Elsevier

Sustainable Cities and Society

Volume 43, November 2018, Pages 395-405
Sustainable Cities and Society

Modeling airborne indoor and outdoor particulate matter using genetic programming

https://doi.org/10.1016/j.scs.2018.08.015Get rights and content

Highlights

  • Monitoring is done at 12 indoors and 5 outdoors locations, spread across university.

  • GP based data mining technique is implemented to identify multi-nonlinear models.

  • Outdoor PM2.5 is much higher than the permissible limits by USEPA and EEA.

  • GP based models are perfectly able to mimic the behavioral trends of outdoor PM.

  • The model predictions are very close to the measured values.

Abstract

Airborne particulate matter (PM) is considered to be an essential indicator of outdoor and indoor air quality. In this study, indoor and outdoor PM1, PM2.5, PM10 concentrations were monitored at different locations within the Tehran University campus. It is found that 10% of PM1, PM2.5 and PM10 concentrations were higher than 36.11, 52.48 and 92.13 μg/m3 for indoors respectively. Genetic programming (GP) based methodology is implemented to identify the influence of outdoor PM on the indoor PM and established significant empirical models. The best GP model is identified based on fitness measure and root mean square error. It was observed that the GP based models are perfectly able to mimic the behavioural trends of outdoor particulate matter for PM1, PM2.5 and PM10 concentrations. The model predictions are very similar to the measured values and their variation was less than ± 8%. This analysis confirms the performance of GP based data driven modeling approach to predict the relationship between the outdoor particulate matter and its influence on the indoor particulate matter concentration.

Introduction

In recent years, the indoor air quality in residential buildings and workplaces has become an increasing concern as the inhabitants spend more time indoors. Airborne particulate matter (PM) is an essential indicator of indoor and outdoor air quality. Epidemiological studies have documented a significant positive correlation between the daily mean concentration of particles (PM1, PM2.5, PM10) and increased mortality and morbidity attributable to respiratory and cardiovascular diseases (Anderson, Atkinson, Peacock, Marston, & Konstantinou, 2004; Hazarika, Srivastava, & Das, 2017). Major human health issues resulting from PM exposure include aggravation of asthma, bronchitis, and other respiratory problems, leading to increased hospital admissions and decreased life expectancy. Smaller particles have increasingly more severe impacts on human health as compared to larger particles. The international agency for research on cancer recently concluded that exposure to PM in outdoor air is carcinogenic to humans and causes lung cancer (Hamra et al., 2014; Shafaghat, Keyvanfar, Manteghi, & Lamit, 2016). Both short-term and long-term epidemiological studies have reported respiratory and cardiopulmonary effects with increased cardiopulmonary mortality (Gehring et al., 2006; Hoek, Brunekreef, Goldbohm, Fischer, & van den Brandt, 2002; Krewski et al., 2004; Le Tertre et al., 2002; McDonnell, Nishino-Ishikawa, Petersen, Chen, & Abbey, 2000; Miller et al., 2007). Ostro (2004) have reported that there is a significant association between PM10 concentrations and the medical visits by children and elderly persons for lower respiratory and upper respiratory symptoms respectively. Other environmental effects of PM are visibility reduction, acidic precipitation, and the transport of pollutants from industrial regions to remote and pristine areas. Few studies have reported that PM is possibly responsible for global climate change through their direct and indirect role in the earth's radiation balance (Levy et al., 2013; Zorpas & Skouroupatis, 2016).

Air pollution caused by industrialization is one of the major environmental challenges and different industries contribute to air pollution in different ways (Bruschweiler et al., 2012; Reisen, Bhujel, & Leonard, 2014). Motor vehicle traffic is also an important source of particulate pollution in cities of the developing world, where rapid growth, coupled with a lack of adequate transport and land use planning, may result in harmful levels of fine particles (PM2.5) in the air (Bereitschaft, 2015; Qiu et al., 2017). The inhabitants in these industrialized cities spend more than 90% of their time in indoor environments carrying various daily activities. Especially students spend approximately 60% of their daytime in their school/college/university campus and hence exposed to closed indoor pollution (An & Yu, 2018). Due to prolonged exposure, they are vulnerable to potential health hazards. Indoor concentrations of air pollutants will vary by time and geographical location. Usually high mobility in the daytime and limited or no movement in the night time is commonly observed inside a regular university campus. Within the university indoors, the main source of airborne particulate matters is due to re-suspension of particles because of mobility of occupants. Any outdoor activities that generate dust or air pollutants might penetrate indoors and remain inside the closed environment. Since students occupy in a closed and confined space over a period of several hours during their lectures/tutorials and laboratories, hence prone to be exposed in long term. Currently, very few studies are reported in open literature about indoor classroom PM concentrations and occupants' exposure patterns to PM in a university establishment (An & Yu, 2018; Manimaran & Narayana, 2018).

