Prediction Of Biochemical Oxygen Demand Research Articles

En-WBF: A Novel Ensemble Learning Approach to Wastewater Quality Prediction Based on Weighted BoostForest

With the development of urbanization, the accurate prediction of effluent quality has become increasingly critical for the real-time control of wastewater treatment processes. The conventional method for measuring effluent biochemical oxygen demand (BOD) suffers from significant time delays and high equipment costs, making it less feasible for timely effluent quality assessment. To tackle this problem, we propose a novel approach called En-WBF (ensemble learning based on weighted BoostForest) to predict effluent BOD in a soft-sensing manner. Specifically, we sampled several independent subsets from the original training set by weighted bootstrap aggregation to train a series of gradient BoostTrees as the base models. Then, the predicted effluent BOD was derived by weighting the base models to produce the final prediction. Experiments on real datasets demonstrated that on the UCI dataset, the proposed En-WBF approach achieved a series of improvements, including by 28.4% in the MAE, 40.9% in the MAPE, 29.8% in the MSE, 18.2% in the RMSE, and 2.3% in the R2. On the Fangzhuang dataset, the proposed En-WBF approach achieved a series of improvements, including by 8.8% in the MAE, 9.0% in the MAPE, 12.8% in the MSE, 6.6% in the RMSE, and 1.5% in the R2. This paper contributes a cost-effective and timely solution for wastewater treatment management in real practice with a more accurate effluent BOD prediction, validating the research in the application of ensemble learning methods for environmental monitoring and management.

Prediction of Biochemical Oxygen Demand in Wastewater Treatment Plants Using Artificial Neural Network and Regression Analysis

Atık su arıtma tesislerinde su kalitesini takip edip müdahale etmek, tesislerin yönetiminde önemli bir rol oynar. Atık su arıtma tesisleri yapılırken ve işletilirken, biyolojik oksijen ihtiyacı değerlerine gereksinim duyulmaktadır. Bu değerin ölçülmesi diğer parametrelere göre daha uzun sürelerde gerçekleşmekte ve deneylerin yapılması da zahmetli ve maliyetli olmaktadır. Bu çalışmada biyolojik oksijen değerinin, atık su arıtma tesislerinde kolayca ölçülebilen diğer parametreler aracılığıyla yapay sinir ağları ve çoklu regresyon analizi teknikleriyle tahmin edilmesi amaçlanmıştır. Çalışmada kullanılan ölçüm sonuçları 2021-2022 yılları arasında Van iline ait bir atık su arıtma tesisinde ölçülen verileri kapsamaktadır. Kullanılan tahmin girdi parametreleri pH, elektriksel iletkenlik, sıcaklık, çözünmüş oksijen, kimyasal oksijen ihtiyacı, askıda katı madde, toplam azot ve toplam fosfor değerleri bağımsız değişken ve biyolojik oksijen değeri ise bağımlı değişken olarak seçilmiştir. Yapılan tahminlerde yapay sinir ağı modeli için MAPE değeri %0.12, MAD değeri 0.04, R değeri %99.83 ve R2 değeri %99.68 olarak elde edilmiştir. Aynı şekilde çoklu regresyon analizi yöntemi ile BOİ tahmin modelinde MAPE değeri %0.68, MAD değeri 0.06, R değeri %96.40 ve R2 değeri %92.92 olarak bulunmuştur. Çalışmada elde edilen bulgular biyolojik oksijen değerinin kolayca ölçülebilen parametreler yardımıyla ileri beslemeli yapay sinir ağları ve doğrusal çoklu regresyon analizi teknikleri ile oluşturulmuş olan modeller kullanılarak tahmin edilmesi mümkündür. Her iki model karşılaştırıldığında ise yapay sinir ağları ile geliştirilmiş olan modelin çoklu regresyon analizi ile geliştirilmiş olan modele göre daha iyi performans sergilediği tespit edilmiştir.

