Description:DESCRIPTION:
Field of the invention:
[0001] The present disclosure generally relates to the technical field of air quality prediction, in specific, relates to a system for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period.
Background of the invention:
[0002] Air quality management is a crucial process that involves monitoring air quality, assessing the impact of human activities, taking measures to improve the situation, and ensuring that these measures are effective. This system is designed to reduce the emission of pollutants and other harmful substances in the atmosphere and to sustain ambient air quality. In recent years, effective methods have been developed globally to understand and summarize the importance of good air quality. One of the best ways to maintain air quality is by computing the air quality index, which helps to categorize the meteorological conditions relevant to the air quality of that country or city.

[0003] Air pollution forecasting is the use of science and technology to predict the composition of air pollution in the atmosphere for a given location and time. Government agencies use air quality index (AQI) to communicate to the public how polluted the air is currently or how polluted it is forecast to become. As air pollution levels rise, so does the AQI, along with the associated public health risk.

[0004] Growing industries in emerging countries contributes significantly to air pollution. Pollutants in the air can be harmful to people’s heath .Every year millions of people die as a result of diseases brought on by exposure to outdoor air pollution. The sources of pollutants are fire and dust transport occurring naturally which also include various anthropogenic activities such as factories, power plants and incineration plants. Domestic heating, industrial plant operations and traffic are the most common sources of pollutants that cause particulate matter.
[0005] The air quality is uniformly indexed and consistently computed throughout India to avoid confusion. However, it alone cannot provide a complete picture of air quality. It does not offer information about the correlation between different pollutants, making it challenging to determine the cause of increased air pollution levels. Therefore, it is recommended to use air quality indexing in combination with other factors for a comprehensive understanding of air quality.

[0006] The existing air pollution prediction systems relies solely on observing specific pollutants like PM2.5, PM10, CO, SO2, NO2, and O3 to calculate the Air Quality Index (AQI). This overlooks the crucial role of meteorological parameters in influencing air quality. While advancements in machine learning and deep learning offer powerful tools, their application to pollution prediction often suffers from limitations like single-algorithm reliance, insufficient accuracy, neglecting meteorological data, and suboptimal learning.

[0007] Multivariate statistical analysis is a useful tool for understanding the factors that contribute to air pollution and for accurately assessing air quality. The multivariate statistical analysis include a factor analysis (FA), component analysis (PCA) and cluster analysis (CA), is a common statistical method for analyzing air quality data. The PCA reduces variables by correlating them and creating new principal components. The cluster analysis groups similar data into clusters with different characteristics. Standardizing data is necessary before using the PCA to avoid false results. Statistical methods do not indicate cause and effect relationships between pollutants and air quality. However, by predicting future pollutant concentrations, residents and policymakers can improve air quality. Artificial neural networking is another mathematical approach for air quality analysis.

[0008] When describing air quality, it is important to report the concentration of all pollutants in that location. Air quality parameters include gases, volatile organic compounds, and meteorological parameters. In India, rapid urban development has led to severe crises in air quality, affecting biology, physics, and economic system.

[0009] By addressing all the above there is a need for a system for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period. There is also a need for a system for predicting air quality index that uses multivariate statistical analysis technique for assessing air quality accurately. There is also a need for a system for predicting air quality index that uses an artificial neural network (ANN) model for improving prediction accuracy, handling missing values, and mapping variable effectively. Further, there is also a need for a system for predicting air quality index that correlates the relation between the pollutant parameters and the meteorological parameters for predicting air quality.
Objectives of the invention:
[0010] The primary objective of the present invention is to provide a system for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period.

[0011] Another objective of the present invention is to provide a system for predicting air quality index that uses multivariate statistical analysis technique for assessing air quality accurately.

[0012] Yet another objective of the present invention is to provide a system for predicting air quality index that uses an artificial neural network (ANN) model for improving prediction accuracy, handling missing values, and mapping variable effectively.

[0013] Further objective of the present invention is to provide a system for predicting air quality index that correlates the relation between the pollutant parameters and the meteorological parameters for predicting air quality.
Summary of the invention:
[0014] The present disclosure proposes a system for predicting air quality index using artificial neural network model and method thereof. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key/critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.

