Artificial Neural Network (ANN) Modeling of Cavitation Mechanism by Ultrasonic Irradiation for Cyanobacteria Growth Inhibition

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

J. Environ. Treat. Tech.

ISSN: 2309-1185

Journal web link: http://www.jett.dormaj.com

Artificial Neural Network (ANN) Modeling of

Cavitation Mechanism by Ultrasonic Irradiation

for Cyanobacteria Growth Inhibition

Esmaeel Salami Shahid , Marjan Salari , Majid Ehteshami , Solmaz Nikbakht Sheibani

Ph.D. Candidate of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran

Department of Civil Engineering, Sirjan University of Technology, Kerman, Iran

Associate Professor, Civil and Environmental Eng. Dept. KN Toosi Univ. of Technology, Tehran, Iran

M.Sc. Candidate, Environment Eng. Dept. Shiraz University, Shiraz, Iran

Received: 03/12/2019

Accepted: 12/02/2020

Published: 20/05/2020

Abstract

Cyanobacteria produce toxins that affect animals and human’s health. Therefore, modeling concentration of this type of algae

is necessary. This study employs artificial neural network (ANN) modeling method to simulate the cavitation mechanism by

ultrasonic irradiation on cyanobacteria concentration variation in treated water. The proposed model used parameters such as power

intensity, frequency and the time of ultrasound irradiation as input variables. The results showed that proportional value of

cyanobacteria concentration to the initial concentration (C/C ). The data obtained from a laboratory experiment and number of data

in the existed study was not enough for ANN modeling, the data expanded to 7280 data sets from the original 28 data sets obtained

by the experimental study. A feed-forward learning algorithm with 20 neurons in the first (hidden) layer and one neuron in the

-5

second layer was developed with the MSE value equals to 2.72×10 . Model results were used for predicting the cell density value.

Furthermore, a novel formulation was presented to correlate the C/C values with the cell density. To verify the accuracy of the

ANN and developed equation, the value of cell density was predicted by studies performed by other researchers. In this case the

MSE was 1.55×10 .

Keywords: Artificial Neural Networks, Cyanobacteria, Modeling, Ultrasonic irradiation, Water Quality

Introduction¹

et al., 2004). Therefore, ultrasonic irradiation, due to its

advantages, including safety, cleanliness and energy

conservation, can be a suitable way to reduce pollution in

aquatic resources containing harmful algae. Only a few

researches have been developed since cyanobacterial bloom

control by ultrasound is a fairly new ﬁeld. Moreover, the

ultrasonic process is not affected by the toxicity and low

biodegradability of compounds (Fu et al., 2007). Ultrasound

waves affect larger molecules faster than smaller ones

Cyanobacteria affect animals and human’s health due to

poison production. The presence of this type of algae in

nutrient-rich drinking water reserves has adverse effects in

terms of taste, smell and health (Leclercq et al., 2014).

Hence, it is necessary worldwide to control the blooms of

algae and to remove the garbage before the distribution of

water in an appropriate manner in accordance with the WHO

guidelines and customer expectations (Leclercq et al., 2014).

Today, ultrasonic devices have various applications in

different fields of science, industry, medical practice and so

forth (Sharma et al., 2011). One of the innovative techniques

for improvement of water/wastewater treatment process is

the application of ultrasonic waves; that does not require the

addition of oxidants or catalyst and does not generate

additional waste streams as compared to adsorption or

ozonation processes (Zhang et al., 2009; Wang and Yuan

(Yamamoto et al., 2015). The ultrasonic method uses

electrical energy to induce physical, chemical and biological

treating effects as shown in Fig.1 (Kasaai, 2013). Sound

waves with a frequency between 20 KHz to 200 MHz are

called ultrasonic waves. When an ultrasonic wave enters a

liquid, it can cause cavitation (Wu et al., 2012). Fig.2

illustrates the ultrasonic range due to its applications. In

addition, bloom-forming cyanobacteria participate in

diverse consortia, and symbioses with a broad array of

microorganisms, higher plants and animals, which help

016). The ultrasonic method is known to have a detrimental

effect on the structure and function of organisms (Hongwei

Corresponding author: Marjan Salari, Department of Civil

Engineering, Sirjan University of Technology, Kerman,

Iran. E-mail: salari.marjan@gmail.com.

