FLUD-TECHNOLOGY STRENGTHS

FLUD is a state-of-the-art urban flood monitoring and forecasting platform which relies on the following technologies in order to deliver its benefits

Advanced Computational Intelligence

Accumulated Knowledge from a variety of diverse inputs

Effective visualization through state of the art GUI and VR environments

Robust sensor design

Warning and optimized recommendations

ADVANCED COMPUTATIONAL INTELLIGENCE

FLUD utilizes a novel approach that is based on deep learning techniques for spatial temporal modeling. This approach is the base for developing intelligent Big Data analysis and decision support systems for deployment in a variety of industrial and commercial application areas. IC approach executes two main functions enabling users to make more informed decisions regarding matters of importance:

Accurate analysis and modeling of historical data Precise forecasting of future events based on the analysis.

IC novel spatial temporal approach is capable of studying complex data in terms of both space and time, thus enabling users to gain better insights about cause-effect relationships. This approach analyzes and models Big Data in real-time, in competitive computational times in order to forecast future outcomes and predict abnormalities in the data, based on the analysis of deviation patterns. IC approach is a robust methodology able to handle uncertainties and noise affecting a process, and features reduced processing times, and increased efficiency and productivity for users.

For FLUD, IC deep learning approach will be combined with state of the art Fuzzy Logic (FL) techniques in order to provide accurate and timely predictions. FL enables FLUD to incorporate expert opinion into the decision-making process of the model. FL is able to combine various input sources, quantify qualitative information, and account for uncertainty in providing accurate forecast and classification results. More importantly, FL is able to deliver its results fast. As it is widely acknowledged FL can produce the same results as other more complex techniques but faster.

ACCUMULATED KNOWLEDGE FROM A VARIETY OF DIVERSE INPUTS

The application of IC novel CI techniques enable FLUD to incorporate and account for information from a variety of input sources including:

  • Weather data (radar data, humidity, barometric pressure, temperature, solar radiation
  • Expert opinion concerning threshold values and specific characteristics of the area of deployment (building density, garbage accumulation, and others)
  • Sensory input (water flow and volume)
  • Morphology of the area of deployment

EFFECTIVE VISUALIZATION THROUGH STATE OF THE ART GUI AND VR ENVIRONMENTS

FLUD platform enables its user to visualize effectively the incoming information and the forecasting results generated by the computational models.

FLUD also includes a virtual construct of the area of deployment where the end user is able to navigate in the UTM area, monitor the rain as it unfolds, and be informed for any warnings related to increased urban flooding probability.

The following videos show draft versions of FLUD’s VR control environment, and UTM’s virtual construct.

ROBUST SENSOR EQUIPMENT DESIGN

During extreme flooding phenomena sensory equipment is tested to its limits. Therefore, as part of the FLUD project the consortium has designed custom sensors able to collect and transmit data under extreme weather conditions.

  • A custom rain gauge for monitoring water volume
  • A custom water flow sensor

WARNING AND OPTIMIZED RECOMMENDATIONS

During extreme weather phenomena the mobile network is down so there has to be alternative ways for warning the public. FLUD takes this reality into account and utilizes new communication technologies and protocols to deliver timely warning to the public Given data related to the local authorities are provided FLUD is also able to provide optimized recommendation related to the allocation of human and public resources to effectively tackle a hazardous developing situation. FLUD achieves this through the deployment of advanced optimization algorithms. These algorithms mimic the theory of evolution to calculate the parameters that provide the optimal solution for a given problem.