[0015] In order to overcome the above deficiencies of the prior art, the present disclosure is to solve the technical problem to provide a system for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period.

[0016] According to one aspect, the invention provides the system for predicting air quality index using multivariate statistical analysis and artificial neural network model comprises a computing device having a processor and a memory for storing one or more instructions executable by the processor, in specific the computing device is in communication with a server via a network.

[0017] In one embodiment, the processor is configured to execute plurality of modules. The plurality of modules comprises an input module, an aggregation module, a regression module, a correlation analysis module, a prediction module, and a computation module.

[0018] In one embodiment, the input module is configured to receive input parameters include plurality of pollutant parameters and plurality of meteorological parameters. In one embodiment, the pollutant parameters include seasonal data of gaseous, particulate matter and volatile organic compounds. In one embodiment, the meteorological parameters include wind speed, wind direction, relative humidity, sun radiation, and atmospheric temperature. In one embodiment, the volatile organic compounds include benzene (C6H6), toluene (C6H5CH3) and xylene ((CH3)2C6H4). The gases include carbon monoxide (CO), sulphur dioxide (SO2), oxides of nitrogen (NO?)

[0019] In one embodiment, the aggregation module is configured to aggregate the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data for each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively.

[0020] In one embodiment, the regression module is configured to analyze the air quality index (AQI) values computed from a central pollution control board (CPCB) method, a Tiwari and Ali method, and a PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively.

[0021] In one embodiment, the correlation analysis module is configured to determine a correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using a multivariate statistical analysis to provide selected input data for developing an artificial neural network (ANN) model. In one embodiment, the multivariate statistical analysis is conducted using plurality of methods, which include a factor analysis (FA), a cluster analysis (CA), and a principal component analysis (PCA).

[0022] In one embodiment, the prediction module is configured to receive the selected input data and predict the air quality of the plurality of pollutant parameters and the plurality of meteorological parameters using the ANN model, thereby providing predicted air quality data. The artificial neural network model utilizes a non-linear time series prediction and model tool with two layered architecture. In one embodiment, the artificial neural network model is a non-linear auto-regression exogenous model (NARX) network that predicts the number of response parameters and the number of predictor parameters.

[0023] In one embodiment, the computation module is configured to generate air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the CPCB method.

[0024] According to another aspect, the invention provides a method for predicting the air quality index using a system. At one step, the input module receives the input parameter include the plurality of pollutant parameters and the plurality of meteorological parameters. At another step, the aggregating module aggregates the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data of each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively.

[0025] At another step, the regression module analyses the air quality index (AQI) values computed from the CPCB method, the Tiwari and Ali method, and the PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively.

[0026] At other step, the correlation analysis module determines the correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using multivariate statistical analysis to provide selected input data for developing an artificial neural network (ANN) model.

[0027] At another step, the prediction module receives the selected input data and predict the air quality of the plurality of pollutant parameters and the plurality of meteorological parameters using the ANN model, thereby providing predicted air quality data.

[0028] Further at other step, the computation module generates the air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the central pollution control board (CPCB) method.

[0029] Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawings.
Detailed description of drawings:
[0030] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, explain the principles of the invention.

[0031] FIG. 1 illustrates a block diagram of a system for predicting air quality index, in accordance to an exemplary embodiment of the invention.

[0032] FIG. 2 illustrates a flow chart of an aggregation module of the system for predicting air quality index, in accordance to an exemplary embodiment of the invention.

[0033] FIGs. 3A-3B illustrate graphs representing the comparison charts of air quality index (AQI) for summer season and monsoon season, in accordance to an exemplary embodiment of the invention.

[0034] FIGs. 3C-3D illustrate a graphs representing the comparison charts of air quality index (AQI) for post-monsoon season and winter season, in accordance to an exemplary embodiment of the invention.

[0035] FIG. 4 illustrates a schematic view of a non-linear auto-regression exogenous model (NARX) network model, in accordance to an exemplary embodiment of the invention.

[0036] FIG. 5 illustrates a schematic view of an architecture of a neural network model, in accordance to an exemplary embodiment of the invention.

[0037] FIG. 6 illustrates a flowchart of a method for predicting the air quality index using a system, in accordance to an exemplary embodiment of the invention.
Detailed invention disclosure:
[0038] Various embodiments of the present invention will be described in reference to the accompanying drawings. Wherever possible, same or similar reference numerals are used in the drawings and the description to refer to the same or like parts or steps.

