Академический Документы
Профессиональный Документы
Культура Документы
Regular article
a r t i c l e i n f o a b s t r a c t
Article history: An artificial neural network (ANN) was first developed to predict the transmembrane pressure in an
Received 17 October 2017 anoxic-aerobic membrane bioreactor (AO-MBR) treating domestic wastewater. A few studies about pre-
Received in revised form 10 January 2018 diction of membrane fouling in MBRs using ANNs have been published so far, even though our recent
Accepted 2 February 2018
work indicates that ANNs show a great potential for this application. In this study, 10 parameters linked to
Available online 5 February 2018
wastewater treatment and measured in the different parts of the AO-MBR system were used as the input
variables of the ANN. The goal was to select the most relevant input parameters to predict the evolution of
Keywords:
the transmembrane pressure based on the performances of the ANN. An ANN model was selected for its
Artificial neural network
Membrane bioreactor
satisfying performances (R2 = 0.850). In conclusion, ANNs could be a valid method to predict membrane
Membrane fouling fouling in AO-MBR systems treating domestic wastewater.
Transmembrane pressure © 2018 Elsevier B.V. All rights reserved.
Wastewater
https://doi.org/10.1016/j.bej.2018.02.001
1369-703X/© 2018 Elsevier B.V. All rights reserved.
48
Table 1
A summarization of the inputs, structure, output, and limitations for the ANNs used in MBRs.
Piron et al. [21] Rt TMP Recurrent Error = 11% • Dynamic simulation of membrane fouling but constant
Crossflow velocity operational parameters, which does not represent real
SS conditions in MBRs
Hamachi et al. [22] Ep TMP Recurrent Error = 10%
J Crossflow velocity
In recent years, many different types of ANNs have been used to nitrogen (NH4 -N), nitrate (NO3 -N), total phosphorus (TP) on this
model membrane systems [21,22,28–32]. In particular, ANNs were target.
developed to predict the evolution of the hydraulic resistance and
the evolution of deposit thickness and permeate flux respectively
2. Materials and methods
with the help of recurrent networks by taking TMP and other influ-
ent characteristics as inputs [21,22]. However, in these ANN models
2.1. MBR operation
only a single set of the initial input values was used to predict by
recurrence the whole set of the outputs. A cascade forward ANN
A lab-scale AO-MBR system has been used in this study (Fig. S1).
has been developed to predict the performance of a small-scale
The working volumes of the anoxic and aerobic basin were 25 and
submerged MBR treating cheese whey. The ANN input variables
50 L, respectively. The HRT of the anoxic and aerobic basins were
were the influent total chemical oxygen demand (COD), ammo-
3 and 6 h, respectively. In order to facilitate nitrogen removal, the
nia, nitrate and total phosphate concentrations, sludge retention
reactor was able to internally recycle the mixed liquor (IR = 300%Q)
time (SRT), hydraulic retention time (HRT), the permeate flux and
using a peristaltic pump (Masterflex, Model GT-300D, Green Tech)
the TMP [23]. A perfect match between measured and estimated
connected to the aerobic and anoxic basins. The anoxic basin was
outputs was obtained for both training and testing phases. How-
equipped with a low speed mixer (Model MS 30120, 0–1500 rpm,
ever, TMP and flux evolutions were taken as inputs, thus the ANN
Lab. Stirrer) to maintain the MLSS in suspension. Five flat mem-
was not considering them as outputs. The effluent characteristics
brane modules (Table S1) were submerged in the aeration basin.
of a SBR-MBR treating synthetic hypersaline wastewater were pre-
The operating flux of the membrane was maintained at 17 L/m2 h
dicted by an ANN which was composed of a single hidden layer.