Modelling biochemical oxygen demand in a large inland aquaculture zone of India: Implications and insights

Water quality surveillance is tough, and a specific timely management is necessary for the inland aquaculture ponds and ecology as well. Real time quality monitoring involves the study of numerous parameters includes physical (turbidity, temperature, and specific conductivity), chemical (pH, calcium, manganese, chlorides, iron, biochemical oxygen demand), and biological (bacteria and algae). It is also crucial to recognize the inter-dependence among the parameters. Alternatively, these relationships can be predicted with statistical and numerical modelling. Organic strength parameter 5-day biochemical oxygen demand (BOD) is a significant parameter to evaluate since its impact is very high on the quality of water, aquatic life, and other biological concerns. This study focuses on the prediction of BOD using six traditional and four boosting algorithms considering ten input physicochemical attributes. The attributes were fine-tuned for highly precise predictions by removing extreme values from the data set using data outlier treatment. The prediction results are compared using performance metrics such as coefficient of determination (R2), root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE). The findings revealed that boosting algorithms outperform the results of traditional models with the highest prediction accuracy. Among the boosting algorithms, eXtreme Gradient Boosting algorithm (XGBM) is found highly appropriate for the inland aquaculture waters with R2 = 0.95, RMSE = 0.31, MSE = 0.09, MAE = 0.1. Finally, this study provides a systematic evaluation of the BOD in the aquaculture waters and has a significant contribution to water management and eco-balance.

Prediction of biochemical oxygen demand with genetic algorithm-based support vector regression

Abstract Five-day biochemical oxygen demand (BOD5) is a vital wastewater contamination strength indicator. The process of measuring BOD5 is to measure the mass of molecular oxygen consumed in 1 L of water at 20 °C over a 5-day incubation period. It is a time-consuming process and often too late for water management agencies to make a timely reaction if the result of measurement shows a water body is seriously polluted. Biosensors can simplify the process of BOD5 measurement; however, the measurement results often deviate significantly from the measured BOD5 values. The main aim of this research is to identify a machine learning model, which could predict BOD5 value from historical data and make it easier to detect water pollution in advance and timely adopt treatment measures. Three machine learning techniques, linear regression, support vector regression (SVR) and multi-layer perceptron (MLP) and two optimization processes have been studied in this research. Four main steps, preprocessing (one-time only), model training, model evaluation (testing) and analysis have been implemented in the experiments. With three feature selection strategies, the results of the experiment showed that SVR with genetic algorithm (GA) optimizer achieved the best performance with R2 of 0.694 and the lowest MAE of 0.109.

Modelling dissolved oxygen and biochemical oxygen demand using data-driven techniques

Precise quantification of Biochemical oxygen demand (BOD) and Dissolved oxygen (DO) are critically important for water quality assessment as well as for development of various management policies. To calculate BOD and DO for any water sample, standard technique Winkler-Azide method is used which is cumbersome and prone to measurement error. Therefore, there is a need to device alternate Data Driven Technique (DDT). In the present study, three different DDT: Artificial Neural Network (ANN), Multi Gene Genetic Programming (MG-GP) and M5 Model Tree (M5T) have been used for DO as well as BOD prediction for 3 separate stretches of Mula-Mutha River situated in Pune, India. Additionally, attempt has been made to predict BOD using modelled DO; which shows possibility of using modelled parameter in development of another model. Performance of the models was assessed through, root mean square error (RMSE); mean absolute relative error (MARE) and coefficient of correlation (R). Results based on 3 stations indicate that ANN and MGGP both outperformed with R above 0.85 and RMSE below 1 mg/L for 2 stations out of 3. MGGP and M5T can grasp the influence parameter which can be seen from the input frequency distribution in MGGP and coefficient of input parameters in M5T.

Prediction of BOD Concentration in Wastewater Treatment Process Using a Modular Neural Network in Combination with the Weather Condition

Since weather has a huge impact on the wastewater treatment process (WWTP), the prediction accuracy for the Biochemical Oxygen Demand (BOD) concentration in WWTP would degenerate if using only one single artificial neural network as the model for soft measurement method. Aiming to solve this problem, the present study proposes a novel hybrid scheme using a modular neural network (MNN) combining with the factor of weather condition. First, discriminative features among different weather groups are selected to ensure a high accuracy for sample clustering based on weather conditions. Second, the samples are clustered based on a density-based clustering algorithm using the discriminative features. Third, the clustered samples are input to each module in MNN, with the auxiliary variables correlated with BOD prediction input to the corresponding model. Finally, a constructive radial basis function neural network with the error-correction algorithm is used as the model for each subnetwork to predict BOD concentration. The proposed scheme is evaluated on a standard wastewater treatment platform—Benchmark Simulation Model 1 (BSM1). Experimental results demonstrate the performance improvement of the proposed scheme on the prediction accuracy for BOD concentration in WWTP. Besides, the training time is shortened and the network structure is compact.