[0039] The present disclosure has been made with a view towards solving the problem with the prior art described above, and it is an object of the present invention to provide a system for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period.

[0040] According to one exemplary embodiment of the invention, FIG. 1 refers to a block diagram of a system 100 for predicting air quality index. In one embodiment herein, the system 100 for predicting air quality index uses multivariate statistical analysis technique for assessing air quality accurately. The system 100 for predicting air quality index uses artificial neural network (ANN) model for improving prediction accuracy, handling missing values, and mapping variable effectively.

[0041] In one embodiment herein, the system 100 for predicting air quality index using multivariate statistical analysis and artificial neural network model comprises a computing device 102 having a processor 104 and a memory 106 for storing one or more instructions executable by the processor 104, in specific the computing device 102 is in communication with a server 126 via a network 124. The system 100 stores input data and other information in a database 122.

[0042] In one embodiment herein, the processor 104 is configured to execute plurality of modules 108. The plurality of modules 108 comprises an input module 110, an aggregation module 112, a regression module 113, a correlation analysis module 114, a prediction module 116, and a computation module 118.

[0043] In one embodiment herein, the input module 110 is configured to receive input parameters include plurality of pollutant parameters and plurality of meteorological parameters. In one embodiment herein, the pollutant parameters include seasonal data of gaseous, particulate matter and volatile organic compounds. In one embodiment herein, the meteorological parameters include wind speed (WS), wind direction (WD), relative humidity (RH), sun radiation (SR), and atmospheric temperature (AT). In one embodiment herein, the volatile organic compounds include benzene (C6H6), toluene (C6H5CH3) and xylene ((CH3)2C6H4). The gases include carbon monoxide (CO), sulphur dioxide (SO2), oxides of nitrogen (NO?)

[0044] In one embodiment herein, the aggregation module 112 is configured to aggregate the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data for each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively.

[0045] In one embodiment herein, the regression module 113 is configured to analyze the air quality index (AQI) values computed from a central pollution control board (CPCB) method, a Tiwari and Ali method, and a PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively.

[0046] In one embodiment herein, the correlation analysis module 114 is configured to determine a correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using a multivariate statistical analysis to provide selected input data for developing an artificial neural network model (ANN) model. In one embodiment herein, the multivariate statistical analysis is conducted using plurality of methods, which includes a factor analysis (FA), a cluster analysis (CA), and a principal component analysis (PCA).

[0047] In one embodiment herein, the prediction module 116 is configured to receive the selected input data and predict the air quality of the plurality of pollutant parameters and the plurality of meteorological parameters using the ANN model, thereby providing predicted air quality data. The artificial neural network model utilizes a non-linear time series prediction and modelling tool with two layered architecture. In one embodiment herein, the artificial neural network model is a non-linear auto-regression exogenous model (NARX) network that predicts the number of response parameters (R) and the number of predictor parameters (P).

[0048] In one embodiment herein, the computation module 118 is configured generate air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the CPCB method.

[0049] According to another exemplary embodiment of the invention, FIG. 2 refers to a flowchart 200 for the input data aggregation by the aggregation module 112 of the system 100 for predicting the air quality index. A study area is selected to predict the ambient air quality, in specific the study area could be industrial areas and residential areas. The industrial areas include small and medium scale chemical, pharmaceutical, electrical and mechanical industries. The connectivity of the study area to neighboring places is via a road, which implies that most of the pollution concentrations are related to vehicular emissions.

[0050] To study the seasonal trends and for a better understanding of the AQI in the study area a comprehensive comparative analysis of AQI is conducted. This is obtained by computing AQI by three distinct formulas for period of 2007-2017.After incorporating the missing values, the hourly data is transformed into daily data for each parameter by calculating the average value of each parameter from the hourly data. The daily data is then averaged into monthly data, then the monthly data is divided into seasonal data.

[0051] In one embodiment herein, the seasonal data relating to the sampling location is collected from Telangana state pollution control board (TSPCB), Hyderabad, Telangana, India for the years 2007-2017. The following Table 1 describes the types of parameters collected from TSPCB for this study.