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

alleviate environmental stresses and limitations (Pilli et al.,

reactors was 1200 and 800 ml cyanobacterial suspension that

was filled in experiments (Lee et al., 2002; Ma et al., 2005).

Cyanobacteria, or blue-green algae, occur worldwide often

in calm, nutrient-rich waters. Some species of cyanobacteria

produce toxins that affect animals and humans. People may

be exposed to cyanobacteria toxins by drinking or bathing in

contaminated water. The most frequent and serious health

effects are caused by drinking water containing toxins or by

ingestion during recreational water contact like swimming.

Cyanobacteria can also cause problems for drinking water

treatment systems. Therefore, studying and modeling the

concentration of this type of algae is necessary to prevent its

adverse effects on humans and animals.

011; Rajasekhar et al., 2012). In this study, the apparent gap

between experimental data and modeling, by trying to

determine if artificial intelligence models can predict

cavitation mechanism for inhibiting algal growth. According

to the best author’s knowledge, artificial neural network

modeling and network performance evaluation are not

considered in order to determine the cavitation mechanism

by ultrasonic irradiation to control cyanobacterial citrate

growth.

.1 Cavitation process

When a liquid is sonicated, dissolved gas molecules are

entrapped by micro-bubbles that grow and expand upon

rarefaction of the acoustic cycle; these micro-bubbles then

release extreme temperatures upon adiabatic collapse (Hao

et al., 2004; Lin et al., 2008). The process is based on the

phenomenon of acoustic cavitation, which involves the

formation, growth, and sudden collapse of micro-bubbles

that generate short-lived, localized “hot spots” in an

irradiated liquid (Wang et al., 2007).

The high-temperatures (5000 K) and pressures (1000

atm) induced by cavitation's in collapsing gas bubbles in

aqueous solution lead to the thermal dissociation of water

molecules into reactive free radicals H0 and OH0 (Shimizu

et al., 2007; Goel et al., 2004 ). There are three possible

reaction sites in ultrasonically irradiated homogeneous

liquids: (i) the gaseous interiors of collapsing cavities; (ii)

the interfacial liquid region between cavitation's bubbles and

the bulk solution, where high-temperatures (ca. 1000-2000

K) and high temperature gradients exist; and (iii) the bulk

solution at ambient temperature, where small amounts of

OH0 diffuse from the interface. The sonochemical effect

takes place at the gas-liquid interface due to the oxidation of

organic molecules by OH0 and, to a lesser extent, in the bulk

solution or the pyrolytic decomposition inside the bubbles

Figure 1: Physical, chemical and biological effects of electrical

energy (Kasaai, 2013)

(Li et al., 2008; Eren, 2012). Hydrophilic and non-volatile

compounds such as dyes mainly degrade through OH0

mediated reactions in the bulk solution and at the bubble-

liquid interface, while hydrophobic and volatile species

degrade thermally inside the bubbles (Merouani et al., 2015;

Wu et al., 2012).

Figure 2: Diagram of the ultrasonic range (Wu et al., 2012)

Cavitation

temperature/pressure fields and free radicals (such as OH

and H , etc) in liquids (Fig. 3) (Pang et al., 2011).

phenomena

produce

high

Material and Methods

Three ultrasonic devices, two beaker systems (operating

2 2

at 200 kHz and 1.7 MHz, respectively) and one horn system

operating at 20 kHz), are employed in this study. The

Cavitation consists of the repetition of three distinct steps:

formation (nucleation), rapid growth (expansion) during the

cycles until it reaches a critical size, and violent collapse in

the liquid in less than a microsecond (Chowdhury and

Viraraghavan, 2009) as shown in Fig. 4. Parameters that

affect cavitation are liquid temperature, external pressure,

liquid viscosity, amount and type of solved gases, the surface

tension of a liquid (Fu et al., 2007). All these parameters are

related to the liquid but the parameters of sound wave that

are effective discuses below:

(

ultrasonic generator designed in our laboratory consists of a

voltage-controlled oscillator (VCO) (Zhang et al., 2009),

power amplifier, matched impedance and feedback unit. The

ultrasound at 20 kHz was emitted from a titanium horn

dipped into the cyanobacterial suspension, and the ultrasonic

at higher frequencies (200 kHz and 1.7 MHz) was emitted

from the piezo-electric discs of lead zirconate titanate fixed

on the underside of the beaker reactors with epoxy. Each

frequency required a specific emitter. Their ultrasonic

powers dissipated in the medium were measured

calorimetrically (Paerl, 2018). The volume of ultrasonic

2.2 Cavitation near surfaces

The most important effects of ultrasonic on liquid-solid

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

systems are mechanical and attributed to asymmetric

cavitation. In addition, shockwaves that have the potential to

create microscopic turbulence are produced within

interfacial films surrounding nearby solid particles (Pang et

al., 2011).

generate larger values of acoustic pressure. An increase in

the value of Pa lead to a more and violent collapse (Hongwei

et al., 2004).

