Real-time flood disaster warning and dynamic decision-making method based on distributed optical fiber

Abstract

The application relates to a kind of flood disaster real-time early warning and dynamic decision-making methods based on distributed optical fiber, as follows: distributed optical fiber sensor is laid in key position of river channel, real-time data acquisition constitutes data set, and disaster grade is marked, the data set after marking is divided into training set, verification set and test set;The fusion weight of each key position measuring point is calculated, and then the data of each measuring point is weighted and spliced according to the fusion weight to generate a fusion spatial feature vector;A flood disaster prediction model is constructed, the model is iteratively trained to obtain a trained model;After the newly collected data is weighted and fused, it is input into the trained model, and the disaster grade prediction result is output, whether to perform early warning is judged according to the prediction result, and the corresponding disaster grade decision is generated according to the prediction result.The application can realize real-time prediction and dynamic decision-making of flood disaster by efficiently using the collected high-frequency, multi-measuring-point and unstructured data.

Images