[0052] Table 1:
S.No Parameters Type of parameters Unit
1 Carbon Monoxide
(CO) Ambient air pollutant
2 Sulphur dioxide
(SO2) Ambient air pollutant
3 Particulate Matter
(PM) Ambient air pollutant
4 Oxides of Nitrogen
(NOx) Ambient air pollutant
5 Benzene
(C6H6) Ambient air pollutant
6 Toluene
(C6H5CH3) Ambient air pollutant
7 Xylene
((CH3)2C6H4) Ambient air pollutant
8 Atmospheric Temperature
(AT) Meteorological Parameter °C
9 Relative Humidity
(RH) Meteorological Parameter %
10 Wind Speed
(WS) Meteorological Parameter
11 Sun radiation
(SR) Meteorological Parameter
12 Wind Direction
WD Meteorological Parameter deg

[0053] According to another exemplary embodiment of the invention, FIGs. 3A-3B refer to graphs (300, 302) representing the comparison charts of air quality index (AQI) for summer season and monsoon season. Once the seasonal data is obtained, the AQI for each season are computed using the formulas by CPCB method. The comparative study obtained from these methods stated that the AQI calculated in the three methods is above the CPCB value for the entire study period in all the seasons expect for summer. According to another exemplary embodiment of the invention, FIGs. 3C-3D refer to graphs (304, 306) representing the comparison charts of air quality index (AQI) for post-monsoon season and winter season.

[0054] In one embodiment herein, the regression module 113 configured to analyze the air quality index (AQI) values computed from the CPCB method, the Tiwari and Ali method, and the PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively.
[0055] . The following Table 2 describes the correlation analysis of the AQI values based on the season.

[0056] Table 2:
Season Index Regression equation R2 Standard deviation Coefficient of variance %
Summer CPCB y = -1.3077x2 + 14.238x + 82.545 0.51 17.48 16
PI y = -1.3112x2 + 14.307x + 68.745 0.65 12.72 24
Tiwari y = -0.3555x2 + 3.6021x + 54.648 0.42 17.61 14
Post-monsoon CPCB y = -0.8415x2 + 9.407x + 35.812 0.46 25.42 18
PI y = -0.5256x2 + 5.5713x + 33.479 0.35 15.38 16
Tiwari y = -0.0559x2 + 1.035x + 21.909 0.19 7.64 18
Monsoon CPCB y = 1.035x2 – 10.601x + 146.09 0.38 19.15 19
PI y = 0.0758x2 – 3.9727x + 122.99 0.26 13.37 12
Tiwari y = 0.7354x2 – 11.325x + 98.939 0.63 5.85 10
Winter CPCB y = 1.1037x2 – 8.2357x + 136.1 0.54 2.87 11
PI y = 0.028x2 – 2.3175x + 122.8 0.22 12.94 20
Tiwari y = 0.1818x2 – 1.2545x + 70.891 0.21 7.31 10

[0057] In the above study the primary pollutant is identified by the maximum index values, which is observed to be “Particulate matter (PM)” through the study period. From correlation analysis Table 2, it can be noted that CPCB method for all the seasons shows good statistical values reflecting the reliability of this method for the study of air quality.

[0058] In one embodiment herein, the multivariate statistical analysis is conducted using three methods includes the factor analysis (FA), the cluster Analysis (CA), and the principal component analysis (PCA). The statistical analysis is carried out only for summer, post- monsoon and winter season data, as the missing data for monsoon season is over 30 %. The parameters are categorised into gaseous (CO, SO2, NO?), particulate matter (PM), volatile organic compounds (benzene, toluene and xylene), statistical analysis is conducted for each group along with the meteorological parameters in the respective seasons of the study area. Table 3 presents the statistical analysis and descriptions of different cases.

