REAL-TIME ENVIRONMENTAL MONITORING AND AIR QUALITY PREDICTION USING IOT AND MACHINE LEARNING IN SMART CITIES

Abstract

REAL-TIME ENVIRONMENTAL MONITORING AND AIR QUALITY PREDICTION USING IOT AND MACHINE LEARNING IN SMART CITIES The method for the development of the data on air quality was gathered over a number of months. Machine learning (ML) algorithms like K-Nearest-Neighbor (KNN), Expectation-Maximization (EM), Multiple Imputation by Chained Equations (MICE), and Autoregressive–Moving-Average (ARMA) are used to address missing data and outliers caused by technical difficulties in order to solve the instability problems of low-cost devices in monitoring. When it came to RMSE, MSE, MAE, R-squared, and execution time, the KNN model performed better than any other. In order to gather actual data on sulfur dioxide, carbon monoxide, nitrogen dioxide, ozone, and particle matters 2.5 and 10 µm, the Internet of Things framework uses MQ9, MQ135, MQ131, and dust or PM sensors equipped with an Arduino microcontroller. The device measures air quality in parts per million (PPM) and detects hazardous gases using two sensors, MQ135 and MQ3. Additionally, the gathered data will be subjected to machine learning analysis.

Specification

Description:REAL-TIME ENVIRONMENTAL MONITORING AND AIR QUALITY PREDICTION USING IOT AND MACHINE LEARNING IN SMART CITIES

Technical Field
[0001] The embodiments herein generally relate to a method for real-time environmental monitoring and air quality prediction using IoT and machine learning in smart cities.
Description of the Related Art
[0002] The sensors can transmit data, communicate with one another, and accurately describe their environment when IoT devices are deployed properly. It is possible to comprehend the surroundings and react to any situation that calls for quick action. IoT can assist with a variety of tasks, including home security and monitoring systems, indoor and outdoor air quality monitoring and management, energy management in public and private buildings, healthcare systems, and micro-climate monitoring systems in particular locations. In order to give a comprehensive understanding of the Air Quality Index (AQI) and its variations across different districts, the dataset measures important air pollutants like sulfur dioxide, carbon monoxide, nitrogen dioxide, ozone, and particulate matter. In addition to identifying areas of concern and potential sources of pollution for focused intervention and pollution management, the dataset's analysis attempts to assess the extent of compliance with current regulatory norms and recommendations. Particulate matter and air pollutants such as sulfur dioxide, nitrogen oxides, and ozone are measured by Air Quality Monitoring (AQM) systems that are equipped with sensors and cutting-edge technologies. These systems gather data that is used to monitor pollution reduction efforts, create policies, and enable the public to make well-informed decisions about their health and well-being. Because there are more sources of pollution and people living in large, densely populated cities are frequently exposed to high levels of air pollution, pollution issues are more common there.
[0003] Our systems used five pieces of equipment outfitted with sensor modules for Carbon Monoxide (CO), Carbon Dioxide (CO2), delicate particulate matter (PM), UV index, humidity, and temperature to collect data on the interior environment on-site. Following receipt of the data, the Arduino Uno R3 board provided the data to the five modules for integration. After that, they use the ESP8266 Wi-Fi module to send the data to Thing Speak for additional computation and analysis. This system's low cost and high accuracy when compared to earlier solutions under the same conditions are its main advantages. automated technology for data cleaning. By comparing real-time air quality data to standard values, the dataset aims to offer significant insights into the overall environmental quality of the locations under investigation. By supporting the creation of prediction models and analytical tools for estimating AQI values and monitoring long-term trends in air quality, it also aims to promote scientific research.
[0004] The main functions of AQM stations are pollution monitoring and the computation of the Air Quality Index (AQI). However, the growth of AQM networks and the accessibility of air pollution data are constrained by the infrastructure needs, operational difficulties, and continuous costs related to these stations. One essential element of IoT technology is smart devices, which enable object connections over pre-existing networks. To improve citizens' quality of life and well-being, they seek to improve city operations like transportation, healthcare, education, water, communication, energy, and competitiveness. Manual data cleansing is no longer practical due to the steadily growing data volumes produced by inexpensive sensors.
SUMMARY
[0001] In view of the foregoing, an embodiment herein provides a method for real-time environmental monitoring and air quality prediction using IoT and machine learning in smart cities. In some embodiments, wherein the development of a device known as "Smart-Air" includes a number of sensors that gather information on air quality, including CO2, CO, smoke, dust, temperature, and humidity. Using a Long-Term Evolution (LTE) modem, the collected data is transmitted to the webserver for additional processing. Similarly, S. McGrath et al. offer an alternative method of air quality monitoring that records four air pollutants: CO, NO2, PM 2.5, and Ozone. To display the state of the air quality, a mobile application and a Firebase web application are being developed. For the device to send the gathered data to the Thing speak, internet connectivity is made possible by the WiFi module. The device is powered by the power source. Electronic sensors and microprocessor/microcontroller chips for signal processing and acquisition are the foundation of contemporary air quality monitoring systems. Typically, these systems gather data to be processed on cloud computing platforms and displayed via web or mobile applications. Big data techniques are typically used in the analysis to glean valuable information from the raw sensor measurements.
[0002] In some embodiments, wherein an alternate model for forecasting air quality that makes use of the Multiple Kernel Learning technique. Four parameters were included in this experimental study: PM, fine suspended particles, respirable suspended particulates, NO2, SO2, and Ozone. The performance of the presented system is compared to that of the multiple-layer perceptron (MLP) neural network, Random Forest (RF), long short-term memory (LSTM), support vector machine (SVM), and classical autoregressive integrated moving average (ARIMA) model. A sensor data acquisition layer, data processing layer, internet connectivity layer, cloud data storage layer, and user interface layer are among the layers that make up the IoT device's software architecture.
[0003] In some embodiments, wherein the two separate weather stations in Illinois's two separate suburban residential districts provided the data. In order to improve performance, the authors of this work suggested updated models for predicting IAQ using complex regularization techniques. A structured method for processing, feature selection, choosing suitable machine learning algorithms, creating and training models, assessing models, and visualizing data is offered by the MQ3 and MQ135 sensors, NodeMCU processor, Arduino IDE, Thing Speak, and Blynk platforms. This framework can assist in offering significant insights for decision-making in a number of applications, including public health and environmental monitoring.
[0004] These and other aspects of the embodiments herein will be better appreciated and understood when considered in conjunction with the following description and the accompanying drawings. It should be understood, however, that the following descriptions, while indicating preferred embodiments and numerous specific details thereof, are given by way of illustration and not of limitation. Many changes and modifications may be made within the scope of the embodiments herein without departing from the spirit thereof, and the embodiments herein include all such modifications.