Hence frequency has a diverse relation with the bubble

sizes, in (relatively) low-frequency irradiations (16-

00KHz) will produce large cavitation bubbles which results

in high temperature and pressure in the cavitation zone

Tsaih et al., 2004). As the frequency increases the cavitation

(

zone becomes less violent and, in the MHz, range no

cavitation's is observed and the main mechanism is acoustic

streaming (Hongwei et al., 2004). I, is the power intensity of

the ultrasonic wave in terms of the energy transmitted per

unit time per unit normal area of fluid:

1

(2)

I  P_a_,_m_a_x(2  c)

where ρ is density of a medium/liquid, c is velocity of sound

in that medium and P is the maximum pressure amplitude

of the wave (Hongwei et al., 2004; Shchukin et al., 2011).

Increasing intensity will increase the cavitation and also

violent collapse of bubbles (Tsaih et al., 2004). This increase

will continue until reaching an optimum point and after that

point increase in power, the intensity will reduce the rate of

cavitation. For example, Mutiarani et al., showed that

removing turbidity increases with increasing power to 60

watts and decreases for powers beyond 60w (Zhang et al.,

Figure 3: Reaction zone in cavitation process (Aadapted from

Duong Pham et al., 2009)

016). Whereby the effect of power on removal efficiency

(of microcystins) was studied using 30, 60 and 90 watt in the

constant frequency of 20 KHz after 1, 5, 10, 20 minutes of

exposure to the ultrasonic samples. The results are shown in

Fig.5.

Figure 4: Growth and implosion of cavitation bubbles in aqueous

solution under ultrasonic irradiation (Merouani et al., 2015)

Asymmetric collapse leads to the micro-jet formation of

solvent that collides with the solid surface at tremendous

force, resulting in newly exposed, highly reactive surfaces as

well as corrosion and erosion. These phenomena increase the

rate of mass transfer near the catalyst surface (Chowdhury

and Viraraghavan, 2009).

.3 Acoustic pressure

The acoustic pressure is a sinusoidal wave dependent on

time (t), frequency (f) and the maximum pressure amplitude

of the wave, Pa, max and is represented by the following

equation (Kuna et al., 2017; Hongwei et al., 2004):

Figure 5: The decrease of Turbidity at 28 kHz and 1 Hour of

Irradiation time with variation of Power (Hatanaka et al., 2002)

This phenomenon may be explained by bubble shielding

effect. When the power intensity is high enough, a dense

cloud of navigational bubbles accumulates around the

ultrasonic transducer. The cavitation bubbles attenuate

sound waves due to both scattering and absorption and thus

impede the propagation of sound waves, especially at the

resonant size (Mutiarani and Trisnobudi 2012).

This study intends to develop the ANN model(s) to

simulate the concentration of microsystems after using

ultrasonic waves to remove them or changing those

(

^Pa ^^P

a,max

sin(2.f .t)

where, Pa is the acoustic pressure which directly

proportional to more and violent collapse of bubbles, Pa,

max is the maximum pressure amplitude of the wave which

directly proportional to the input power of the transducer, t

and fare time and frequency of ultrasonic waves,

respectively (Hongwei et al., 2004; Shchukin et al., 2011). A

sufficiently large increase in the intensity of ultrasound will

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

pollutants to removable compounds. Input variables such as

time, frequency and power of the applied ultrasonic are used

to determine the proportional (C/C0) as the target value of

the model(s). The data from Ma, et al., are used in this study

(

Rajasekhar et al., 2012). Then the results are verified by

distinct study made by Hao et al., 2004 and Zhang et al.,

016. Ultrasonic wave may Influence Cyanobacteria in

several ways: (a) sinking Cyan bacteria by rapture vehicles

that filled with gases which causes the flotation of

Cyanobacteria; (b) disruption of photosynthesis; (c) damage

of cell membranes due to lipid peroxidation; and (d)

differential susceptibility to ultrasonic waves at different

stages in the cell division cycle (Gerde et al., 2012; Zhang et

al., 2006). Most experiments in this field are performed in

frequencies between 20-28 KHz and in some researchers

used frequencies up to 1.7 MHz (Lee et al., 2002; Ma et al.,

005). Therefore, this study evaluates the effect of different

ultrasonic frequencies such as 20,150,410 and 1700 KHz (in

constant power of 30w) after 1,5,10 and 20 minutes of

irradiation on Microcystis as the typical representative of

bloom-forming algae.