[0059] Table 3:
Cases No Description of different cases
1 FA: Gaseous parameters, PM, Meteorological parameters (Summer season)
2 FA: Gaseous parameters, PM, Meteorological parameters (Post-monsoon season)
3 FA: Gaseous parameters, PM, Meteorological parameters (Winter season)
4 FA: VOC parameters, Meteorological parameters (Summer season)
5 FA: VOC parameters, Meteorological parameters (Post-monsoon season)
6 FA: VOC parameters, Meteorological parameters (Winter season)
7 CA: Gaseous parameters, PM, Meteorological parameters (Summer season)
8 CA: Gaseous parameters, PM, Meteorological parameters (Post-monsoon season)
9 CA: Gaseous parameters, PM, Meteorological parameters (Winter season)
10 CA: VOC parameters, Meteorological parameters (Summer season)
11 CA: VOC parameters, Meteorological parameters (Post-monsoon season)
12 CA: VOC parameters, Meteorological parameters (Winter season)
13 PCA: Gaseous parameters, PM, Meteorological parameters (Summer season)
14 PCA: Gaseous parameters, PM, Meteorological parameters (Post-monsoon season)
15 PCA: Gaseous parameters, PM, Meteorological parameters (Winter season)
16 PCA: VOC parameters, Meteorological parameters (Summer season)
17 PCA: VOC parameters, Meteorological parameters (Post-monsoon season)
18 PCA: VOC parameters, Meteorological parameters (Winter season)

[0060] In one embodiment herein, the multivariate analysis is conducted to understand the similarity and seasonal variations of the pollutant parameters. In the cases 1 to 6 where FA is conducted no factors are produced therefore no further extractions are observed. But from case 7 to 18. The PCA states that highly influencing gaseous parameters is PM, and the most correlated gaseous parameters are NO? - CO, as for VOC parameters Toluene is identified as the highly influencing parameter and Toluene – Xylene are most correlated in all the respective seasons. This method also stated that most influencing meteorological parameters are “WS, AT, RH, SR” for all the seasons.

[0061] In one embodiment herein, the CA shows the highest correlated parameters are CO and WS for gaseous and PM parameters in all the seasons. While the CA for the VOC parameters showed “Benzene and WS” exhibits strong correlation in summer and winter, and for summer it is between “Xylene and Benzene”. After analysing all the dendrograms and agglomeration tables for gaseous, PM and VOC, the parameters “WS, AT, RH and SR” from a prominent cluster indicates a significant influence on the air quality in the study area.

[0062] The findings from the PCA and the CA shows a seasonal variation in their influence, indicating that the meteorological parameters vary with respect to season. Additionally, the type of the influencing meteorological factors has pollutant parameters changed with the pollutant categories, therefore it can be concluded that the influencing meteorological factors vary with respect to season and the type of pollutant.

[0063] In one embodiment herein, the average of the VOC parameters is computed for every year. Once the average is computed, interspecies ratios T/B and X/B are analyzed. The Table 4 below gives interspecies ration with respect to the season.

[0064] Table 4:
Season Year Benzeneµgm³ Tolueneµgm³ Xyeleneµgm³ T/B ratio X/B ratio
Summer season 2007 5.02 17.55 2.16 3.50 0.43
2008 4.68 14.08 3.13 3.01 0.67
2009 6.49 37.17 6.74 5.73 1.04
2010 4.60 38.11 9.35 8.29 2.03
2011 1.69 10.19 2.12 6.02 1.25
2012 1.07 6.06 0.67 5.64 0.62
2013 1.69 10.19 2.12 6.02 1.25
2014 1.07 6.06 0.67 5.64 0.62
2015 1.29 14.00 2.72 10.83 2.10
2016 0.86 7.89 0.97 9.14 1.13
2017 0.31 4.38 0.47 13.99 1.52
Post-Monsoon season 2007 5.13 39.69 3.37 7.74 0.66
2008 4.26 18.90 2.66 4.44 0.63
2009 29.67 143.68 26.95 4.84 0.91
2010 6.16 41.28 8.52 6.70 1.38
2011 1.31 14.53 2.22 11.07 1.69
2012 6.83 52.53 11.91 7.69 1.74
2013 1.31 14.53 2.22 11.07 1.69
2014 6.83 52.53 11.91 7.69 1.74
2015 3.14 25.15 0.26 8.01 0.08
2016 0.87 9.82 0.82 11.25 0.94
2017 1.84 36.46 2.25 19.83 1.22
Winter season 2007 7.34 30.47 5.59 4.15 0.76
2008 5.09 19.13 2.85 3.76 0.56
2009 8.71 53.57 9.32 6.15 1.07
2010 7.48 66.88 12.77 8.94 1.71
2011 3.95 27.10 5.02 6.86 1.27
2012 2.45 18.78 4.62 7.66 1.88
2013 3.95 27.10 5.02 6.86 1.27
2014 2.45 18.78 4.62 7.66 1.88
2015 2.68 30.06 4.25 11.20 1.58
2016 1.38 14.44 0.90 10.44 0.65
2017 1.55 20.27 1.49 13.12 0.97

