Machine Learning Techniques with IoT in Agriculture

Dam plays a crucial role in water security. A sustainable dam intends to balance a range of resources involves within a dam operation. Among the factors to maintain sustainability is to maintain and manage the water assets in dams. Water asset management in dams includes a process to ensure the planned maintenance can be conducted and assets such as pipes, pumps and motors can be mended, substituted, or upgraded when needed within the allocated budgetary. Nowadays, most water asset management systems collect and process data for data analysis and decision-making. Machine learning (ML) is an emerging concept applied to fulfill the requirement in engineering applications such as dam water researches. ML can analyze vast volumes of data and through an ML model built from algorithms, ML can learn, recognize and produce accurate results and analysis. The result brings meaningful insights for water asset management specifically to strategize the optimal solution based on the forecast or prediction. For example, a preventive maintenance for replacing water assets according to the prediction from the ML model. We will discuss the approaches of machine learning in recent dam water research and review the emerging issues to manage water assets in dams in this paper.

Section: Machine Learning Approachesmentioning

confidence: 99%

Machine Learning in Dam Water Research: An Overview of Applications and Approaches

Yahya¹,

Fazli²,

Sallehudin³

et al. 2020

“…The various IoT frameworks are studied for agriculture in the research work done by Sujeet More et al [22].…”

Section: Introductionmentioning

confidence: 99%

Cloud-Based Smart Water Quality Monitoring System using IoT Sensors and Machine Learning

Jha¹

2020

Low water quality is a major concern in urban as well as rural areas. Consumption of contaminated water leads to several health hazards. Early water quality detection can prevent most of such health-related issues. Parameters such as conductivity, pH, nitrate, biochemical oxygen demand, fecal coliform are significant parameters in deciding the quality of water. These parameters which are collected from groundwater samples at different places are highly correlated to each other. Therefore, machine learning algorithms are used for classification. The data collected from sensors are further analyzed using a cloud-based environment Ubidots to support distributed computing. The cloud environment is connected to display units and mobile devices. To predict the quality of water it is necessary to check the values associated with the quality attributes and for that reason, a decision tree classification model is used. The dataset is broken into subsets that have decision nodes and leaf nodes to decide classifications. The IoT based sensors are deployed in the water tank to measure the quality parameters which are further sent to the cloud. The proposed framework predicts the water quality and assesses the performance of the decision tree classifier. Decision Tree is used to infer decision rules based on various parameters read through sensors.

“…[2] Therefore, to reduce if not eliminate this problem, the inefficient fertilizer application practices among farmers must be addressed [3], thus the emergence of various intelligent farming practices. [4] [5] One of the important nutrients that improve plant growth and yield in rice and maize is Nitrogen (N). The N fertilizer is applied multiple times to ensure that the crop's N need is adequate during its growing stage.…”

Section: Introductionmentioning

confidence: 99%

Nitrogen Deficiency Level Assessment Device for Rice (Oryza sativa L.) and Maize (Zea mays L.) using Classification Algorithm-based Spectrophotometry

Tabal¹

2020

In this paper, a nitrogen deficiency level assessment device (NDLAD) for rice and maize is presented. The proposed device was based on the functionality of the 4-window panel Leaf Color Chart (LCC) for assessing the nitrogen content of rice and maize plants. The principle of spectrophotometry was implemented using a TCS3200 color sensor module along with a hardware-deployed nearest neighbor algorithm in an Arduino Nano microcontroller for leaf color classification. The objective of the NDLAD is to eliminate the subjective nature of using an LCC in assessing the nitrogen levels in rice and maize. Based on the tests performed, it was revealed that the proposed device can provide a faster and higher detection accuracy rate compared with using an LCC. The performance results make NDLAD a cheaper and promising alternative to other existing electronic crop nutrient assessment tools that are currently available in the market.