Figure 7: (a) Effect of ultrasonic irradiation (20 kHz, 30 W) for

different time on Microcystis suspension, including changes of

Microcystis biomass (Lee et al., 2002) and (b) Cyanobacterial

biomass as a function of time during ultrasonic irradiation at 20 kHz,

0W (Ma et al., 2005)

.4 Artificial Neural Network Method

The artificial neural network (ANN), as its name

implies, is a technique for simulation of the human brain

functions during the problem–solving process, which has

been developed and originated about 60 years ago. The

neural network approach can be applied to the powerful

computation of complex nonlinear relationships, just as

humans apply knowledge gained from past experience to

new problems or situations (Tang et al., 2004). Thus, for

modeling parameters that don’t have a simple (linear)

relationship with input data, ANN method can be employed

effectively. The MLF (multilayer feed-forward) networks

trained with back-propagation algorithm are the most

popular type of networks (Salami et al., 2016 a,b; Salari et

al., 2018). For example, models that marked by (*) in Table2

have used feed-forward networks for their development

Figure 6: (a) The removal of microcystins dissolved in water after

ultrasonic irradiation at 20 kHz and various powers of 30, 60, and

0 W with time (Hatanaka et al., 2002) and (b) The removal of

microcystins dissolved in water after ultrasonic irradiation at 30W

and various frequencies of 20, 150, 410 kHz, and 1.7 MHz with time

(Tang et al., 2004).

(Hao et al., 2004)

.5 Structure of the Networks

The basic architecture consists of three types of neuron

Also, some research worked on the effect of ultrasonic

layers: input, hidden, and output layers. Fig. 10 shows a two-

layer network. In feed-forward ANN networks, the signal

flow from input to output units, strictly in a feed-forward

direction (Samani et al., 2007; Koncsos, 2010). Hidden

layers consist of a different number of neurons. Fig. 8 shows

a parameter such as “a” is the output of neuron and “p” is

the input. Parameters w and p are weight and bias

respectively. All parameters denoted as matrices, and can be

expressed as (Salami et al., 2015).

irradiation on cyanobacteria. Their work is very similar to

Ma et al., 2005). In both works (see Fig.6) they used UV–

(

vis spectrophotometer (Ultra-Spec 2000, Amersham

Biosciences AB, Uppsala, Sweden). Furthermore, Ma et al.,

calibrated the device on 684 nm and Hao et al., found an

optical density of cell suspension of 560 nm as an optimum.

The initial concentration in both experiments was 2µg/L.

Therefore, it is possible to compare the results of the current

ANN model with (Ma et al., 2005; Hao et al., 2004).

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

(

l1)

(l)

i, j

e(w,b)





i, j

(l)

(5)

(6)



i, j

(

l1)

(l)

i, j

e(w,b)

(l)

^bi, j





i, j







e(w,b)



(i)

(l)

i, j





(l)^e⁽^w^,^b^;

x y

) 

(





m i1







(7)

i, j

i. j



Figure 8: A two layer feed-forward network







e(w,b)



(i)

(l)

i, j





(l)^e⁽^w^,^b^;

x y

) 

(



^bi, j

m



^bi. j





(8)



a  f (net)  f (n)  f (w .p  b)  f ( w_R.p  b)

(3)



i1

푛

푚푠푒 = ∑ 푒²

푖

(9)

∑(푡 − 푎 )

푖

(

p 



p , p ,..., p



, w 



w , w ,..., w



푚

푖ꢀꢁ

where α is the learning rate. In this study, the mean square

error (MSE) is the criterion for comparing the outputs.

Where ti is the target (real) value and ai is the network

output. Two feed-forward networks with back propagation

learning rule (Eqs.9-12) are used to develop the models in

MATLAB environment. The design parameters of the

networks have been. Other training parameters of the models

are shown in Table 1.