Biochemical oxygen demand prediction: development of hybrid wavelet-random forest and M5 model tree approach using feature selection algorithms.

Supplying adequate water to individuals and maintaining water supplies to support human life, particularly to rapidly urbanizing communities, are of paramount importance in the development of urban areas in each country worldwide. In turn, maintaining water resource quality and avoiding permanent damage as a consequence of environmental pollution and unsustainable off-take from sources such as rivers and aquifers should be considered as important as the water supply quantity. In this study, random forest (RF) and M5 model tree (M5) models were used to predict water biochemical oxygen demand (BOD). Having decomposed the input variables by wavelet transform, based on the feature selection algorithms (FS) (relief (RA), correlation (CA), principal component analysis (PCA), and ant colony optimization (ACO) algorithms), the important components were recognized and inserted into the RF and M5 models. The proposed approach was applied to Karun River in Ahvaz station on a monthly basis from 2006 to 2018. The results showed that the RF model had better performance with R = 0.872, MAE = 0.0312, and RMSE = 0.0332 values for the variable of BOD compared with the M5 model with R = 0.751, MAE = 0.0377, and RMSE = 0.0468 values. In addition, comparing RF and hybrid models, the purposed hybrid models were considered as viable options to improve the prediction accuracy of BOD. The findings also showed that, among the hybrid models, the WRF-PCA model with R = 0.927, MAE = 0.0198, and RMSE = 0.0241 values was the best model for the prediction of BOD values.

Assessing the biochemical oxygen demand using neural networks and ensemble tree approaches in South Korea

The biochemical oxygen demand (BOD), one of widely utilized variables for water quality assessment, is metric for the ecological division in rivers. Since the traditional approach to predict BOD is time-consuming and inaccurate due to inconstancies in microbial multiplicity, alternative methods have been recommended for more accurate prediction of BOD. This study investigated the capability of a novel deep learning-based model, Deep Echo State Network (Deep ESN), for predicting BOD, based on various water quality variables, at Gongreung and Gyeongan stations, South Korea. The model was compared with the Extreme Learning Machine (ELM) and two ensemble tree models comprising the Gradient Boosting Regression Tree (GBRT) and Random Forests (RF).Diverse water quality variables (i.e., BOD, potential of Hydrogen (pH), electrical conductivity (EC), dissolved oxygen (DO), water temperature (WT), chemical oxygen demand (COD), suspended solids (SS), total nitrogen (T-N), and total phosphorus (T-P)) were utilized for developing the Deep ESN, ELM, GBRT, and RF with five input combinations (i.e., Categories 1–5). These models were evaluated by root mean square error (RMSE), Nash-Sutcliffe efficiency (NSE), coefficient of determination (R2), and correlation coefficient (R). Overall evaluations suggested that the Deep ESN5 model provided the most reliable predictions of BOD among all the models at both stations.

Improved Water Quality Prediction with Hybrid Wavelet-Genetic Programming Model and Shannon Entropy

Prediction of biochemical oxygen demand (BOD) as the main pollution indicators of organic pollution in freshwater resources is necessary. In the present work, a hybrid wavelet-genetic programming (WGP) method was implemented to improve prediction of BOD. The Shannon entropy was used to identify the optimal input combinations of WGP. In addition, an investigation was done to find which functions of wavelet and decomposition levels have better results in conjunction with genetic programming (GP). For comparison of WGP efficiency, five machine learning methods consisting of WANN (wavelet-artificial neural network), ANN (artificial neural network), GP, DT (decision tree) and BN (Bayesian network) were considered. Experiments on wavelet-processed data revealed that the best results were obtained when the models WGP and WANN were calibrated at three levels of decomposition using the Dmey mother wavelet function. The WGP model created rational forecasts for the peak BOD values. The results show that the use of Shannon entropy is suitable for determining the optimal composition of inputs to machine learning methods. Comparison of the results indicate that the WGP model is superior to the GP, ANN, DT, BN and WANN models based on data from the Varian Hotel and Dam Input stations.