(LMH) by a peristaltic pump (Masterflex, Model GT-150D, Green
Nevertheless, the prediction of membrane fouling characteristics
Tech). The pump was operated using cycles of 10 min of suc-
was not considered in the study [24]. Similarly, feed forward neu-
tion followed by 2 min of pause. In order to minimize membrane
ral networks composed of a single hidden layer were developed to
fouling due to the cake layer formation, an aeration system was
predict the COD, NH4 + -N and PO4 3− -P concentrations in the efflu-
placed below the membrane module, with the cross velocity of
ent of a submerged MBR treating municipal wastewater. However,
uplifting air flow maintained within the range 20–22 cm/s. The DO
the membrane fouling characteristics such as TMP or permeate
in the aerobic basin was maintained within the range 3–4 mg/L.
flux were also not considered in the model [25]. Geissler et al. [26]
In order to control the DO concentration in the anoxic basin at
developed an ANN to predict the permeate flux evolution in a pilot
less than 0.5 mg/L for efficient denitrification, a degassing tank
MBR. The ANN consists of a single hidden layer. The inputs were
(Volume = 12.5 L; HRT = 30 min) was added in the internal recycle
composed of TMP, rate of TMP change, filtration cycle length, back-
line, before the anoxic basin. The system was operated for about
wash cycle length, SRT, total suspended solids, temperature and
180 days to treat domestic wastewater. The wastewater character-
oxygen decay rate in the aerobic zone. The permeate flux was the
istics used in this study are shown in Table 2.
output of the ANN. However, in this study, the parameters related
to wastewater characteristics such as COD and dissolved oxygen
(DO) concentrations or related to the mixed liquor such as mixed 2.2. ANN modelling development
liquor suspended solids (MLSS) concentration were not considered,
although they have a significant influence on membrane fouling. An algorithm sketch of the model development process in this
Moreover, the predicted values of the permeate flux were likely to study was presented in Fig. 1. The procedure includes several steps.
overfit the experimental data, because the number of hidden neu- (i) Data collection and preprocessing: The main objective of the
rons was high. More recently, Mirbagheri et al. [27] investigated data preprocessing is to determine the suitable locations for the
membrane fouling in a pilot submerged MBR plant. The main goal data required for the modelling activities. In this step, the inputs
of the study was to determine the effect of simultaneous aeration on and target outputs were normalized between 0 and 1 using the
submerged MBR performances by developing an ANN which would function mapminmax. (ii) Model design and network training: The
be able to predict the TMP and the permeability of the membrane model architecture, training method, and training rates were deter-
over time. The problem was approached with two different mod- mined using a trial-and-error approach. In this study, the Neural
els including a multilayer perceptron neural network and a radial Network Toolbox V9.1 of MATLAB® mathematical software was
basis function neural network. For both simulations, the input vari- used to develop ANNs with the aim of predicting the TMP evolu-
ables were the operational time, total suspended solids, COD in tion in the AO-MBR by considering different input variables related
the influent, SRT and MLSS. The outputs consisted of permeabil- to the wastewater treatment. The overall structure of a basic ANN
ity and TMP. The ANN models were composed of one input layer, along with the diagram of a neuron are represented on Fig. 2. In
one output layer and a single hidden layer. Then the prediction of the first phase of the ANN development, simple ANNs were created
the permeability and TMP was considered. Nevertheless, the whole using the fitnet function, which returns a basic backpropagation
experiment was based on a single series of measures which means network presenting a single hidden layer. For each test, the num-
that the ANN models obtained could possibly not be able to predict ber of hidden neurons was fixed to 5 in order to study the influence
membrane fouling for various working conditions of the reactor. of the input variables chosen to feed the ANN and optimize the
The development of ANNs to predict membrane fouling in a selection of the wastewater treatment parameters which have to
basic filtration system stayed until now limited to simple models be taken into account for TMP prediction. Defining a small number
taking few and constant parameters as inputs. To the best of our of neuron allows indeed to avoid overfitting, which is the conse-
knowledge, a few studies about prediction of membrane fouling in quence of a large number of weights which tends to give the ANN
MBRs using ANNs have been published so far. Our recent work indi- the capacity to be flexible and easily fit the training dataset [34]. As
cates that ANNs show a great potential to predict membrane fouling seen in Fig. 2, each connection between the nodes (or neurons) is
in MBRs [33]. Therefore, the aim of this study was to develop an ANN assigned with a weight and a bias, specific to each neuron, is added
model to predict membrane fouling from long-term anoxic-aerobic to the sum of the weighted inputs. Moreover, an activation function
MBR (AO-MBR) operational data. In particular, the ANN was trained is defined for each layer of the ANN. Its role is to compute the output
to predict the TMP considering the effect of different inputs such of a neuron in function of its input and to introduce non-linearity
as pH, alkalinity (Alk), MLSS, COD, total nitrogen (TN), ammoniacal to the network, since these functions are commonly sigmoids. The
activation functions are specific to a layer and can be different for
50 F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58
Table 2
Characteristics of the influent wastewater used in the study.