The most common "f" functions are presented in Fig. 9.

These transfer functions transfer output of each layer to a

simpler more useful expression for calibrating the wi and bi

(

s) in next layer/step.

Table 1: Training parameters

Feed-Forward back

0.001

0.1

Network type

propagation

Trainlm (Levenberg-

Marquardt)

α decrees

Training function

Adaptive learning

function

α increase

Train GDM

maximum

E+10

Performance function

Transfer function

MSE

min grad

1.00E-10

Tensing(x)

The validation of all equations/models after

development, have been tested with precision parameters

such as R or R* and MAE. R* is used in cases that we had

negative values of R. In other words, when푦 >

푦̅_푗, R will be

푗

negative.

y  y _j





Figure 9: Transfer functions

The training process determines the ANN weights and is

(10)

j1

y _j

R 



y _j

similar to the calibration of a mathematical model. In order

to perform training correctly, we must iteratively continue

and repeat the process of calibrating and optimizing the wi,

bi(s), with the final target of minimizing the mean square

error (MSE) value as possible as it is, the process will

continue until the required precisions reached. In the

following procedure, weights and biases will change every

time the process is repeated. The calibration process for wi,

bi(s) is as (Abraham et al., 2005):







for y  y  R 

j 1

 _y

y _j

for y  y  R 

 _y





R  R

R 

푖

|푦_ꢃ^ꢄ − 푦_ꢃ|

ꢅ

ꢂ

ꢃꢀꢁ

푀퐴퐼 =

(11)

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

^풙풎,풆

^풙풎

풂 =

풆

(15)

Results and Discussion

.1 Data development

where xm,0 is the C/C

cell density in t = 0 , obtained from Fig. 6a, xm,e is C/C

end time (T) , calculated with Eq. 11, x

in t = 0 , calculated with Eq. 11, x

Fig.6 shows measured data that can be used in this study.

They are just 28 sets (12 sets obtained from Fig.6a and 16

sets obtained from Fig.6b). Since the size of data sets did not

look enough for developing a reliable ANN model, we used

a simple interpolation method to expand these 28 sets of data

to 7280 sets (Hao et al., 2004). To implement the procedure

all three (input) parameters of time (t), power (I) and

frequency (f) were prepared with smaller intervals compared

to Fig 6 (a,b). Actually, the number of data was increased to

a reasonably sufficient amount by simple interpolation. For

example, in step 1, the irradiation duration was modified

from 1 to 20 min, one minute by minute. In step 2, power

changed from 30W to 90W by 10W each time, and in step 3

the frequency range expanded from 20 to 400 KHz by 10

KHz and from 500 KHz to 1.7 MHz by 100 KHz each time.

Furthermore, all possible compositions of input parameters

were considered in the potential range.

is cell density in end

time (T), obtained from Fig. 6a. Fig. 12 shows the results of

the model and developed equation (Eq. 13) and their

comparison with the real data (Fig.6a). The input parameters

in Fig.6.b such as (I = 40W, f = 30 KHz, t = 0, 1, 2, …, 10

min) were used in the generated model and the results of

model (r) is converted to cell density using Eq. 13. Finally,

the results of the proposed procedure is compared with the

results of Hao et al., (2004) which is shown at Fig. 12b.

In order to ensure the model accuracy, the original data

(

28 data sets from Ma et al., (2005) removed from learning

data and just were used to validate the models after the

development. Since Ln(C/C ) exhibited a rather linear

relationship with time, it was chosen as the target value to

obtain a more accurate and reliable model. The target values

of 7280 data sets prepared and adjusted for modelling are

presented in Fig. 10.

Figure 10: 7280 obtained target values of 7280 data sets prepared

for modeling

In order to obtain the optimum removal rate at each

condition, we consider all possible combinations of input

values (time (t), power (I), frequency (f)) within the desired

ranges which can describe the oscillating pattern of the target

data C/C . The primary proposed models developed and

verified assuming linear or polynomial correlations between

input values time (t), power (I) and frequency (f) to the target

parameter Ln(C/C ). However, the analysis showed a weak

correlation for linear and polynomial models. Therefore,

MATLAB software was used to generate ANN models by

applying the input parameters such as I,t and f as feed data

to predict Ln(C/C ) as the target element.

