Artificial neural network onto eight bit microcontroller for Secchi depth calculation

Abstract

A comprehensive study aimed at evaluating the occurrence, significance of concentrations and spatial distribution of priority pollutants (PPs) along the Comunidad Valenciana coastal waters (Spain) was carried out in order to fulfil the European Water Framework Directive (WFD). Additionally, PPs concentrations were also analysed in the effluent of 28 WWTPs distributed along the studied area, since these infrastructures are usually considered a significant point source of many of these toxic substances.In coastal waters 36 organic pollutants of the 71 analysed, including 26 PPs (8 of them priority hazardous substances) were detected although many of them with low frequency of occurrence. Only 13 compounds, which belong to four different classes (VOCs, organochlorinated pesticides, phthalates and tributyltin compounds (TBT)) showed a frequency of occurrence above 20% in coastal waters. Toluene was the most ubiquitous pollutant (detected in 100% of the samples) with maximum concentration of 1.6 g/l. All analysed coastal waters showed the presence of PPs, being di(2-ethylhexyl)phthalate (DEHP) the most frequent one (detected in 57% of the samples) reaching concentrations up to 15 g/l. In the results obtained until now, there are four compounds that exceed the established Environmental Quality Standards (EQS): octylphenol, pentachlorobenzene, DEHP and TBT exceeded the annual average concentration (EQS-AAC), and only TBT surpassed the maximum allowable concentration (EQS-MAC).Despite its prohibition, endosulfan and hexachlorocyclohexane (HCH) were detected in 9% and 30% of the samples, respectively, and their concentrations were relatively close to their EQS-AAC value (around the 70 %). The most frequent contaminants determined in coastal waters were also present in WWTP's effluents. Generally, effluents from WWTPs exhibited higher frequency of occurrence and concentration than in coastal waters, except for DEPH.

Freeze-Damage Detection in Lemons Using Electrochemical Impedance Spectroscopy

“…Actually, this type of development [48] has already been successfully applied using simplified ANNs, [49,50] which are even simpler, computationally less demanding, quick to program, easy to use and very reliable [51]. The goal was to work with a prediction system that is more flexible, adaptive and versatile than the traditional statistical data treatment methods [52,53]. It was particularly interesting considering that the samples were natural fruits that were subject to a wide set of variables that could have generated more or less diversity among them despite belonging to the same species and the same batch (ripeness, acidity, sugar contents, size, time since they were collected, etc.).…”

Early Detection of Freeze Damage in Navelate Oranges with Electrochemical Impedance Spectroscopy

“…In order to carry out a more precise, flexible, and adaptive prediction model than those traditionally conducted by PLS [35,36], an ANN model was developed. To do so, the software Alyuda Neurointelligence 2.2 © (Alyuda Research Inc., Cupertino, CA, USA) was used.…”

An Electrochemical Impedance Spectroscopy-Based Technique to Identify and Quantify Fermentable Sugars in Pineapple Waste Valorization for Bioethanol Production

“…A commercial ANN software (Alyuda Neurointelligence 2.2 © , Alyuda Research Inc., Los Altos, CA, USA) was used throughout this study in order to create alternative, flexible and more adaptive predictive models to PLS [ 35 , 38 , 39 ]. Multi-layer feed forward neural networks and a single hidden layer ANN structure were selected and on-line back propagation training algorithms were used for fitting the network.…”

“…Similarly, several trials suggested the selection of logistic-type transfer functions for the output layer neurons and hyperbolic tangent-type functions for the hidden nodes. Random data division was used by Alyuda Neurointelligence 2.2 © in order to select the samples for training (70%), validation (15%) and test (15%) data [ 38 , 39 , 40 ]. In addition, overfitting was avoided by using proportional number of nodes in the network architecture [ 45 ], cross validation and early-stopping in the training phase, so that the difference between training and validation mean square errors was minimal.…”

“…In addition, ANNs are able to learn in their training process in order to improve themselves and find the optimal conditions to work showing high flexibility and adaptive capacity. In addition, ANNs are being used in a wide range of applications such as electronic noses [ 34 , 35 ] and tongues [ 36 , 37 ] and they are showing very interesting prediction models in several fields such as water [ 38 ], food [ 39 ] and the environment [ 40 , 41 ].…”