[0065] According to another exemplary embodiment of the invention, FIG. 4 refer to a schematic view of a non-linear auto-regression exogenous model (NARX) network model 400. The artificial neural network (ANN) model has several advantages over statistical methods with an accurate prediction. The prediction is carried out for the air pollutants such as CO, SO2, NO?, PM, benzene, toluene, and xylene. The prediction is carried out by using meteorological parameters as set of inputs. The input data is obtained from the PCA and the CA outputs.

[0066] FIG. 4 represents the model structure developed in the present invention shows a two layered simple feed forward network. It comprises of four sections. The first section includes y (t) (number of response parameters (R)) and x (t) (number of predictor parameters (P)). The second section consists of a hidden layer that comprises of weights (w), bias (b) neurons, and input variables. The network training takes place in second section. The third section includes output variables that are influenced by the weights (w) and the bias (b). The network validation and testing are conducted in the third section. The final stage involves obtaining predicted variables through prediction by comparing mean squared error (MSE) and R-squared (R2) values.

[0067] According to another exemplary embodiment of the invention, FIG. 5 refer to a schematic view of an architecture 500 of the artificial neural network model. The input data for the ANN model varies based on the season and model. The number of neurons and hidden layers in the network architecture effect the output, therefore different sets of neurons (n) are used, specifically 10,15 and 20. The input data for each model is segregated into three categories training data (70%), validation data (15%), and test data (15%), with a time delay of 2. The training algorithm Levenberg- Marquardt is applied for all the models.

[0068] The comparison is observed in the tabular form to facilitate seasonal study. Therefore, two set of data sets are used for the same purpose where the predicted data is observed from 2024 – 2026 (as shown in Table 6) and for the verification of model is done with the data 2017-2023 (as shown in Table 5).

[0069] Table 5:
Season Year Month AQI Predicted AQI CPCB
Value Quality status Value Quality status
SUMMER 2017 MARCH 128 MODERATE 90 SATISFACTORY
APRIL 91 SATISFACTORY 107 MODERATE
MAY 79 SATISFACTORY 73 SATISFACTORY
JUNE 62 SATISFACTORY 38 GOOD
2018 MARCH 67 SATISFACTORY 122 MODERATE
APRIL 99 SATISFACTORY 83 SATISFACTORY
MAY 102 MODERATE 69 SATISFACTORY
JUNE 99 SATISFACTORY 45 GOOD
2019 MARCH 117 MODERATE 89 SATISFACTORY
APRIL 110 MODERATE 69 SATISFACTORY
MAY 107 MODERATE 77 SATISFACTORY
JUNE 57 SATISFACTORY 52 SATISFACTORY
2020 MARCH 111 MODERATE 66 SATISFACTORY
APRIL 92 SATISFACTORY 56 SATISFACTORY
MAY 75 SATISFACTORY 63 SATISFACTORY
JUNE 57 SATISFACTORY 34 GOOD
2021 MARCH 143 MODERATE 128 MODERATE
APRIL 117 MODERATE 90 MODERATE
MAY 140 MODERATE 48 GOOD
JUNE 88 SATISFACTORY 36 GOOD
2022 MARCH 62 SATISFACTORY 123 MODERATE
APRIL 116 MODERATE 81 SATISFACTORY
MAY 120 MODERATE 74 SATISFACTORY
JUNE 63 SATISFACTORY 47 GOOD
2023 MARCH 143 MODERATE 74 SATISFACTORY
APRIL 117 MODERATE 82 SATISFACTORY
MAY 140 MODERATE 67 SATISFACTORY
JUNE 88 SATISFACTORY 46 GOOD
POST-MONSOON 2017 OCTOBER 114 MODERATE 113 MODERATE
NOVEMBER 135 MODERATE 107 MODERATE
2018 OCTOBER 123 MODERATE 86 SATISFACTORY
NOVEMBER 145 MODERATE 163 MODERATE
2019 OCTOBER 116 MODERATE 93 SATISFACTORY
NOVEMBER 108 MODERATE 188 MODERATE
2020 OCTOBER 79 SATISFACTORY 103 MODERATE
NOVEMBER 107 MODERATE 126 MODERATE
2021 OCTOBER 116 MODERATE 94 SATISFACTORY
NOVEMBER 134 MODERATE 106 MODERATE
2022 OCTOBER 102 MODERATE 97 SATISFACTORY
NOVEMBER 121 MODERATE 178 MODERATE
WINTER 2017 JANUARY 125 MODERATE 173 MODERATE
FEBRUARY 114 MODERATE 203 POOR
DECEMBER 129 MODERATE 243 POOR
2018 JANUARY 125 MODERATE 229 POOR
FEBRUARY 110 MODERATE 125 MODERATE
DECEMBER 192 MODERATE 202 POOR
2020 JANUARY 146 MODERATE 211 POOR
FEBRUARY 140 MODERATE 100 MODERATE
DECEMBER 122 MODERATE 176 MODERATE
2021 JANUARY 107 MODERATE 104 MODERATE
FEBRUARY 103 MODERATE 95 SATISFACTORY
DECEMBER 123 MODERATE 192 MODERATE
2022 JANUARY 122 MODERATE 159 MODERATE
FEBRUARY 136 MODERATE 158 MODERATE
DECEMBER 136 MODERATE 185 MODERATE
2023 JANUARY 138 MODERATE 141 MODERATE
FEBRUARY 147 MODERATE 107 MODERATE
DECEMBER 110 MODERATE 155 MODERATE