Figure 11: Comparison between model results and real data

In this regard, 7280 sets of data were used for modelling

in which 70% for training, 15% for validation and the other

.2 Cell density simulation

Input parameters such as Fig. 6a data, (I = 30 W, f = 20

5% for testing. The network that is used has one hidden

layer with 20 neurons. After developing the model, all 7280

sets of) input data were used to estimate Ln(C/C ) for each

KHz, t = 0, 1, 2..., 9 min) were used as input parameters in

the ANN model and C/C for all 10 conditions were

simulated. Then a novel formula such as Eq. (13-15) is

developed to show the relationship between the C/C and

cell density.

(

data set, the overall average MSE was 2.72×10-5 which

implies how accurate the model is. Figure 11 shows all 28

sets of original data that were used for modelling compared

with model results. The average MSE for the following 28

data sets was 3.15×10-4. Accordingly, the model outcome

x_m

cell density

is Ln of C/C .

t.(a  a )

(13)

a₀

퐶

= 푒^푟

푥_ꢆ

(12)

퐶0

^xm,0

a₀

(14)

x₀

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

Conclusion

In recent times wastewater treatment by ultrasonic has

become a popular treatment technique due to its ability to

degrade pollutants at the end of treating products, i.e. CO

water and organic acids. Ultrasonic dye degradation is a

complete, irreversible degradation process that provides an

appropriate, safe wastewater treatment method with non-

toxic and stable products. The main achievement of this

study is an ANN model that can simulate the value of

Ln(C/C ), for cyanobacteria at minutes of ultrasound

irradiation such as t; with frequency of ‘f’ and power

intensity of ‘I’. By using parameters such as I, f and t (as

input parameters), model will simulate the value of

Ln(C/C ). Accuracy of the model demonstrated in all

domains as Time: between 0 to 20 minutes, Power intensity:

between 30 to 90 watts, Frequency between 20 KHz to 1.7

MHz, Initial concentration of cyanobacteria: 2 µg/L. The

model has the ability of adaptation with new data and can be

updated/calibrated with new/different data. Result of the

ANN model and Salami’s equation (that both made by data

from Ma et al., 2005, verified completely with an

outstanding, but similar work such as Hao et al., 2004. This

study also shows a method for expanding data with a simple

interpolation technique that can increase the number of data

(sets) by:

Figure 12: (a) the results of the model and Eq. 12 compared with

real data and (b) Model verification using the comparative results of



Breaking the differences between (existed) input

parameters to smaller fractions and find values of target

parameters (in those points) by interpolation

Combining different experiment that is done in similar

conditions

Considering all possible combinations of input

parameters; and it is a procedure that can be used to

describe experimental data sets.

(Hao et al., 2004)

Whereby, the input parameters in Fig.6.b (I = 40W, f =



0 KHz, t = 0, 1, 2, …, 10 min) were entered to model and

the results (r) converted to cell density using Eq. 12 which

can be compared to the results of (Hao et al., 2004). It worth

to note that the model predicts the value of C/C closely near

to 1 in time zero while the data less than one minute were

not used at the model training stage. The model results can

be used for:

Acknowledgments

The authors are grateful and thankful of Civil and

Environmental Laboratory at Shiraz University for

providing facilities and apparatus for the current study and

analyses.

Optimizing problems

Calibrating measurement equipment

Finding errors in measurements, caused by

operator/equipment fault

Reproducing missing/bad data

The other main outcome of the current study is the

Ethical issue

Authors are aware of, and comply with, best practice in

publication ethics specifically with regard to authorship

relation between C/C0-removal and cell density which leads

to develop a promising state-of-the-art equation called

Salami’s equation:

(avoidance of guest authorship), dual submission,

manipulation of figures, competing interests and compliance

with policies on research ethics. Authors adhere to

publication requirements that submitted work is original and

has not been published elsewhere in any language.

C /C0

t.(a  a )

(16)

SalamiEq .



a₀

where α is a factor that depends on the passed time, initial

and endpoint values of the (C/C ) and it’s proportional to the

value of cell density. The value of the (C/C ) can be obtained

from the experiment and/or the presented ANN model. The

model simulates C/C very close to 1 at time t equals to zero.

It is obvious that C in time zero should be equals to C and

thus the C/C must be 1. However, it should be considered

that the model is just using times equals or above one minute

and the fact that it can predict the behavior of target in time

zero, indicate that the model has some kind of intelligence

for predicting data that has never encountered before.

Competing interests

The authors declare that there is no conflict of interest

that would prejudice the impartiality of this scientific work.