Range pH DO (mg/L) COD (mg/L) MLSS (mg/L) TN (mg/L) TP (mg/L) Alkalinity (mgCaCO3 /L) NO3 -N (mg/L) NH4 -N (mg/L)
Max 7.68 1.24 290.0 112.0 66.4 5.80 280.0 1.40 57.10
Min 7.36 0.79 154.0 18.0 42.8 3.80 230.0 0.10 30.10
Average 7.51 1.05 226.7 61.2 55.2 4.93 252.6 0.52 42.55
Standard deviation 0.07 0.11 27.0 20.8 5.4 0.37 10.5 0.34 6.43
Fig. 2. Graphical representation of the structure of a basic ANN (A) and of a neuron (B).
the hidden layers and the output layer. Eq. (1) gives the expression In Eq. (2), X is the normalized input, X is the original input and
of the output yi obtained from the i-th neuron of a defined layer as Xmin and Xmax are the minimum and maximum values respectively
a function of the input xj coming from the j-th neuron of the previ- of the input variables over the whole range of input values of the
ous layer (over n neurons) and entering this neuron, the weight wij dataset. The learning (or training) phase consists in adapting the
specific to the interconnection between the i-th neuron and the j-th weights and the biases of the ANN which are randomly defined at
neuron of the previous layer, the bias bi and the activation function the creation of the network, in order to minimize its performance
f. factor often defined as the mean squared error (MSE). The learning
procedure was launched with the function train which initialized
j=1 the random initial weights and biases for the ANN and applied the
yi = f bi + xj × wij (1) LM algorithm to minimize the MSE. The MSE, whose expression is
n given by Eq. (3), allows to compare, for the same inputs, the pre-
dicted outputs given by the ANN with the target outputs measured
In order to fit the demands of the activation functions, inputs and during the experiment.
targets of the ANN are normalized between 0 and 1. The normal-
ization function is given by Eq. (2).
i=1 2
(i)
X − Xmin MSE = 1/N Y (i) − Yt (3)
X = (2)
Xmax − Xmin N
F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58 51
Table 3 Table 4
Summary of the parameters used as input variables for the study. Variations in the performances of the ANN trained 10 times.
Influent Degassing tank Anoxic tank MBR Effluent 1 0.39 0.201 2−1-4-3
2 0.43 0.192 4−2-3-1
COD (mg/L) x x x x x 3 0.55 0.171 1-2-3-4
MLSS (mg/L) x x x x x 4 0.28 0.231 2-4−1-3
MLVSS (mg/L) x x x 5 0.43 0.194 4−2-1-3
pH x x x x 6 0.31 0.212 1-4-3-2
DO (mg/L) x x x 7 0.54 0.176 4−1-2-3
Alkalinity (mgCaCO3 /L) x x x x 8 0.36 0.209 3−1-4-2
TN (mg/L) x x x x 9 0.42 0.197 2−1-3-4
TP (mg/L) x x x x 10 0.42 0.195 3-4−1-2
NO3 -N (mg/L) x x x x
NH4 -N (mg/L) x x x x
(i)
In Eq. (3), Y (i) and Yt are the i-th predicted and target outputs sensors and gauges installed in the system. All sensors and gauges
respectively over N data. were cleaned and calibrated weekly.