[0070] Table 6:
Season Year Month AQI Predicted
Value Quality status
SUMMER 2024 MARCH 62 SATISFACTORY
APRIL 116 MODERATE
MAY 119 MODERATE
JUNE 62 SATISFACTORY
2025 MARCH 44 SATISFACTORY
APRIL 57 SATISFACTORY
MAY 41 SATISFACTORY
JUNE 59 SATISFACTORY
2026 MARCH 56 SATISFACTORY
APRIL 50 GOOD
MAY 37 GOOD
POST-MONSOON 2023 OCTOBER 116 MODERATE
NOVEMBER 135 MODERATE
2024 OCTOBER 102 MODERATE
NOVEMBER 119 MODERATE
2025 OCTOBER 62 SATISFACTORY
NOVEMBER 56 SATISFACTORY
2026 OCTOBER 53 SATISFACTORY
NOVEMBER 81 SATISFACTORY
WINTER 2024 JANUARY 121 MODERATE
FEBRUARY 138 MODERATE
DECEMBER 136 MODERATE
2025 JANUARY 139 MODERATE
FEBRUARY 145 MODERATE
DECEMBER 107 MODERATE
2026 JANUARY 88 SATISFACTORY
FEBRUARY 73 SATISFACTORY
DECEMBER 64 SATISFACTORY

[0071] According to another exemplary embodiment of the invention, FIG. 6 refers to a flowchart 600 of a method for predicting the air quality index using the system 100. At step 602, the input module 110 receives the input parameters include the plurality of pollutant parameters and the plurality of meteorological parameters. At step 604, the aggregating module 112 aggregates the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data of each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively.

[0072] At step 606, the regression module 113 analyses the air quality index (AQI) values computed from the central pollution control board (CPCB) method, the Tiwari and Ali method, and the PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively.

[0073] At step 608, the correlation analysis module 114 determines the correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using multivariate statistical analysis, thereby providing the selected input data for developing the artificial neural network (ANN) model.

[0074] At step 610, the prediction module 116 receives the selected input data and predict the air quality of the plurality of pollutant parameters and the plurality of meteorological parameters using the ANN model, thereby providing predicted air quality data.

[0075] Further at step 612, the computation module 118 generates air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the CPCB method.