Authors’ contribution

All authors of this study have a complete contribution

for data collection, data analyses and manuscript writing.

Journal of Environmental Treatment Techniques

0202, Volume 8, Issue 2, Pages: 625-633

Merouani S, Ferkous H, Hamdaoui O, Rezgui Y, Guemini M.

A method for predicting the number of active bubbles in

sonochemical reactors. Ultrasonics sonochemistry. 2015 Jan

References

Abbasi M, Asl NR. Sonochemical degradation of Basic Blue 41

dye assisted by nanoTiO2 and H2O2. Journal of hazardous

materials. 2008 May 30;153(3):942-7.

;22:51-8.

Muthukumaran S, Kentish SE, Stevens GW, Ashokkumar M.

Application of ultrasound in membrane separation processes: a

review. Reviews in chemical engineering. 2006;22(3):155-94.

Mutiarani IM., and Trisnobudi A. Ultrasonic Irradiation in

Decreasing Water Turbidity, 2009..

Ma B, Chen Y, Hao H, Wu M, Wang B, Lv H, Zhang G.

Influence of ultrasonic field on microcystins produced by

bloom-forming algae. Colloids and Surfaces B: Biointerfaces.

Abraham A. Artificial neural networks. Handbook of

measuring system design. 2005 Jul 15.

Barzamini R, Menhaj MB, Kamalvand S, Tajbakhsh A. Short

term load forecasting of Iran national power system using

artificial neural network generation two. In2005 IEEE Russia

Power Tech 2005 Jun 27 (pp. 1-5). IEEE.

Chowdhury P, Viraraghavan T. Sonochemical degradation of

chlorinated organic compounds, phenolic compounds and

organic dyes–a review. Science of the total environment. 2009

Apr 1;407(8):2474-92.

005 Mar 25;41(2-3):197-201.

Pilli S, Bhunia P, Yan S, LeBlanc RJ, Tyagi RD, Surampalli

RY. Ultrasonic pretreatment of sludge: a review. Ultrasonics

sonochemistry. 2011 Jan 1;18(1):1-8.

Pang YL, Abdullah AZ, Bhatia S. Review on sonochemical

methods in the presence of catalysts and chemical additives for

treatment of organic pollutants in wastewater. Desalination.

Pham TD, Shrestha RA, Virkutyte J, Sillanpää M. Recent

studies in environmental applications of ultrasound. Canadian

Journal of Civil Engineering. 2009 Nov;36(11):1849-58.

Eren Z. Ultrasound as a basic and auxiliary process for dye

remediation: a review. Journal of Environmental Management.

011 Aug 15;277(1-3):1-4.

012 Aug 15;104:127-41.

Paerl HW. Mitigating toxic planktonic cyanobacterial blooms

in aquatic ecosystems facing increasing anthropogenic and

climatic pressures. Toxins. 2018 Feb;10(2):76.

Rajasekhar P, Fan L, Nguyen T, Roddick FA. A review of the

use of sonication to control cyanobacterial blooms. Water

research. 2012 Sep 15;46(14):4319-29.

Shimizu N, Ogino C, Dadjour MF, Murata T. Sonocatalytic

degradation of methylene blue with TiO2 pellets in water.

Ultrasonics sonochemistry. 2007 Feb 1;14(2):184-90.

Shchukin DG, Skorb E, Belova V, Moehwald H. Ultrasonic

cavitation at solid surfaces. Advanced Materials. 2011 May

Fu H, Suri RP, Chimchirian RF, Helmig E, Constable R.

Ultrasound-induced destruction of low levels of estrogen

hormones in aqueous solutions. Environmental science &

technology. 2007 Aug 15;41(16):5869-74.

Goel M, Hongqiang H, Mujumdar AS, Ray MB. Sonochemical

decomposition of volatile and non-volatile organic

compounds—a comparative study. Water Research. 2004 Nov

;38(19):4247-61.

Gerde JA, Montalbo-Lomboy M, Yao L, Grewell D, Wang T.

Evaluation of microalgae cell disruption by ultrasonic

treatment. Bioresource technology. 2012 Dec 1;125:175-81.

Hao H, Wu M, Chen Y, Tang J, Wu Q. Cavitation mechanism

in cyanobacterial growth inhibition by ultrasonic irradiation.

Colloids and Surfaces B: Biointerfaces. 2004 Feb 15;33(3-

;23(17):1922-34.