The training phase is operated with a precise learning algo-
rithm, the most common one being the Levenberg-Marquardt (LM)
algorithm, used for various ANN applications. Generally, a part of 3. Results and discussion
the input database is used to stop the learning phase when the
minimum MSE over this part of the database has reached a local 3.1. Selection of the input parameters
minimum, even if the MSE over the training dataset keeps decreas-
ing. This part forms the validation dataset and is mainly used to The different tests could not be performed by studying only one
avoid overfitting of the network [35]. Another part of the dataset ANN model each time. The performances of the trained ANN indeed
is used to test the network after the training phase and evaluate highly depend on the initial values of weights and biases which are
its accuracy. The performances of the ANN can be evaluated using randomly defined at the first step of the training phase. The division
the root-MSE (RMSE) along with the coefficient of determination of the database operated by the fitnet function has also an impact on
R2 (Eq. (4)) and the relative error (Eq. (5)). the final values of weights and biases. To illustrate this characteris-
tic, 10 successive correlation tests for TMP were carried out using 4
i=1 (i)
2
variables as inputs: the permeate flux (J), COD concentration in the
1− N
Yt − Y (i)
R2 = MBR (CODmbr ), MLSS concentration in the MBR (MLSSmbr ) and the
i=1 2 (4)
(i) OLR calculated from CODin and J, associated with numbers 1, 2, 3
Yt − Ŷ
N and 4 respectively. Table 4 presents the values of R2 and RMSE for
these tests. The order of importance of the variables according to
(i)
Yt − Y (i) the Garson’s algorithm, developed by Garson in 1991 [37], are also
Error (%) = | | × 100 (5)
(i) indicated. This algorithm allows to determine the relative impor-
Yt
tance of each input variable of the ANN by partitioning the neural
In Eq. (4), Ŷ is the average of the predicted outputs over the N data. network connection weights [38].
(iii) Model testing: The actual values of the inputs in the testing Despite a similar ANN architecture, each test led to different
datasets are compared to predicted values by the ANN models, to performances. R2 ranged from 0.28 to 0.55 with a mean of 0.413
evaluate the models performance. In this study, the whole exper- and a standard deviation of 0.0864. For these 10 successive tests,
imental database was adapted for each test, taking into account the order of importance of the variables was never the same, despite
the parameters used as the input variables of the ANN and their that several tests led to similar performances such as tests N◦ 2 and
frequency of measure, so that each input matched its correspond- 5 and tests N◦ 9 and 10. For this last case, the importance of each
ing output. Then the input database was divided in a training, a variable for one test was even at the opposite of the one for the
validation and a test dataset with ratios of 0.70, 0.15 and 0.15 other test. Thus the order of importance does not seem to be linked
respectively for the learning process. The whole database was dis- to the correlation efficiency of the ANN. The Garson’s algorithm
tributed randomly to these 3 datasets. Table 3 indicates the input was therefore not considered to study the influence of the input
variables which were available to use with the ANN predicting the variables on the prediction of TMP.
TMP evolution in the AO-MBR. For the next section, the ANN model selected for each test on
different sets of input variables was chosen to present a coefficient
2.3. Analysis methods of determination close to the mean of the coefficients of determi-
nation obtained over 20 successive ANNs trained using the same
Samples for the analyses of soluble constituents were cen- inputs. Each parameter presented in Table 3 was studied sepa-
trifuged at 5000 rpm for 5 min, and the supernatant then filtered rately and the procedure was the same for each test: a first ANN
through a membrane with a pore size of 0.45 mm (GD/X PVDF, was trained using the whole variables whose concentrations were
Whatman). MLSS, mixed liquor volatile suspended solids (MLVSS), measured in the AO-MBR, then different tests were carried out by
COD, pH and nitrate were determined in accordance with Standard considering several combinations of these same variables in order
Methods [36]. TN and TP were determined using a Hach spec- to determine which one was the most relevant for TMP prediction.
trophotometer (Model: DR-2800). The ammonia concentrations in Finally, the last step of this part of the study consisted in combin-
the influent and effluent were measured using an ion-selective ing the selected variables together in order to obtain an ANN model
electrode (Thermo Orion, Model 95-12). The cross flow velocity of which presents a high potential to predict TMP in the AO-MBR. In
uplifting air was measured with an electromagnetic flow veloc- the next section, the different parts of the AO-MBR were referenced
ity meter (ACM-250-DL, Alec Electronics, Japan). Other operational as –in for the influent, -deg for the degassing tank, -an for the anoxic
parameters such as permeate flow rate and TMP were monitored by tank, -mbr for the MBR and –eff for the effluent.
