A comparative analysis of artificial neural networks and wavelet hybrid approaches to long-term toxic heavy metal prediction

Abstract

the occurrence of toxic metals in the aquatic environment is as caused by a variety of contaminations which makes difficulty in the concentration prediction. In this study, conventional methods of backpropagation neural network (Bpnn) and nonlinear autoregressive network with exogenous inputs (NARX) were applied as benchmark models. Explanatory variables of Fe, pH, electrical conductivity, water temperature, river flow, nitrate nitrogen, and dissolved oxygen were used as different input combinations to forecast the long-term concentrations of As, Pb, and Zn. The wavelet transformation was applied to decompose the time series data, and then was integrated with conventional methods (as WNN and WNARX). The modelling performances of the hybrid models of WNN and WNARX were compared with the conventional models. All the given models were trained, validated, and tested by an 18-year data set and demonstrated based on the simulation results of a 2-year data set. Results revealed that the given models showed general good performances for the long-term prediction of the toxic metals of As, Pb, and Zn. The wavelet transform could enhance the long-term concentration predictions. However, it is not necessarily useful for each metal prediction. Therefore, different models with different inputs should be used for different metals predictions to achieve the best predictions. Due to the nature of ubiquity, toxicity at a trace level, and hard biodegradation, elevated metals in aquatic environments are a global concern 1,2. A long-term exposure of toxic metals by the ingestion of the contaminated water and fish can cause chronic diseases 3,4. For example, Arsenic (As) destroys the redox capacity of cells, affects the normal metabolism, causes tissue damage and body disorders, and even directly causes poisoning death when ingested in small quantities 5. Lead (Pb) affects nerves, digestion, urinary, reproductive and developmental, cardiovascular, endocrine, immune, bone, and other organ systems. More serious is that Pb affects the growth and mental development of infants and young children, impairs brain function such as cognition 6. In addition, a high level of Zinc (Zn) weakens immune function, leads to iron deficiency anaemia, affects the function of the digestive system, and causes damage to blood vessels 7. Due to the special significance to water quality, As, Pb, and Zn are included in the priority pollutant list by the United States Environmental Protection Agency. Therefore, it is essential to understand the environmental behaviours of toxic metal in rivers to protect the drinking water intake. Traditionally, the toxic metals in trace levels were required to be routinely sampled and determined in the laboratory. However, there are some constraints for the environmental managers to adequately and timely receive the metal contents and respond to the metal pollution, such as (i) expense of field monitoring, (ii) staffs availability and resources, (iii) field safety issues, and (iv) large time intervals between data collection, r...

BOD5 prediction using machine learning methods

“…Alvarez-Guerra et al (Chan et al 2021) studied the use of traditional statistical methods in prediction of amphipod toxicity in contaminated sediments and reported machine learning methods have achieved significantly higher prediction accuracy compared to any of the statistical methods. Li et al 2020 studied the use of nonlinear autoregressive exogenous model (NARX) in predicting concentrations of three toxic metals in the Elbe river in Europe and demonstrated it was inferior to other statistical methods. Cipullo et al (2019) presented the research work of predicting the bioavailable concentration of chemicals in soil samples with a better prediction performance using random forest (RF) method.…”

“…Among machine learning methods, artificial neural networks (ANNs) and support vector machines (SVMs) has been found widely used in prediction of chemical components. Counter-propagation artificial neural networks (CP-ANN) in (Chan et al 2021), back-propagation neural networks (BPNN) in (Li et al 2020) and deep neural network (DNN) in (Chan et al 2021) were found to have excellent performance and superior to those from traditional statistical methods. Even traditional neural networks (NNs) in (Park et al 2014;Cipullo et al 2019;Chou et al 2018) are also shown good performance in prediction.…”

“…Latest research work on optimization algorithms have been presented in (Cai et al 2021) and (Deng et al 2020). Li et al (2020) tested hybrid methods of wavelet BPNN hybrid model (WNN) and wavelet NARX hybrid model (WNARX). It was found that hybrid models produce predictions with best accuracy, with the exact better model of the two depending on the toxic metal being predicted.…”

Water, Soil and Air Pollutants’ Interaction on Mangrove Ecosystem and Corresponding Artificial Intelligence Techniques Used in Decision Support Systems - A Review

“…WNN can achieve convergence despite the divergence effect of multiple inputs. In such, the activation function retains its sensitivity to predict extreme values and display better adjustments [124]. Subsequently, wavelet decomposition poses extraction properties of the input's division signals bringing positive effect for heavy metal content prediction.…”

Overview of Smart Aquaculture System: Focusing on Applications of Machine Learning and Computer Vision

“…Supervised learning is commonly used for classification and regression, where using data as a sample after trained by machine learning model which have the same target values [21]. From the theory of machine learning as well as its advantages, there are several implements in aquaculture recently such as biomass fish detection [22], size estimates [23][24][25], weight estimates [26][27][28], count [29][30][31][32], fish recognition [33][34][35][36][37][38], age detection [39,40], sex identification [34,[41][42][43], fish species classification [44][45][46][47][48][49][50], feeding behavior [51,52], group behavior [53], abnormal behavior [54,55], univariate prediction [38,[56][57][58][59], multivariate prediction [60][61][62], with the high accuracy rate.…”