An Approach for Dam Monitoring and Alerting System Using IoT

A dam is a barrier constructed across a river to control the flow of water. Dams are used for harvesting energy and further utilizing this energy for generating hydroelectric power. Dams are also used for protection against flooding conditions by controlling the flow of water in the river. Apart from this dams can release water to the river to facilitate agricultural activities like irrigation. The majority of the dams are still being manually controlled which is an inefficient and time-consuming way of controlling dams. Due to this, the mismanagement of dams happens which can lead to devastating damage. This can be evident from a few recent cases of dam failures like the Dhauliganga dam failure case of 2021, the Tiwari dam failure of 2019, and the Uttarakhand dam failure of 2013 where dam failure lead to flash floods causing severe property damages and loss of many lives. To solve this issue an IoT-based dam monitoring and alerting system has been proposed that can reduce the human intervention from the existing dam control system thereby making the system more efficient, and more accurate hence reducing the chances of dam failures. This system uses an efficient algorithm and industry-applicable software to detect flood conditions early using different water level checkpoints. The water level of the river is continuously monitored using ultrasonic sensors and this real-time data is used by the decision algorithm of the system to make decisions for controlling the dam. Once this data is collected node-red is used to represent this real-time data using graphical representations on a web-based dashboard. Also node-red is used to further send the sensor data to the ThingSpeak cloud. This will allow the dam operator to remotely monitor the dam working. The system uses the IFTTT platform to send alerts to the dam operator with a delay of less than 2 min whenever a flood condition triggers so that possible measures can be taken to control the situation. We have used a machine learning model in our system to categorize the water level of the river with an overall accuracy of 84%.