Any studies related to identify correlations between the indoor and outdoor PM will be a great boon for student’s society and accordingly can lead to take necessary primitive measures to minimize the effect. Since the PM data can be available from various measuring devices, efficient data mining techniques can be implemented to understand the behavioural patterns and identify the complex characteristics of the system. In general, various data mining techniques were studied to identify models ranging from algebraic to differential equations system using processed data (Dong et al., 2009; Fernandez-Camacho et al., 2015; Karri, 2011; Karri, Jayakumar, & Sahu, 2017; Jayawardene, 2012; Tzima, Karatzas, Mitkas, & Karathanasis, 2007; Rao, Srinivasan, & Venkateswarlu, 2010; Riga, Tzima, Karatzas, & Mitkas, 2009; Voukantsis et al., 2011). It was found from the literature that genetic programming (GP) based data mining technique is a powerful tool for system identification when little is known about the underlying model structure in the data (Nelles, 2013). The applications on algebraic modeling using the GP approach is extensively available (Hassani, Tjahjowidodo, & Do, 2014; Kandpal, Kalyan, & Samavedham, 2013; Nelles, 2013; Pombeiro, Machado, & Silva, 2017). However, there have been little or no reported applications of genetic programming as a system identification tool to identify the state space model especially in assessing indoor and outdoor air quality.

The main objective of this research study is to identify the influence of outdoor PM on the indoor PM. Indeed, the authors initially tried to identify and establish empirical relation representing outdoor PM as dependent variables. It was found that indoor PM has significant strong correlation with only one outdoor monitored PM concentration. This is due to the fact that the PM concentrations at outdoor locations are far apart, and due to geometrical orientations of the rooms, PM at indoor locations (which are in the vicinity) had similar (strong) correlation with only one outdoor location. This thus led to a simple straight forward linear relation. But it was observed that PM measured at a particular outdoor location is consequently affecting the indoor concentration of many indoor rooms. In this research study, it was observed that the particulate matters entering classrooms through the doors and windows had stronger influence with the outdoor monitoring location in its vicinity. So in the established state space models, the influence of outdoor PM on the respective indoor can be estimated based on the significance of the coefficients with respect to the corresponding indoor location. Hence this approach will help us to identify the indoor locations which are strongly influenced by particular outdoor pollution. Therefore, primary objective of this study is to model indoor and outdoor PM concentrations using the genetic programming-based framework and identify the correlation between them.

Section snippets

Materials and methods

Airborne particulate matters with size 1, 2.5 and 10 microns respectively (PM1, PM2.5 and PM10) are measured at specified indoor locations consisting of four classrooms, four corridors, and four faculty offices in the Payambar Azam Campus of the Mazandaran University of Medical Sciences, Iran. Samples are collected at all the monitoring locations using continuous sampling instruments during weekday mornings and afternoons during the month of April until September 2014, wherein this sample

GP implementation procedure

Due to the varied parameters' contribution to PM, we attempted to identify the correlation between PM1, PM2.5 and PM10 indoors and outdoors respectively. Identification of a relation between the input and output variables, falls under the category of system identification problem. In this framework, input-output behaviour of the unknown process system is approximated using an appropriate model. Appropriate modeling components must be selected to ensure that a model can accurately reproduce the

Results and discussion

The descriptive statistics of the indoor and outdoor sampling locations for PM1, PM2.5 and PM10 concentrations respectively are shown in Table 1. The mean indoor PM1 concentration (18.70 μg/m3) was much higher than the outdoor mean PM1 concentration (16.46 μg/m3). Similarly, it was also observed that the mean indoor PM10 concentration (60.22 μg/m3) was much higher than the corresponding outdoor mean PM10 concentration (53.42 μg/m3). These observations are definitely alarming, as the students in

Conclusions

Since students usually spend approximately 60% of their daytime in the university and are regarded as particularly vulnerable to potential health hazards due to longer exposure to closed indoor pollution. In this research study, indoor and outdoor PM1, PM2.5, PM10 concentrations were monitored at 12 indoors and 5 outdoors locations, spread across the university campus. It was found that the mean indoor PM1, PM10 concentrations (18.70, 60.22 μg/m3) are much higher than corresponding outdoor

Conflict of interest

Authors have no conflict of interest.

References (43)

  • M. Oliveira et al.

    Assessment of air quality in preschool environments (3–5 years old children) with emphasis on elemental composition of PM10 and PM2.5

    Environmental Pollution

    (2016)
  • D.S. Pandey et al.

    Multi-gene genetic programming based predictive models for municipal solid waste gasification in a fluidized bed gasifier

    Bioresource Technology

    (2015)
  • H. Pombeiro et al.

    Dynamic programming and genetic algorithms to control an HVAC system: Maximizing thermal comfort and minimizing cost with PV production and storage

    Sustainable Cities and Society

    (2017)
  • Z. Qiu et al.

