Fly ash (kg/m3),
3. Blast furnace slag (kg/m3),
4. Water (kg/m3),
5. Superplasticizer (kg/m3),
6. Coarse aggregate (kg/m3),
7. Fine aggregate (kg/m3)
8. Age at the time of testing (days).
506 D. Dutta and S. V. Barai

Table 1 Feature summary of the data set


Min 1st Quartile Median Mean 3rd Quartile Max
Cement 102 160.1 260.5 265.4 322.2 540
Slag 0 0 94.3 86.3 160.1 359.4
Ash 0 0 59.5 62.8 120.0 200.1
Water 121.7 171 185 183.1 193.2 247
S.P. 0 0 7.8 7.0 10.3 32.2
C.A. 801 882.2 953 956.1 1012.4 1145
F.A. 594 712 769.2 764.4 810.9 992.6

The set includes data from several sources. This verifies the reliability of the data
[2]. Only, the 28-day samples have been taken for the study. The summary of the
425 samples is shown in Table 1.
The data set was randomised before each iteration. 43 samples ( 10%) were
kept aside for testing, while the remaining was used for training the algorithm. The
same sets of randomised data were utilised to conduct the experiment across the
three algorithms to ensure a perfect comparison between the three.

4 Model Development

4.1 Instance-Based Learning

The weighting scheme used in the analysis gave each neighbour a weight, which
was the inverse of the Euclidean distance between the simulation point and the
neighbour. The analysis was run for 1-NN through 6-NN models. The average of
the absolute error in the prediction of the output had been calculated and a graph of
predicted versus expected values of the output was generated.

4.2 Artificial Neural Network

The feedforward neural network is trained using seven inputs in the input layer and
one output in the output layer. Training is the process of presenting testing data to
the network and has it to modify its weights to accurately produce the strength
when presented with the mix proportions. The performance function in the net-
works is mean squared normalised error (MSE). The parameters for the training of
the network are as follows:
For LM,
• No. of hidden layers = 1,
• No. of hidden units = 7,
• Initial momentum (mu) = 0.001,
Prediction of Compressive Strength of Concrete: Machine … 507

• mu increase factor = 0.1,


• mu decrease factor = 10,
• Max. validation failures = 6,
• Max. epochs to train = 1000.
For GDX,
• No. of hidden layers = 1,
• No. of hidden units = 7,
• Learning rate (lr) = 0.01,
• Ratio to increase lr = 1.05,
• Ratio to decrease lr = 0.7,
• Momentum constant (mc) = 0.9,
• mu increase factor = 0.1,
• mu decrease factor = 10,
• Max. validation failures = 6
• Max. epochs to train = 1000.

5 Model Performance

The absolute error percentage (AEP) was calculated for each testing data point and
subsequently, the mean of that was calculated (MAEP). The MAEP thus produced
by each algorithm (IBL, LM and GDX) was statistically analysed using hold-out
and cross-validation. These give an estimate of the accuracy of the performance of
the models in a practical setting [6].
In hold-out (HO), the data was divided randomly into training and testing sets
consisting of 90% and 10% of the entire data. The errors in the output of the testing
set were used for the analysis. Ten iterations of each k-NN, LM and GDX were
evaluated using HO.
Cross-validation (CV) is preferred for prediction models [6]. In k-fold
cross-validation, the data set is equally split into k subsets, each of which is used as
the test set while the others (k-1) are used together as the training set. The average
error of the k trials is calculated. A 10-fold cross-validation has been used for the
analysis.

6 Results

6.1 Instance-Based Learning

1-NN through 6-NN models were run and the experiment was iterated 10 times and
analysed. Table 2 shows the results of the hold-out experiments. The 2-NN model
has shown the best performance and its summary is as follows:
508 D. Dutta and S. V. Barai

Table 2 Arithmetic mean MAEP SD


and standard deviation for the
absolute error percentage in NN 1 9.24 2.56
the hold-out analysis of the NN 2 8.50 1.55
k-NN model NN 3 9.00 1.72
NN 4 8.99 1.78
NN 5 8.77 1.82
NN 6 8.71 1.79

Mean ¼ 8:5%
S:D: ¼ 1:55
Minimum ¼ 5:81%
Maximum ¼ 10:87%
1st Quartile ¼ 7:60%
Median ¼ 8:25%
3rd Quartile ¼ 10:08%

The 10-fold cross-validation analysis on the k-NN model too shows that the
2-NN algorithm performs the best. Table 3 shows evaluation of the analysis, and its
summary is presented as follows:

Mean ¼ 9:32%
S:D: ¼ 0:78
Minimum ¼ 8:52%
Maximum ¼ 11:49%
1st Quartile ¼ 8:99%
Median ¼ 9:17%
3rd Quartile ¼ 9:33%

A scatter plotted graph of the expected values versus the predicted values of the
fifth experiment of 2-NN using hold-out is shown in Fig. 1.