52 F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58
3.2. Correlation of pH and alkalinity input variables on TMP input variables (R2 = 0.673). But according to the tests, only CODan
prediction and CODmbr can be considered as input variables because the result-
ing performances are overall similar to those obtained with the best
During the experiment, pHin , pHan , pHmbr and pHeff have been correlation model, with R2 a bit lower (0.659 against 0.673) but a
monitored. A test was carried out to assess the relative influence of RMSE slightly lower too (0.151 cmHg against 0.153 cmHg) (Table
pH on TMP prediction. Fig. 3 shows the correlation performances S2). It can be inferred from the previous tests that COD concentra-
obtained by considering the 4 pH input variables for the ANN tions in the anoxic tank and in the MBR have much more impact on
(R2 = 0.427 and RMSE = 0.194cmHg). After using the ANN with com- TMP prediction than pH, Alk and MLSS ones.
binations of the pH variables, pHan and pHmbr are the most influent
pH variables for TMP prediction as the ANN reached the highest
performances by considering only these 2 variables as inputs, with
R2 = 0.431 and RMSE = 0.203cmHg (Table S2). 3.5. Correlation of TN, NO3-N, NH4-N input variables on TMP
The alkalinity in each part of the AO-MBR (in the influent, in the prediction
anoxic tank, in the MBR and in the effluent) has been measured
along the experiment. A test using these concentrations (measured Removal of TN from the influent represents a major issue in
as mg CaCO3 /L) as input variables of the ANN has been carried wastewater treatment. During the experiment, the TN amounts
out, the results being shown on Fig. 4. The performances of the were monitored in the AO-MBR. A first correlation test was car-
ANN reached R2 = 0.617 and RMSE = 0.158cmHg. The best correla- ried out using the 4 TN variables as inputs. The correlation results
tion results were reached considering Alkan , Alkmbr and Alkeff as are shown on Fig. 7. This test led to the best correlation results
input variables, with R2 = 0.645 and RMSE = 0.152cmHg. However, until now, with R2 = 0.820 and RMSE = 0.109cmHg. This means that
the performances reached with Alkmbr and Alkeff only were almost TN measurements in the AO-MBR present a high potential to pre-
similar (R2 = 0.626 and RMSE = 0.159cmHg) (Table S2). This combi- dict TMP evolution using ANNs. The performances reached by using
nation can be selected as it allows to get rid of one variable, i.e. of one TNin and TNeff as input variables (R2 = 0.792 and RMSE = 0.118cmHg)
series of measurements during the operation. The influence of the were practically similar to those obtained by using the 4 TN input
DO variables was quickly analysed, as the performances were very variables. These results showed the potential of these variables for
low for all the tests. They reached R2 = 0.162 and RMSE = 0.239cmHg TMP prediction and highlighted the link between membrane foul-
when the 3 DO variables where taken as input variables (Table S2). ing and the TN removal rate in the AO-MBR. The TN removal rate
DO used as input has no impact on membrane fouling. These results increased along with the TMP during the experiment from 47% to
were predictable because the amounts of DO in the MBR and in the 66% approximately (Figure S2). However, a transitory phase has
anoxic tank were controlled over the whole operation, and thus to be noticed during the first 40 days, most likely due to the period
were independent from the membrane fouling phenomenon. needed to achieve the normal operating conditions for the AO-MBR.
Thus, as a matter of fact, membrane fouling tends to improve TN
3.3. Correlation of MLSS input variables on TMP prediction removal in the AO-MBR. In a second phase, the input variables TNin
and TNan on one hand and TNin and TNmbr in the other hand have
Then MLSS concentrations have been used as input vari- been considered to study the influence of the TN removal rate in
ables for TMP prediction. The input database was composed the anoxic tank and in the MBR respectively on TMP prediction. The
of MLSSin , MLSSdeg , MLSSan , MLSSmbr and MLSSeff . The predic- performances reached by the ANN were not satisfying in both cases,
tion results are shown on Fig. 5. The correlation performances with R2 close to 0.60. Thus the TN removal rates in the anoxic tank
when these variables were considered were very poor (R2 = 0.163 and in the MBR are not reliable variables to be used for the ANN.
