Optimization of artificial neural network models through genetic algorithms for surface ozone concentration forecasting

Introduction

This study proposes three methodologies to define artificial neural network models through genetic algorithms (GAs) to predict the next-day hourly average surface ozone (O₃) concentrations. GAs were applied to define the activation function in hidden layer and the number of hidden neurons.

Methods

Two of the methodologies define threshold models, which assume that the behaviour of the dependent variable (O₃ concentrations) changes when it enters in a different regime (two and four regimes were considered in this study). The change from one regime to another depends on a specific value (threshold value) of an explanatory variable (threshold variable), which is also defined by GAs. The predictor variables were the hourly average concentrations of carbon monoxide (CO), nitrogen oxide, nitrogen dioxide (NO₂), and O₃ (recorded in the previous day at an urban site with traffic influence) and also meteorological data (hourly averages of temperature, solar radiation, relative humidity and wind speed). The study was performed for the period from May to August 2004.

Results and discussion

Several models were achieved and only the best model of each methodology was analysed. In threshold models, the variables selected by GAs to define the O₃ regimes were temperature, CO and NO₂ concentrations, due to their importance in O₃ chemistry in an urban atmosphere.

Conclusion

In the prediction of O₃ concentrations, the threshold model that considers two regimes was the one that fitted the data most efficiently.     

Methods of characterizing the distribution of exhaust emissions from light-duty, gasoline-powered motor vehicles in the U.S. fleet

Mobile sources significantly contribute to ambient concentrations of airborne particulate matter (PM). Source apportionment studies for PM10 (PM < or = 10 microm in aerodynamic diameter) and PM2.5 (PM < or = 2.5 microm in aerodynamic diameter) indicate that mobile sources can be responsible for over half of the ambient PM measured in an urban area. Recent source apportionment studies attempted to differentiate between contributions from gasoline and diesel motor vehicle combustion. Several source apportionment studies conducted in the United States suggested that gasoline combustion from mobile sources contributed more to ambient PM than diesel combustion. However, existing emission inventories for the United States indicated that diesels contribute more than gasoline vehicles to ambient PM concentrations. A comprehensive testing program was initiated in the Kansas City metropolitan area to measure PM emissions in the light-duty, gasoline-powered, on-road mobile source fleet to provide data for PM inventory and emissions modeling. The vehicle recruitment design produced a sample that could represent the regional fleet, and by extension, the national fleet. All vehicles were recruited from a stratified sample on the basis of vehicle class (car, truck) and model-year group. The pool of available vehicles was drawn primarily from a sample of vehicle owners designed to represent the selected demographic and geographic characteristics of the Kansas City population. Emissions testing utilized a portable, light-duty chassis dynamometer with vehicles tested using the LA-92 driving cycle, on-board emissions measurement systems, and remote sensing devices. Particulate mass emissions were the focus of the study, with continuous and integrated samples collected. In addition, sample analyses included criteria gases (carbon monoxide, carbon dioxide, nitric oxide/nitrogen dioxide, hydrocarbons), air toxics (speciated volatile organic compounds), and PM constituents (elemental/organic carbon, metals, semi-volatile organic compounds). Results indicated that PM emissions from the in-use fleet varied by up to 3 orders of magnitude, with emissions generally increasing for older model-year vehicles. The study also identified a strong influence of ambient temperature on vehicle PM mass emissions, with rates increasing with decreasing temperatures.     

Inferring episodic atmospheric iron fluxes in the Western South Atlantic

Iron (Fe) and other trace elements such as Zn, Mn, Ni and Cu are known as key-factors in marine biogeochemical cycles. It is believed that ocean primary productivity blooms in iron deficient regions can be triggered by iron in aeolian dust. Up to now, scarce aerosol elemental composition, based on measurements over sea at the Western South Atlantic (WSA), exist. An association between the Patagonian semi-desert dust/Fe and chlorophyll-a variability at the Argentinean continental shelf is essentially inferred from models. We present here experimental data of Fe enriched aerosols over the WSA between latitudes 22°S–62°S, during 4 oceanographic campaigns between 2002 and 2005. These data allowed inferring the atmospheric Fe flux onto different latitudinal bands which varied from 30.4 to 1688 nmolFe m^?2 day^?1 (October 29th–November 15th, 2003); 5.83–1586 nmolFe m^?2 day^?1 (February 15th–March 6th, 2004) and 4.73–586 nmolFe m^?2 day^?1(October 21st–November 5th, 2005).     

Particulate organic compounds emitted from experimental wildland fires in a Mediterranean ecosystem

Fine (PM_2.5) and coarse (PM_2.5–10) smoke particles from controlled biomass burnings of a shrub-dominated forest in Lousã Mountain, Portugal, enabled the quantification by chromatographic techniques of several molecular tracers for the combustion of Mediterranean forest ecosystems, which could be conducive to source apportionment studies. The major organic components in the smoke samples were pyrolysates of vegetation cuticles, mainly comprising steradienes and sterol derivatives, carbohydrates from the breakdown of cellulose, aliphatic lipids from vegetation waxes and methoxyphenols from the lignin thermal degradation. Most of these compounds are chiefly found in fine particles. Polycyclic aromatic hydrocarbons (PAH) were also present as minor constituents. Anhydrosugar and PAH molecular diagnostic ratios were applied as source assignment tools. Some biomarkers are reported for the first time in biomass burning smoke.     

Evaluation of the efficiency of a Venturi scrubber in particulate matter collection smaller than 2.5 µm emitted by biomass burning

Environmental Science and Pollution Research - Energy demand has increased worldwide, and biomass burning is one of the solutions most used by industries, especially in countries that have a great...     

Elaeis guineensis-activated carbon for methylene blue removal: adsorption capacity and optimization using CCD-RSM

Environment, Development and Sustainability - The activated biochar is a carbonaceous material obtained from thermal degradation of biomass with high adsorption potential. On the other hand, the...     

Biological activities of the essential oil from the leaves of Lantana montevidensis (Spreng) Briq. in mice

Environment, Development and Sustainability - This work aimed to analyze the chemical composition and evaluate the anti-inflammatory and antinociceptive properties of the essential oil obtained...     

Evaluation of bisphenol A levels in Nigerian thermal receipts and estimation of daily dermal exposure

Environmental Science and Pollution Research - Bisphenol A (BPA) is a high production volume chemical that has wide industrial applications, especially as a color developer in thermal papers. The...     

Strategies for applying gray water effluent on ornamental sunflower crops

Environmental Science and Pollution Research - In a context of scarcity of good quality water, reuse is a mandatory practice to increase water availability, thus allowing the exploitation of more...     

Strategies for competitiveness and sustainability: Adaptation of a Brazilian subsidiary of a Swedish multinational corporation

The competitiveness, in terms of macro and micro levels of an enterprise, is often dictated by its capacity to innovate, ability to respond to the needs of the market, and resilience to challenges from competition. This paper reviews the literature on the adaptation of research and development (R&D) units in subsidiaries of multinational corporations. It focuses mainly on the impact of the technological dynamic on sustainability performance of the Brazilian subsidiary of the Swedish multinational Ericsson. Through a conceptual framework, it explores information and communication technologies (ICT) towards a transversal and comprehensive vision of levels of innovation and sustainability.