Salami ES, Salari M, Ehteshami M, Bidokhti NT, Ghadimi H.

Application of artificial neural networks and mathematical

modeling for the prediction of water quality variables (case

study: southwest of Iran). Desalination and Water Treatment.

):151-6.

Hatanaka SI, Yasui K, Kozuka T, Tuziuti T, Mitome H.

Influence of bubble clustering on multibubble

016 Dec 1;57(56):27073-84.

Salari M, Shahid ES, Afzali SH, Ehteshami M, Conti GO,

Derakhshan Z, Sheibani SN. Quality assessment and artificial

neural networks modeling for characterization of chemical and

physical parameters of potable water. Food and Chemical

Toxicology. 2018 Aug 1;118:212-9.

Norman ES, Dunn G, Bakker K, Allen DM, De Albuquerque

RC. Water security assessment: integrating governance and

freshwater indicators. Water Resources Management. 2013 Jan

sonoluminescence. Ultrasonics. 2002 May 1;40(1-8):655-60.

Hao H, Wu M, Chen Y, Tang J, Wu Q. Cavitation mechanism

in cyanobacterial growth inhibition by ultrasonic irradiation.

Colloids and Surfaces B: Biointerfaces. 2004 Feb 15;33(3-

):151-6.

Kasaai MR. Input power-mechanism relationship for ultrasonic

irradiation: Food and polymer applications.

Koncsos T., The application of neural networks for solving

complex optimization problems in modeling. In Conference of

Junior Researchers in Civil Engineering, 2010. pp. 97-102.

Kuna E, Behling R, Valange S, Chatel G, Colmenares JC.

;27(2):535-51.

Samani N, Gohari-Moghadam M, Safavi AA. A simple neural

network model for the determination of aquifer parameters.

Journal of Hydrology. 2007 Jun 30;340(1-2):1-1.

Salami ES, Ehteshami M. Simulation, evaluation and prediction

modeling of river water quality properties (case study: Ireland

Rivers). International journal of environmental science and

technology. 2015 Oct 1;12(10):3235-42.

Chen D, Sharma SK, Mudhoo A. Handbook on applications of

ultrasound: sonochemistry for sustainability. CRC press; 2011

Jul 26.

Tsaih ML, Tseng LZ, Chen RH. Effects of removing small

fragments with ultrafiltration treatment and ultrasonic

conditions on the degradation kinetics of chitosan. Polymer

degradation and stability. 2004 Oct 1;86(1):25-32.

Tang JW, Wu QY, Hao HW, Chen Y, Wu M. Effect of 1.7 MHz

ultrasound on a gas-vacuolate cyanobacterium and a gas-

vacuole negative cyanobacterium. Colloids and Surfaces B:

Biointerfaces. 2004 Jul 15;36(2):115-21.

Wang J, Jiang Y, Zhang Z, Zhang X, Ma T, Zhang G, Zhao G,

Zhang P, Li Y. Investigation on the sonocatalytic degradation

of acid red B in the presence of nanometer TiO2 catalysts and

Sonocatalysis:

a potential sustainable pathway for the

valorization of lignocellulosic biomass and derivatives.

InChemistry and Chemical Technologies in Waste Valorization

017 (pp. 1-20). Springer, Cham.

Lee TJ, Nakano K, Matsumura M. A novel strategy for

cyanobacterial bloom control by ultrasonic irradiation. Water

Science and Technology. 2002 Sep;46(6-7):207-15.

Li M, Li JT, Sun HW. Decolorizing of azo dye Reactive red 24

aqueous solution using exfoliated graphite and H2O2 under

ultrasound irradiation. Ultrasonics Sonochemistry. 2008 Jul

;15(5):717-23.

Leclercq DJ, Howard CQ, Hobson P, Dickson S, Zander AC,

Burch M. Controlling cyanobacteria with ultrasound.

InINTER-NOISE and NOISE-CON Congress and Conference

Proceedings 2014 Oct 14 (Vol. 249, No. 3, pp. 4457-4466).

Institute of Noise Control Engineering.

Lin JJ, Zhao XS, Liu D, Yu ZG, Zhang Y, Xu H. The

decoloration and mineralization of azo dye CI Acid Red 14 by

sonochemical process: rate improvement via Fenton's reactions.

Journal of Hazardous Materials. 2008 Sep 15;157(2-3):541-6.