and RMSE = 0.238cmHg). After having gathered the performances As well as TN, NO3 -N and NH4 -N concentrations have been mea-
obtained for several combinations of MLSS input variables, MLSS sured in the influent, in the anoxic tank, in the MBR and in the
concentrations in the AO-MBR seemed not to be suitable for TMP effluent. Four NO3 -N variables were first considered and taken as
prediction, as the values of R2 stayed very low, under 0.20. Nev- inputs of the ANN to predict TMP. The results of this test are shown
ertheless, MLSSan and MLSSmbr taken as the two input variables on Fig. 8. The correlation performances reached R2 = 0.626 and
presented the best correlation results. Additionally, when MLVSS RMSE = 0.165cmHg. The amount of nitrate production in each part of
concentrations in the anoxic tank and in the MBR were added the AO-MBR seemed to have a mitigated influence on TMP predic-
to MLSSan and MLSSmbr as input variables, correlation perfor- tion. In other terms, membrane fouling did slightly affect nitrate
mances reached R2 = 0.210 and RMSE = 0.228cmHg and were thus production throughout the whole treatment. Indeed, R2 did not
slightly better than those reached without considering MLVSS. But exceed 0.673 among all the tests considering several combinations
the improvement is not significant enough to definitely consider of variables. These performances were reached using Nitratembr
MLVSS as input variables of the ANN. and Nitrateeff as input variables. The concentrations of nitrate in
the MBR and in the effluent decreased from 18.0 mg/L at the begin-
3.4. Correlation of COD input variables on TMP prediction ning of the operation to around 6.0 mg/L until 180 days (Figure S3).
Thus nitrification in the MBR seemed to be limited by the increase in
The importance of COD concentrations as input variables for membrane fouling. However, no link can be established between
TMP prediction was assessed. The input database was composed the nitrification rate from the MBR to the effluent and the TMP
of CODin , CODdeg , CODan , CODmbr and CODeff . The results of the test because the two corresponding concentrations were practically the
are shown on Fig. 6. As seen on Fig. 6, when the five COD variables same and evolved similarly.
were considered, the correlation performances were R2 = 0.709 and In a second time, the 4 variables related to NH4 -N concentrations
RMSE = 0.143cmHg. Knowing CODin and CODeff is not sufficient to in the AO-MBR have been used as input variables of the ANN. The
predict TMP evolution (R2 = 0.165). In other words, membrane foul- results of this test are represented on Fig. 9. The performances of
ing which is directly linked to the value of TMP has no impact this model were poor and lower than those obtained with the NO3 -
on the COD removal rate. If CODmbr is added to the two previous N variables. Indeed, the ANN gave R2 = 0.450 and RMSE = 0.201cmHg.
COD input variables, the prediction of TMP is improved but stays Moreover, the performances obtained with several combinations
very limited (R2 = 0.409). The best correlation performances were of NH4 -N variables did not exceed R2 = 0.404. Thus, NH4 -N con-
obtained by using an ANN taking CODdeg , CODan and CODmbr as centrations throughout the whole treatment are not likely to be
F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58 53
Fig. 3. Correlation plot (A) and evolution (B) of the target and the output TMP for pH input variables.
Fig. 4. Correlation plot (A) and evolution (B) of the target and the output TMP for alkalinity input variables.
Fig. 5. Correlation plot (A) and evolution (B) of the target and the output TMP for MLSS input variables.
54 F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58
Fig. 6. Correlation plot (A) and evolution (B) of the target and the output TMP for COD input variables.
Fig. 7. Correlation plot (A) and evolution (B) of the target and the output TMP for TN input variables.
Fig. 8. Correlation plot (A) and evolution (B) of the target and the output TMP for NO3 -N input variables.
F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58 55
Fig. 9. Correlation plot (A) and evolution (B) of the target and the output TMP for NH4 -N input variables.
Fig. 10. Correlation plot (A) and evolution (B) of the target and the output TMP for TP input variables.
Fig. 11. Correlation plot (A) and evolution (B) of the target and the output TMP obtained with the ANN taking the operational time as input variable.