Biochemical oxygen demand prediction in wastewater treatment plant by using different regression analysis models

Biochemical oxygen demand prediction in wastewater treatment plant by using different regression analysis models

Oxygen uptake prediction in rivers and streams: A stochastic approach

Dissolved oxygen fluctuations in a river over a short period of time were assumed to be caused by the microbial growth dynamics, and a stochastic model was built for oxygen uptake. As a case study, biochemical oxygen demand (BOD) was measured in water from the Ura River, Oita River, and Otozu River flowing through the urban district of Oita, Japan. Water samples were taken from each river and partitioned into BOD bottles. BOD was measured in five of these bottles on each of nine days. The experimental results show that the average daily BOD decreased exponentially as expressed by the Streeter-Phelps equation. A wide range of the measured five daily BOD-values was expressed by the difference between the maximum and minimum BOD-values on each day for each river. After the first few days the range became smaller. The proposed stochastic model describes the observed experimental fluctuation of BOD over time. Eighty to ninety percent of the experimental BOD plots are within the 80% probability range given by the model. The uncertainty of BOD prediction can be expressed by the error which is the non-dimensional ratio of the range to the median. Modeled and experimental results reveal that the error is about 0.5–1 (50–100% of expected value) after a few days. This suggests that the BOD predicted by deterministic water quality models can include uncertainty, i.e. the actual BOD can be a quarter or double of the simulated value, for the time scale of a few days. For a longer period, e.g. more than a week, the error can become even more significant.

Uncertainty assessment of the multilayer perceptron (MLP) neural network model with implementation of the novel hybrid MLP-FFA method for prediction of biochemical oxygen demand and dissolved oxygen: a case study of Langat River

Accurate prediction of the chemical constituents in major river systems is a necessary task for water quality management, aquatic life well-being and the overall healthcare planning of river systems. In this study, the capability of a newly proposed hybrid forecasting model based on the firefly algorithm (FFA) as a metaheuristic optimizer, integrated with the multilayer perceptron (MLP-FFA), is investigated for the prediction of monthly water quality in Langat River basin, Malaysia. The predictive ability of the MLP-FFA model is assessed against the MLP-based model. To validate the proposed MLP-FFA model, monthly water quality data over a 10-year duration (2001–2010) for two different hydrological stations (1L04 and 1L05) provided by the Irrigation and Drainage Ministry of Malaysia are used to predict the biochemical oxygen demand (BOD) and dissolved oxygen (DO). The input variables are the chemical oxygen demand (COD), total phosphate (PO4), total solids, potassium (K), sodium (Na), chloride (Cl), electrical conductivity (EC), pH and ammonia nitrogen (NH4-N). The proposed hybrid model is then evaluated in accordance with statistical metrics such as the correlation coefficient (r), root-mean-square error, % root-mean-square error and Willmott’s index of agreement. Analysis of the results shows that MLP-FFA outperforms the equivalent MLP model. Also, in this research, the uncertainty of a MLP neural network model is analyzed in relation to the predictive ability of the MLP model. To assess the uncertainties within the MLP model, the percentage of observed data bracketed by 95 percent predicted uncertainties (95PPU) and the band width of 95 percent confidence intervals (d-factors) are selected. The effect of input variables on BOD and DO prediction is also investigated through sensitivity analysis. The obtained values bracketed by 95PPU show about 77.7%, 72.2% of data for BOD and 72.2%, 91.6% of data for DO related to the 1L04 and 1L05 stations, respectively. The d-factors have a value of 1.648, 2.269 for BOD and 1.892, 3.480 for DO related to the 1L04 and 1L05 stations, respectively. Based on the values in both stations for the 95PPU and d-factor, it is concluded that the neural network model has an acceptably low degree of uncertainty applied for BOD and DO simulations. The findings of this study can have important implications for error assessment in artificial intelligence-based predictive models applied for water resources management and the assessment of the overall health in major river systems.

Prediction of biochemical oxygen demand at the upstream catchment of a reservoir using adaptive neuro fuzzy inference system

The aim of this study is to examine the potential of adaptive neuro fuzzy inference system (ANFIS) to estimate biochemical oxygen demand (BOD). To illustrate the applicability of ANFIS method, the upstream catchment of Feitsui Reservoir in Taiwan is chosen as the case study area. The appropriate input variables used to develop the ANFIS models are determined based on the t-test. The results obtained by ANFIS are compared with those by multiple linear regression (MLR) and artificial neural networks (ANNs). Simulated results show that the identified ANFIS model is superior to the traditional MLR and nonlinear ANNs models in terms of the performance evaluated by the Pearson coefficient of correlation, the root mean square error, the mean absolute percentage, and the mean absolute error. These results indicate that ANFIS models are more suitable than ANNs or MLR models to predict the nonlinear relationship within the variables caused by the complexity of aquatic systems and to produce the best fit of the measured BOD concentrations. ANFIS can be seen as a powerful predictive alternative to traditional water quality modeling techniques and extended to other areas to improve the understanding of river pollution trends.