[0076] Numerous advantages of the present disclosure may be apparent from the discussion above. In accordance with the present disclosure, a system 100 for predicting air quality index using multivariate statistical analysis and artificial neural network model is disclosed. The proposed invention provides the system 100 for predicting air quality index by developing an artificial neural network model using real time data and detecting the significant meteorological factors affecting the concentration of air pollutants in a given period. The system 100 for predicting air quality index uses multivariate statistical analysis technique for assessing air quality accurately. The system 100 for predicting air quality index uses the artificial neural network (ANN) model for improving prediction accuracy, handling missing values, and mapping variable effectively. The system 100 for predicting air quality index correlates the relation between the pollutant parameters and the meteorological parameters for predicting air quality.

[0077] It will readily be apparent that numerous modifications and alterations can be made to the processes described in the foregoing examples without departing from the principles underlying the invention, and all such modifications and alterations are intended to be embraced by this application.
, Claims:CLAIMS:
I/We Claim:
1. A system (100) for predicting air quality index using multivariate statistical analysis and artificial neural network model, comprising:
a computing device (102) having a processor (104) and a memory (106) for storing one or more instructions executable by the processor (104), wherein the computing device (102) is in communication with a server (126) via a network (124),
wherein the processor (104) is configured to execute plurality of modules (108), wherein the plurality of modules (108) comprises:
an input module (110) configured to receive input parameters include plurality of pollutant parameters and plurality of meteorological parameters;
an aggregation module (112) configured to aggregate the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data for each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively;
a regression module (113) configured to analyze the air quality index (AQI) values computed from a central pollution control board (CPCB) method, a Tiwari and Ali method, and a PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively;
a correlation analysis module (114) configured to determine a correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using a multivariate statistical analysis to provide selected input data for developing an artificial neural network (ANN) model;
a prediction module (116) configured to receive the selected input data and predict air quality based on the plurality of pollutant parameters and the plurality of meteorological parameters using the ANN model, thereby providing predicted air quality data; and
a computation module (118) configured to generate air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the central pollution control board (CPCB) method.
2. The system (100) for predicting air quality index as claimed in claim 1, wherein the multivariate statistical analysis is conducted using plurality of methods, which includes a factor analysis (FA), a cluster analysis (CA), and a principal component analysis (PCA).
3. The system (100) for predicting air quality index as claimed in claim 1, wherein the pollutant parameters include the seasonal data of gaseous, particulate matter and volatile organic compounds.
4. The system (100) for predicting air quality index as claimed in claim 1, wherein the meteorological parameters include wind speed, wind direction, relative humidity, sun radiation, and atmospheric temperature.
5. The system (100) for predicting air quality index as claimed in claim 2, wherein the volatile organic compounds include benzene (C6H6), toluene (C6H5CH3) and xylene ((CH3)2C6H4).
6. The system (100) for predicting air quality index as claimed in claim 1, wherein the gases include carbon monoxide (CO), sulphur dioxide (SO2), oxides of nitrogen (NO?).
7. The system (100) for predicting air quality index as claimed in claim 1, wherein the ANN model utilizes a non-linear time series prediction and modelling tool with two layered architecture.
8. The system (100) for predicting air quality index as claimed in claim 1, wherein the artificial neural network model is a non-linear auto-regression exogenous model (NARX) network that predicts the number of response parameters and the number of predictor parameters.
9. A method for predicting the air quality index using a system (100), comprising:
receiving, by an input module (110), the input parameters include plurality of pollutant parameters and plurality of meteorological parameters;
aggregating, by an aggregating module (112), the plurality of pollutant parameters and the plurality of meteorological parameters, thereby obtaining seasonal data of each parameter of the plurality of pollutant parameters and the plurality of meteorological parameters, respectively;
analyzing, by a regression module (113), air quality index (AQI) values computed from a central pollution control board (CPCB) method, a Tiwari and Ali method, and a PI method, thereby correlating the analysis of the AQI values according to the seasons, respectively;
determining, by a correlation analysis module (114), a correlation between the plurality of pollutant parameters and the plurality of meteorological parameters using multivariate statistical analysis to provide selected input data for developing an artificial neural network (ANN) model;
receiving, by a prediction module (116), the selected input data and predicting the air quality of the plurality of pollutant parameters and the plurality of meteorological parameters using the at least one artificial neural network model, thereby providing predicted air quality data; and
generating, by a computation module (118), the air quality index (AQI) values based on the seasonal data of the each parameter for each season and the air quality data using the CPCB method.