Background
Droughts and floods are increasing demographic, economic and climate change is increasing social vulnerability, tacit knowledge with aging people is disappearing and the sustainability gap is growing. People and society need reforms to adapt to the changes that can be brought about through experimentation and research.
Can artificial intelligence, combined with tacit and open knowledge, solve the world's water problems? Can we advise on access to clean drinking water? And can we tell when the algae situation is bad and we can’t swim, what about droughts and floods? What if we could help those who can’t afford to get expert help or those who can’t read? Where in all could artificial intelligence help the water sector?
Maa- ja vesitekniikan tuki ry. made it possible to start this work. The project supports the continuation of the Artificial Intelligence and IoT Water Risks and Water Resources Management (ÄlyVesi) project.
Aim
The aim is to study the utilization of artificial intelligence in water sector:
- In forecasting (water level): Does artificial intelligence replace traditional modeling methods? With open information, can anyone make a prediction?
- In Counseling (Virtual Assistant): Is the answer reliable? How many questions can be answered?
- In citizen detection (image recognition): Can observations be automatically categorized and facilitate the use or monitoring of Algae Watch type applications in general?
- In quality management (automated stations): Could deep learning and statistical models improve data quality?
- In Education (Virtual Teacher): Does Artificial Intelligence Affect Human Interaction - Basic Education? Quality?
- The development of an intelligent virtual assistant requires that an interface be created for the developed models, eg. image recognition, from which the model can be utilized by the virtual assistant.
Methods
The research uses, among other things, feedback neural network, rational flow modeling, natural language processing methods, classification, image recognition methods, guided teaching methods, both IoT water elevation positions and statistical models.
Invasive species in water the city of Lahti in July 2020 (200 training set, with GSD 3-5 mm) gave us about IoU 0.4.© Kuva: Lari Kaukonen.