BRIEF DESCRIPTION OF THE DRAWINGS
[0001] The embodiments herein will be better understood from the following detailed description with reference to the drawings, in which:
[0002] FIG. 1 illustrates a method for real-time environmental monitoring and air quality prediction using IoT and machine learning in smart cities according to an embodiment herein; and
[0003] FIG. 2 illustrates a method for workflow of designed method according to an embodiment herein.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
[0001] The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
[0002] FIG. 1 illustrates a method for real-time environmental monitoring and air quality prediction using IoT and machine learning in smart cities according to an embodiment herein. In some embodiments, connecting to WiFi, connecting to the Thing speak server, and reading and uploading data to Thing speak are the states through which the program runs. Every five minutes, the sensors in the system are set up to read and transmit data. Using its digital and analog inputs, the Arduino UNO R3 gathers all of the sensor data, arranges it into preset data structures, and sends it to the WiFi Chip, which uses the existing WiFi infrastructure to connect to the internet. Before applying ML models, we divide the dataset into two parts, referred to as dependent and independent. In classification, AQI is an independent column that aims to reach the AQI Range, whereas in regression, we let AQI be the goal column. The dataset is separated into training and test data before machine learning is applied. The MQ 3 gas sensor for alcohol and the MQ 135 gas sensor for volatile organic compounds (VOCs) were chosen as the sensors for the system. By subjecting the sensors to known concentrations of contaminants and modifying the readings to correspond with the anticipated values, the sensors were calibrated. An Arduino microcontroller, a power source, MQ 135, MQ 3 gas sensors, an ESP8266, and a Wi-Fi module were also included in the hardware design. Using a collection of sensors, the lower layer also referred to as the perception layer is in charge of collecting the data. These sensors are a component of end-point devices, which are typically built on embedded systems with the ability to communicate with upper layers and perform pre-processing. Because the devices are connected by the network layer, the choice of communication protocols in this layer will affect the system's overall performance.
[0003] In some embodiments, a dependable and educational method for handling missing data in datasets is Multiple Imputation by Chained Equations (MICE). The method fills in a missing value in a dataset by using a series of iterative prediction models. In every iteration, the other parameters in the dataset are used to impute each given variable. Until convergence seems to have been reached, these iterations should be repeated. With an incredibly high R2 value of 0.995, it appears to fit the data quite well. With an RMSE of 18.73 and an MAE of 9.00, the RFR yields superior results. The R2 value of 0.930 indicates that the model can explain. The strategy that performs the best and makes the fewest mistakes is gradient boosting. The code's functions included reading sensor data, processing it, and wirelessly sending it to a server in the cloud. Data from the sensors is periodically gathered and stored by the system. In order to forecast future pollution levels, machine learning algorithms were used to analyze the gathered data. The frequent use of temperature and humidity sensors may be due to their frequent correlation with sensor calibration. Other specific uses include solar radiation, ammonia, hydrocarbons, and volatile organic compounds. The specific conditions of the city to be monitored, such as the primary air pollutants present in the city area, typically determine which variables to include.
[0004] In some embodiments, the timeseries data is highly prone to autocorrelation and is tightly sequential. Before attempting to forecast the data, the time-series models would use the data to train and learn about its overall behavior. A regular observation will have a prediction that is as close to the actual value as is practical; an anomalous observation will have a forecast that is as far from the actual value as is practical; the outlier in the data can be found by examining the forecast errors. About 60% of the variance can be explained by the model, according to the R2 value of 0.60. Gradient Boosting outperforms all other models, achieving the lowest errors and the highest R2 value. These results demonstrate how well gradient boosting works in the regression problem. The Air Quality Index, or AQI for short, is the target variable in this code. It is a measurement of the quality of the air based on the concentration of airborne pollutants. The dataset has many records because it includes daily air quality readings for various cities. Between the network and the application layer is the service layer, which is in charge of offering services to "things" or applications. Cloud development is typically involved in the service layer's implementation. The systems examined in this study primarily describe the perception and network layers, the first two layers of the IoT architecture. As a result, the service layer was poorly described, with insufficient implementation details provided.