Artificial Neural Network Modelling of Biochemical Oxygen Demand and Dissolved Oxygen of Rivers: Case Study of Asa River

Water quality assessment involves the determination of a number of parameters using several analytical methods which are often tedious and time consuming. Artificial Neural Network (ANN) was used in this study to model the relationship between fifteen (15) water quality parameters used to predict other two (2) related parameters in other to reduce the burden of long experimental procedures. Water samples were collected from six (6) point and non point sources of pollution along Asa River in Ilorin during the peak of rainy season (June–Aug, 2014) and peak of dry season (Nov–Jan, 2015). Physical and chemical parameters inputted into the models include pH, turbidity, total dissolved solids, temperature, electrical conductivity, dissolved oxygen, biochemical oxygen demand, chemical oxygen demand, hardness, chloride, sulphate, phosphate, calcium, magnesium and nitrate. The output models include: biochemical oxygen demand (BOD) and dissolved oxygen (DO). The three layer feed-forward model with back-propagation multi-layer perception (MLP) models architecture of 15-9-1 for BOD and 15-13-1 for DO yielded optimal results with 9 and 13 neurons in hidden layer for BOD and DO respectively. The ANN was successfully trained and validated with 83% and 17% of the data sets respectively. Performance of the models was evaluated by statistical criteria of average error (AE) and mean square error (MSE). The correlation coefficients of ANN models for prediction of BOD and DO were 0.9525 and 0.9556 respectively. Sensitivity analysis was also carried out to identify the most significant input-output relationship. Hence, the ANNs was able to show remarkable prediction performance to predicting the BOD and DO in Asa River, Ilorin.DOI: http://dx.doi.org/10.5755/j01.erem.72.3.14120

Evaluation of multivariate linear regression and artificial neural networks in prediction of water quality parameters.

This paper examined the efficiency of multivariate linear regression (MLR) and artificial neural network (ANN) models in prediction of two major water quality parameters in a wastewater treatment plant. Biochemical oxygen demand (BOD) and chemical oxygen demand (COD) as well as indirect indicators of organic matters are representative parameters for sewer water quality. Performance of the ANN models was evaluated using coefficient of correlation (r), root mean square error (RMSE) and bias values. The computed values of BOD and COD by model, ANN method and regression analysis were in close agreement with their respective measured values. Results showed that the ANN performance model was better than the MLR model. Comparative indices of the optimized ANN with input values of temperature (T), pH, total suspended solid (TSS) and total suspended (TS) for prediction of BOD was RMSE = 25.1 mg/L, r = 0.83 and for prediction of COD was RMSE = 49.4 mg/L, r = 0.81. It was found that the ANN model could be employed successfully in estimating the BOD and COD in the inlet of wastewater biochemical treatment plants. Moreover, sensitive examination results showed that pH parameter have more effect on BOD and COD predicting to another parameters. Also, both implemented models have predicted BOD better than COD.

Prediction of BOD, COD, and Total Nitrogen Concentrations in a Typical Urban River Using a Fluorescence Excitation-Emission Matrix with PARAFAC and UV Absorption Indices

The development of a real-time monitoring tool for the estimation of water quality is essential for efficient management of river pollution in urban areas. The Gap River in Korea is a typical urban river, which is affected by the effluent of a wastewater treatment plant (WWTP) and various anthropogenic activities. In this study, fluorescence excitation-emission matrices (EEM) with parallel factor analysis (PARAFAC) and UV absorption values at 220 nm and 254 nm were applied to evaluate the estimation capabilities for biochemical oxygen demand (BOD), chemical oxygen demand (COD), and total nitrogen (TN) concentrations of the river samples. Three components were successfully identified by the PARAFAC modeling from the fluorescence EEM data, in which each fluorophore group represents microbial humic-like (C1), terrestrial humic-like organic substances (C2), and protein-like organic substances (C3), and UV absorption indices (UV220 and UV254), and the score values of the three PARAFAC components were selected as the estimation parameters for the nitrogen and the organic pollution of the river samples. Among the selected indices, UV220, C3 and C1 exhibited the highest correlation coefficients with BOD, COD, and TN concentrations, respectively. Multiple regression analysis using UV220 and C3 demonstrated the enhancement of the prediction capability for TN.

