Citation

Kok, Adrian Eng Hock (2024) Real time metrological grade monitoring stations for MASS Deployment and Analytics with AI Algorithms. Masters thesis, Multimedia University.

Full text not available from this repository.

Abstract

Weather conditions in Malaysia are not only limited to drought and monsoon seasons but we are also strategically located where our neighbouring countries practices open air burning for their crops cycle every year. This will result in heavy pollution levels when the wind direction is unfortunately blown towards this area. The frequent experience on severe haze occurs almost bi-yearly. Measuring and reporting of Air Pollution Index (API) is limited to the number of Metrological stations in the country. By the time our country alerts the people on those hazardous particles, it will be too late and slow to react upon to put contingency plans in place. Precise and accurate measurement of the API index is another open question that is not transparent to the public as the readings or warning signs of the Haze are only a handful such as: “Good”, “Moderate”, “Unhealthy”, “Very Unhealthy” and “Hazardous”. As can be seen and read, this is qualitative information and is quite vague. The research will bring a step closer to enable monitoring of air pollutions particularly on the particle measurements with accuracy of metrological stations. The Internet of Things (IoT) connected device hardware will be designed and constructed to be able to deploy easily and universally. The sensor protocols and data formatting adapt to standards ensuring compatibilities. Connected sensors data will be sent to cloud and processed with artificial intelligence algorithms to make better sense of data gathered alongside predictive algorithms than just monitoring. The research method conducted into two main phases. The heart of the research is the design, development, implementation and deployment of the IoTs on site. Total deployment of the IoT sensors is 36 systems. The second phase is concentrated on the data collection, analytics and adapting machine learning algorithms for prediction of the particle measurement data. The presented results demonstrated the cost-effective particle sensors used are correlated with the metrological standard with correlation of 0.85 and root mean square error (RMSE) of 10.6. Lastly the neural network type machine learning is used for prediction of the air quality data with an RMSE of 9.72 using a year worth of data generated from the IoT systems deployed.

Item Type:	Thesis (Masters)
Additional Information:	Call No.: Q335 .K65 2024
Uncontrolled Keywords:	Artificial intelligence
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	03 Feb 2025 07:00
Last Modified:	03 Feb 2025 07:00
URII:	http://shdl.mmu.edu.my/id/eprint/13349

Downloads

Edit (login required)