| CPC G08B 21/10 (2013.01) | 8 Claims |
|
1. A method for a flood early warning in a smart city based on an Internet of Things (IoT), comprising:
obtaining precipitation information of a target area within a target time period through a meteorological management platform;
obtaining drainage capacity information of the target area through a water affairs platform;
determining regional ponding information of the target area within the target time period based on the precipitation information and the drainage capacity information;
generating a first flood early warning information of the target area based on the regional ponding information;
when the regional ponding information of the target area meets a first preset condition,
obtaining road network information of the target area through a traffic management platform;
obtaining historical regional ponding information of the target area and historical road ponding information of each road in the target area through the water affairs platform;
determining road ponding information of each road in the target area in the target time period through processing road network diagram structure data based on a road ponding information determination model, wherein the road network diagram structure data is constructed based on the road network information, the precipitation information, the drainage capacity information, the historical regional ponding information, and the historical road ponding information, nodes of the road network diagram structure data correspond to each intersection in the target area, attributes of the nodes reflect relevant feature of the corresponding intersection, the attributes of the nodes include a congestion prone feature, a drainage capacity feature, and a precipitation feature, edge of the road network diagram structure data corresponds to each road in the target area, attribute of the edge reflects relevant feature of the corresponding road, a direction of the edge is determined according to a altitude difference of intersections corresponding to two connected nodes, the attribute of the edge includes a precipitation feature, a drainage feature, a slope, a width, and a length of the corresponding road; wherein the road ponding information determination model is a graph neural network (GNN) model, the road ponding information output by the road ponding information determination model is represented by a ponding situation vector, and element value of the ponding situation vector represents a probability at each road ponding risk level;
determining ponding sections in the target area based on the road ponding information;
generating a second flood early warning information of the target area based on the ponding sections in the target area;
determining a first confidence of a flood section based on the probability of the road ponding risk level, wherein the first confidence degree represents the probability of flood disaster on the floow section;
determining a actual disaster situation of the flood section based on a comparison between monitoring information of the flood section and a normal image when the first confidence is less than a first threshold;
determining a second confidence based on a resolution of the monitoring information and a accuracy of a algorithm of the actual disaster situation, wherein the second confidence represents a accurate probability of the actual disaster situation; and
determining the actual disaster situation through manually retrieving the monitoring information of the flood section when the second confidence is less than a second threshold.
|