Prediction of Pollutant Concentration in Laborec River Station, Slovak Republic

A model that predicts pollutant concentrations in a river has been developed and is presented in the paper. The developed model that determines concentrations of pollutants specifically biochemical oxygen demand (BOD5) and total nitrogen (N) in water stream is based on dimensional analysis. Fundamentals of the modeling of the pollutants prediction in water stream consist in derivation of function dependency from expressed non-dimension arguments. Non-dimension arguments are stated from variables, which influence the occurrence of pollutants. Model for prediction of biochemical oxygen demand and nitrogen concentrations in water stream has been developed for Laborec River, eastern part of Slovakia. The differences between the calculated concentrations from developed model described in this paper and measured concentrations in Laborec river stations are also discussed. The developed model might be used for prediction of any pollutant concentration in river.

Prediction of the effluent from a domestic wastewater treatment plant of CASP using gray model and neural network.

When a domestic wastewater treatment plant (DWWTP) is put into operation, variations of the wastewater quantity and quality must be predicted using mathematical models to assist in operating the wastewater treatment plant such that the treated effluent will be controlled and meet discharge standards. In this study, three types of gray model (GM) including GM (1, N), GM (1, 1), and rolling GM (1, 1) were used to predict the effluent biochemical oxygen demand (BOD), chemical oxygen demand (COD), and suspended solids (SS) from the DWWTP of conventional activated sludge process. The predicted results were compared with those obtained using backpropagation neural network (BPNN). The simulation results indicated that the minimum mean absolute percentage errors of 43.79%, 16.21%, and 30.11% for BOD, COD, and SS could be achieved. The fitness was higher when using BPNN for prediction of BOD (34.77%), but it required a large quantity of data for constructing model. Contrarily, GM only required a small amount of data (at least four data) and the prediction results were analogous to those of BPNN, even lower than that of BPNN when predicting COD (16.21%) and SS (30.11%). According to the prediction, results suggested that GM could predict the domestic effluent variation when its effluent data were insufficient.

Using Synchronous Fluorescence Technique as a Water Quality Monitoring Tool for an Urban River

The development of a monitoring tool for predicting water quality and tracing pollution sources are important for the management of sustainable aquatic ecosystems in urban areas. In this study, synchronous fluorescence technique was applied to 18 sampling sites of a typical urban watershed in Korea, some of which are directly affected by the effluent from a wastewater treatment plant (WWTP), to investigate the capability of the technique for biochemical oxygen demand (BOD) prediction and source discrimination. Sampling was conducted three times at the same sites during the low flow period between October and November, 2005. Protein-like fluorescence intensities of the samples showed a positive linear relationship with the BOD values (Spearman’s rho = 0.90, p < 0.0001). The BOD prediction capability was superior to other monitoring tools such as UV absorption and conductivity measurements particularly for the upstream sites from the WWTP, which ranged from 0.0 to 5.0 mg/l as BOD. The protein-like fluorescence and a ratio of protein-like/fulvic-like fluorescence were suggested as good fluorescence signatures to discriminate different sources of dissolved organic matter (DOM). The samples collected from four different DOM source regions including upstream sites from the WWTP, downstream sites, discharge from a reservoir, and headwater were distinguished from one another by varying ranges of the two selected fluorescence signatures. Our results suggest that the synchronous fluorescence technique has the potential to be developed into a real-time water quality management tool for the comprehensive monitoring of urban rivers.

Application of steady-state and dynamic modeling for the prediction of the BOD of an aerated lagoon at a pulp and paper mill: Part II. Nonlinear approaches

Neural networks can provide effective predictive models for complex processes that are poorly described by first principle models, such as wastewater biological treatment systems. In this paper multilayer perceptron (MLP) and functional-link neural networks (FLN) are developed to predict inlet and outlet biochemical oxygen demand (BOD) of an aerated lagoon operated by International Paper of Brazil. In Part I, predictive models for both inlet and outlet BOD for the aerated lagoon were developed using linear multivariate regression techniques. For the current case study, MLP networks are the best choice for the prediction models. When only a relatively small number of samples is available, substantial improvement in inlet and outlet BOD prediction is shown for both FLN and MLP modeling using a reduced input variable set that was generated using partial least squares (PLS). Thus, this paper provides a novel approach for developing PLS–FLN model structures.