[0005] FIG. 2 illustrates a method for workflow of designed method according to an embodiment herein. In some embodiments, the proposal of gating systems such as Long Short-term Memory (LSTM), which allow nodes to ignore or pass memory if it is not being used while maintaining enough error to allow updates, significantly addressed the problem of vanishing gradients throughout Backpropagation Through Time (BPTT) updates. In every parameter, Random Forest Classifiers received the highest score, indicating strong classification performance. The Random Forest model's precision, recall, and F1 scores were all 0.972%, 0.972%, and 0.972%, respectively. This code has also removed the missing values from the dataset. Furthermore, a few of the dataset's features exhibit skewness, which may affect how well machine learning models perform. The code in this example predicts the AQI based on the concentration of pollutants using three distinct regression models: Random Forest, Linear Regression, and Decision Tree. The screened papers lacked sufficient information about network protocols to illustrate this study. Despite being an intriguing choice in a smart city setting, it is important to highlight that the cutting-edge V2I infrastructure has not been utilized much for air monitoring. The most popular technologies for transmitting sensed data are traditional ones like cellular and Wi-Fi.
[0006] In some embodiments, the data is made possible by a sequence of gates that receive the processed repetitive input. Additionally trained are weight values that pass input to gate summations. The addition of LSTM to the RNN model allows for the preservation of long-term dependencies in the data when working with sequences. These results provide useful insight into the performance of different algorithms, suggesting that RFC might be particularly well-suited to the classification task. When choosing efficient algorithms for similar professions in the future, researchers and practitioners might consider these findings. Whereas RMSE calculates the square root of the average of the squared differences between the actual and predicted values, MAE calculates the average absolute difference between the two. DK-based for prototypes that are implemented using hardware development kits, such as Arduino, Raspberry, and others, and specific purpose for prototypes that are developed using a hardware platform created especially for the intended use.
[0007] In some embodiments, the one step ahead can be predicted by a neural net with a single output unit. The neural net forecast model shown in the figure predicts series values in steps ahead of time by using words from past series values. The data series are normalized to [0,1] before being fed into the neural nets. Prior to computing evaluation criteria, predictions from the neural net output were converted to the original data scale. The dataset was gathered over the course of nearly five years from a densely populated area using IoT devices. The system frequently encountered disruptions as a result of the lengthy data collection series. However, premium cloud storage was not used for the data collection process, and it was only carried out in three locations. When the target variable has a wide range of values, RMSLE is used to calculate the root mean squared logarithmic error between the predicted and actual values. A statistical metric called R-squared shows how much of the variance in the dependent variable can be accounted for by the independent variables in the regression model. Cloud computing is significantly more popular than other options like edge and node. Because the processing units usually micro-controllers have very limited processing capacity, node processing is difficult. Although edge computing is a relatively new approach, its potential for these kinds of applications has not been fully investigated.
, Claims:1. A method for real-time environmental monitoring and air quality prediction using iot and machine learning in smart cities, wherein the method comprises;
detecting and measuring pollutants, such as particulate matter (PM2.5, PM10), nitrogen dioxide (NO₂), carbon monoxide (CO), and sulfur dioxide (SO₂), as well as temperature, humidity, and other factors that affect air quality;
collecting data from various locations roadsides, industrial areas, parks, and residential zones creating a comprehensive and granular map of the city's air quality conditions;
analyzing pollutant data over time to detect trends, identify pollution hotspots, and determine peak times for different pollutants;
using historical and real-time environmental data, machine learning algorithms are predicting future air quality levels, helping to anticipate periods of poor air quality and potential health risks;
integrating data from traffic, industry, and weather, the system is identifying sources of pollution and determining the impact of specific activities, such as traffic congestion or industrial operations, on air quality;
alerting residents and city authorities when air quality reaches unsafe levels;
optimizing traffic light signals, rerouting traffic, and suggesting temporary restrictions on industrial activities to reduce pollution levels during peak times;
providing tailored health recommendations to vulnerable populations, such as children, the elderly, and individuals with respiratory conditions;
improving air quality forecast models, making them more accurate and adaptive to changes in pollution sources and climate conditions; and
supporting urban planners in making data-driven decisions on green spaces, traffic flow, and industrial zoning, promoting a healthier urban environment for future generations.

Documents