Selected variables R2 Class The final step for the selection of input variables of the ANN
CODan – CODmbr 0.659 Medium predicting the TMP consisted in combining the different couples of
MLSSan – MLSSmbr 0.196 Weak variables gathered in Table 6 together and select the optimal com-
pHan – pHmbr 0.431 Weak bination, i.e. the combination which led to the best performances
TNin – TNeff 0.792 Strong after training the ANN. Only the variables classified as medium and
TPin – TPan 0.817 Strong
strong were considered, the weak variables being inappropriate for
Nitratembr – Nitrateeff 0.673 Medium
Ammoniain – Ammoniaan 0.404 Weak TMP prediction. Table 7 presents a synthesis of the results obtained
Alkmbr – Alkeff 0.626 Medium from this study. According to the results obtained, the most relevant
DOin – DOmbr 0.159 Weak combination of variables to use for TMP prediction should be TNin ,
TNeff , Alkmbr and Alkeff , whose performances reached R2 = 0.875 and
RMSE = 0.091cmHg. However, it has been decided to let the Alk vari-
ables apart because the input database used when Alk variables
were considered was constituted of 46 data against 69 data for the
other variables (Table 6). Thus if for a couple of variables, R2 < 0.60,
other variables database. This is due to less frequent measures of Alk
it could be considered that it was too low to have any potential, even
in the AO-MBR compared to the other parameters. Following this
coupled to other variables. Then if for a single couple of variables,
choice, two combinations of variables presented equivalent perfor-
R2 > 0.75, then they present a high potential for the final perfor-
mances: TN – TP – COD and TN – TP – Nitrate. The TN – TP – COD
mances when coupled to other variables. For the medium potential,
model reached a higher R2 than the TN – TP – Nitrate one (0.855
it was proposed to take the range between weak and strong (tran-
against 0.850) but its RMSE was higher too (0.101 cmHg against
sition range). It was admited that the choice of the values (0.6 and
0.099 cmHg). Eventually, the TN – TP – Nitrate model was selected
0.75) was a bit empirical.
F. Schmitt et al. / Biochemical Engineering Journal 133 (2018) 47–58 57
[27] S.A. Mirbagheri, M. Bagheri, Z. Bagheri, A.M. Kamarkhani, Evaluation and [33] F. Schmitt, K.-U. Do, Prediction of membrane fouling using artificial neural
prediction of membrane fouling in a submerged membrane bioreactor with networks for wastewater treated by membrane bioreactor technologies:
simultaneous upward and downward aeration using artificial neural bottlenecks and possibilities, Environ. Sci. Pollut. Res. 24 (2017) 22885–22913.
network-genetic algorithm, Process Saf. Environ. Prot. 96 (2015) 111–124. [34] O. Giustolisi, D. Laucelli, Improving generalization of artificial neural
[28] M. Dornier, M. Decloux, G. Trystram, A. Lebert, Dynamic modeling of crossflow networks in rainfall–runoff modelling, Hydrol. Sci. J. 50 (2005) 439–457.
microfiltration using neural networks, J. Membr. Sci. 98 (1995) 263–273. [35] A.P. Piotrowski, J.J. Napiorkowski, A comparison of methods to avoid
[29] S. Chellam, Artificial neural network model for transient crossflow overfitting in neural networks training in the case of catchment runoff
microfiltration of polydispersed suspensions, J. Membr. Sci. 258 (2005) 35–42. modelling, J. Hydrol. 476 (2013) 97–111.
[30] Q.-F. Liu, S.-H. Kim, Evaluation of membrane fouling models based on [36] APHA, Standard Methods for the Examination of Water and Wastewater, 21st
bench-scale experiments: a comparison between constant flowrate blocking ed., American Water Works Association, Water Pollution and Control
laws and artificial neural network (ANNs) model, J. Membr. Sci. 310 (2008) Federation, Washington DC, USA, 2005.
393–401. [37] G.D. Garson, Interpreting neural-network connection weights, AI Expert 6
[31] Q.-F. Liu, S.-H. Kim, S. Lee, Prediction of microfiltration membrane fouling (1991) 46–51.
using artificial neural network models, Sep. Purif. Technol. 70 (2009) 96–102. [38] J.D. Olden, D.A. Jackson, Illuminating the black box: a randomization approach
[32] B.K. Nandi, A. Moparthi, R. Uppaluri, M.K. Purkait, Treatment of oily for understanding variable contributions in artificial neural networks, Ecol.
wastewater using low cost ceramic membrane: comparative assessment of Modell. 154 (2002) 135–150.
pore blocking and artificial neural network models, Chem. Eng. Res. Des. 88
(2010) 881–892.