Multivariate time series based back propagation neural network modeling of air quality inside a public transportation bus using available software

Abstract

Time series and artificial neural networks (ANNs) are two distinct methodologies that present environmental researchers and managers with the resources to develop reliable and accurate indoor air quality (IAQ) models. The development of valid IAQ models is a fundamental component in the development of risk assessment and mitigation plans that ensure an occupant's exposure to indoor air contaminants are within the permissible IAQ guidelines. This software review paper provides a detailed step‐by‐step description of the methodology on how one may integrate the use of the multivariate time series and the back propagation neural network (widely used ANN) methods simultaneously in the development of valid in‐bus carbon dioxide and carbon monoxide models using available software. © 2015 American Institute of Chemical Engineers Environ Prog, 34: 1259–1266, 2015

Multivariate time series based radial basis function neural network modeling of air quality inside a public transportation bus using available software

“…The neurons in the hidden layer of CO 2 and CO RBFNN MTS models are added up until the desired accuracy is reached for a given spread. The RBFNN MTS database were normalized within a range of [0,1] using Equation to avoid the overflow of network due to large or small weights produced for the dataset and to eliminate the influence of dimensions of data on the network . where,…”

“…Table summarizes the best performing multivariate transfer function ARMAX and ARIMAX time series models for CO 2 and CO that were identified to be based on a set of five and three independent variables, respectively, from an initial estimate of 12 independent variables [indoor temperature, indoor relative humidity, ambient temperature, ambient relative humidity, wind speed, precipitation, passenger density, light vehicles (cars/SUVs), heavy vehicles (buses/trucks), bus condition/ventilation settings (run/close, idle/open, idle/close)]. Kadiyala and Kumar provided more details on the ste‐by‐step approach to choosing the best performing MTS models using the SPSS software. This software review paper utilized the identified input variables from MTS methods (refer Table ) to develop valid CO 2 and CO RBFNN MTS models in accordance with the identified ARMAX and ARIMAX models of the orders (1,0,5) and (2,1,2) as noted by Kadiyala and Kumar .…”

“…Kadiyala and Kumar software review papers successfully demonstrated the modeling of in‐vehicle air quality with univariate, multivariate, and vector time series methods. They extended the research to develop valid hybrid back propagation neural network models with inputs optimized from the use of univariate , multivariate , and vector time series methods. Kadiyala and Kumar developed valid hybrid RBFNN models with inputs optimized from the use of univariate time series methods.…”

“…Kadiyala and Kumar developed valid hybrid RBFNN models with inputs optimized from the use of univariate time series methods. This study is a follow‐up to the prior in‐vehicle studies on MTS and emphasizes on providing a step‐by‐step approach to the evaluation of the proposed methodology in developing RBFNN MTS models for the monitored contaminants of carbon dioxide (CO 2 ) and carbon monoxide (CO) inside a 20% grade biodiesel (BD20) operated public transportation bus in Toledo, Ohio using available software (Microsoft® Excel 2013, SPSS Statistics 17, MathWorks ® MATLAB 2011b, BOOT v2.0, MINITAB ). Environmental modelers and managers may use their own datasets to develop corresponding RBFNN MTS models by following the steps outlined in this software review paper.…”

Applications of Python to evaluate environmental data science problems

“…The path to the working directory may be set up by providing the path information in a dialog box that opens up when one accesses the options of "Run ! (11) print (data.head (25)) . Working Directory" located within the Spyder application Menu Toolbar.…”