Table 3 Arithmetic mean MAEP SD


and standard deviation for the
absolute error percentage in E 1 9.26 1.53
the cross-validation of the E 2 9.02 2.15
2-NN model E 3 8.98 2.17
E 4 8.52 1.39
E 5 9.35 1.76
E 6 8.65 1.68
E 7 9.25 3.00
E 8 9.08 1.56
E 9 9.62 1.54
E 10 11.49 3.03
Prediction of Compressive Strength of Concrete: Machine … 509

Fig. 1 Scatter plot of expected versus predicted values for Exp. 5 of hold-out for 2-NN

Table 4 Arithmetic mean and standard deviation of ANN models using hold-out
LM GDX
MAEP 15.18 20.18
SD 2.36 6.93

6.2 Artificial Neural Networks

A hold-out analysis was performed on the artificial neural network models. Table 4
shows the AEPs and MAEP as calculated from the LM and GDX algorithms. The
summary of the findings are as follows:
For LM,

Mean ¼ 15:18%
S:D: ¼ 2:36
Minimum ¼ 11:39%
Maximum ¼ 18:67%
1st Quartile ¼ 13:80%
Median ¼ 15:54%
3rd Quartile ¼ 16:78%
510 D. Dutta and S. V. Barai

Table 5 Arithmetic mean MAEP SD


and standard deviation for the
absolute error percentage in E 1 9.26 1.53
the cross-validation of the LM E 2 9.02 2.15
algorithm E 3 8.98 2.17
E 4 8.52 1.39
E 5 9.35 1.76
E 6 8.65 1.68
E 7 9.25 3
E 8 9.08 1.56
E 9 9.62 1.54
E 10 11.49 3.03

Table 6 Arithmetic mean MAEP SD


and standard deviation for the
absolute error percentage in E 1 29.86 13.02
the cross-validation of the E 2 27.82 14.00
GDX algorithm E 3 21.43 7.37
E 4 23.04 11.29
E 5 25.49 11.82
E 6 21.61 7.17
E 7 19.77 5.63
E 8 23.66 8.00
E 9 24.71 6.75
E 10 19.85 4.71

For GDX,

Mean ¼ 20:18%
S:D: ¼ 6:93
Minimum ¼ 12:89%
Maximum ¼ 38:39%
1st Quartile ¼ 16:84%
Median ¼ 18:62%
3rd Quartile ¼ 20:49%

Tables 5 and 6 show the evaluation of the 10-fold analysis of the fifth experi-
ment of neural network method using LM and GDX, respectively. Their summaries
are as follows:
Prediction of Compressive Strength of Concrete: Machine … 511

For LM,

Mean ¼ 14:2%
S:D: ¼ 1:2
Minimum ¼ 12:83%
Maximum ¼ 16:33%
1st Quartile ¼ 13:1%
Median ¼ 14:08%
3rd Quartile ¼ 14:75%

For GDX,

Mean ¼ 19:45%
S:D: ¼ 4:19
Minimum ¼ 14:52%
Maximum ¼ 30:29%
1st Quartile ¼ 17:42%
Median ¼ 18:62%
3rd Quartile ¼ 20:56%

Scatter plotted graphs of the expected values versus the predicted values of the
seventh experiment of LM and GDX using hold-out are shown in Figs. 2 and 3,
respectively.

Fig. 2 Scatter plot of


expected versus predicted
values for Exp. 7 of hold-out
for LM
512 D. Dutta and S. V. Barai

Fig. 3 Scatter plot of


expected versus predicted
values for Exp. 7 of hold-out
for GDX

7 Comparative Study

The analysis shows that the ML algorithms can predict the compressive strength of
28-day hardened concrete with varying accuracy. A comparison of the data in
Tables 2 and 4 is shown in Fig. 4.
Figure 4 shows that IBL performs considerably better than either of the neural
network algorithms. It follows that the 2-NN model performs slightly better than the
other k-NN models with an MAEP of 8.50%. The 2-NN model also has the lowest
SD at 1.55.

Fig. 4 Comparison of the MAEPs and SDs of the three algorithms


Prediction of Compressive Strength of Concrete: Machine … 513

The variable-learning-rate-with-momentum gradient descent algorithm (GDX) has


a very high-performance error at 20.18%. The standard deviation of 6.93 indicates that
it may not be dependable to produce consistent results.
The Levenberg–Marquardt algorithm (LM) performs better than GDX and other
gradient descent algorithms. It has an error of 15.18% and a standard deviation of
2.36.

8 Conclusions

HPC is a complex material and that makes the modelling and predicting its
behaviour difficult. The study was an attempt to demonstrate the possibilities of
using machine learning to predict the compressive strength of concrete. The ini-
tialisation values were not optimised, and hence there is a scope for further study of
the ANN models. Optimising the initial weights and constants may give better
results.
It must be noted that none of the explored methods is applicable to extrapolation.
Hence, the minimum and maximum of the training set must be selected carefully.
Also, more training instances will produce more accurate predictions. Hence, larger
training data sets should be utilised for modelling.
Based on the comparison of the models, the following can be concluded:
• Instance-Based Learning (IBL) is a preferred choice for the prediction of the
compressive strength with the seven parameters.
• The neural network models are comparatively easier to train and hence easier to
simulate the predictions, but their performance is relatively poor.
• The Levenberg–Marquardt (LM) algorithm can be used for approximations
where the tolerance for error is high (preferably above 15%).
• It is advisable to not use the variable-learning-rate-with-momentum gradient
descent algorithm (GDX) for the prediction.

References

1. Kleiger, P. (1994). Significance of tests and properties of concrete and concrete-making


materials (Issue 169, Part 3). ASTM.
2. Yeh, I.-C. (1998). Modeling of strength of high performance concrete using artificial neural
networks. Cement and Concrete Research, 28(12), 1797–1808.
3. Gani, M. S. J. (1997). Cement and concrete. Chapman & Hall.
4. Lourakis, M. I. A. (2005). A brief description of the Levenberg-Marquardt algorithm
implemented by levmar. Foundation of Research and Technology, 4, 1–6.
5. Yeh, I. C. (2007, August 03). UCI machine learning repository: Data set. Irvine: University of
California.
6. Reich, Y., & Barai, S. V. (1999). Evaluating machine learning models for engineering
problems. Artificial Intelligence Engineering, 13.

Das könnte Ihnen auch gefallen