    Pedestrian exposure to traffic PM on different types of urban roads: A case study of Xi’an, China

    Sustainable Cities and Society

    (2017)
  • N.M.M. Ramos et al.

    Knowledge discovery of indoor environment patterns in mild climate countries based on data mining applied to in-situ measurements

    Sustainable Cities and Society

    (2017)
  • K.R. Rao et al.

    Soft sensor based nonlinear control of a chaotic reactor

    IFAC Proceedings Volumes

    (2009)
  • K.R. Rao et al.

    Mathematical and kinetic modeling of biofilm reactor based on ant colony optimization

    Process Biochemistry

    (2010)
  • F. Reisen et al.

    Particle and volatile organic emissions from the combustion of a range of building and furnishing materials using a cone calorimeter

    Fire Safety Journal

    (2014)
  • A. Shafaghat et al.

    Environmental-conscious factors affecting street microclimate and individuals’ respiratory health in tropical coastal cities

    Sustainable Cities and Society

    (2016)
  • D. Voukantsis et al.

    Intercomparison of air quality data using principal component analysis, and forecasting of PM 10 and PM 2.5 concentrations using artificial neural networks, in Thessaloniki and Helsinki

    Science of the Total Environment

    (2011)
  • A.A. Zorpas et al.

    Indoor air quality evaluation of two museums in a subtropical climate conditions

    Sustainable Cities and Society

    (2016)
  • Cited by (12)

    • Study on the influencing factors on indoor PM<inf>2.5</inf> of office buildings in beijing based on statistical and machine learning methods

      2023, Journal of Building Engineering
      Citation Excerpt :

      Considering the outdoor meteorological variables, the influence of wind direction (1.66 ± 0.63%) is larger than that of wind speed (1.17 ± 1.29%) at all the sites. The influence of a specific wind direction depends on the window orientation of the site [37,67]. As for other meteorological factors, indoor temperature and humidity which affect the pattern of personnel activities play more important roles in the variation of indoor PM2.5 than outdoor ones which affect the infiltration of outdoor particles.

    • Modeling Wastewater Treatment Process: A Genetic Programming Approach

      2021, Soft Computing Techniques in Solid Waste and Wastewater Management
    • Artificial Intelligence Models for Forecasting of Municipal Solid Waste Generation

      2021, Soft Computing Techniques in Solid Waste and Wastewater Management
    • Dynamic Multi-objective Optimization of Integrated Waste Management Using Genetic Algorithms

      2021, Soft Computing Techniques in Solid Waste and Wastewater Management
    • Evaluation of machine learning models for predicting the temporal variations of dust storm index in arid regions of Iran

      2021, Atmospheric Pollution Research
      Citation Excerpt :

      The activity level of dust storms is usually evaluated based on changes in the various criteria such as dust concentration (Shao and Wang, 2003), frequency of dust days (Ekhtesasi an Sepehr, 2009), aerosols optical depth index (Butt et al., 2017) and dust storm index (O'Loingsigh et al., 2014). Various models have been applied to predict dust events and air quality, including multivariate linear regression−MLR (Kaboodvandpour et al., 2015; Ebrahimi Khusfi et al., 2020a), multivariate adaptive regression splines−MARS (Kisi et al., 2017), least absolute shrinkage and selection operator−Lasso (Son et al., 2018), k-nearest neighbors−kNN (Dragomir, 2010), neuro-fuzzy inference system−ANFIS (Rahmati et al., 2020b), artificial neural networks−ANN (Li et al., 2017), genetic programming−GP (Karri et al., 2018), conditional inference trees−Cforest (Gholami et al., 2020b), a hybrid genetic algorithm-artificial neural network−GA-ANN (Kouchami- Sardoo et al., 2020), deep learning−DL (Xayasouk and Lee, 2018), and random forest−RF (Souri and Vajedian, 2015; Hu et al., 2017; Huang et al., 2018; Stafoggia et al., 2019). For instance, Gholami et al. (2020b) indicated the Cforest and Cubist models performed better than the gradient boosted decision trees−XGBoost, MARS, ANFIS, and Elasticnet models for predicting wind erosion-sensitive lands in southeastern Iran.

    • Mapping spatial distribution of particulate matter using Kriging and Inverse Distance Weighting at supersites of megacity Delhi

      2020, Sustainable Cities and Society
      Citation Excerpt :

      PM2.5 also affects the indoor air quality in places which are densely populated and residential zones are located closer to traffic and highways (Martins & da Graça, 2018). Modelling techniques have been used to predict the effect of outdoor air quality on the indoor air quality highlighting the effect (Karri et al., 2018). Prediction of pollutant concentration using Kriging and IDW techniques has acceptable dependability and the future scope of combining it with other spatial interpolation techniques is vast.

    View all citing articles on Scopus
    View full text