A method and system for predicting direct economic loss of urban waterlogging

By using a two-stage LSTM model and fixed asset spatialization technology, the problems of computational time and insufficient accuracy in urban flooding prediction are solved. This enables rapid and accurate prediction of flood depth and economic losses, as well as the quantification of uncertainty, and is applicable to urban disaster prevention, mitigation, and planning decisions.

CN122243024APending Publication Date: 2026-06-19HENAN CHUSHANDIAN RESERVOIR IRRIGATION DISTRICT ENGINEERING CO LTD +4
0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HENAN CHUSHANDIAN RESERVOIR IRRIGATION DISTRICT ENGINEERING CO LTD
Filing Date
2026-02-24
Publication Date
2026-06-19

AI Technical Summary

⚠Technical Problem

Existing methods for predicting urban flooding are computationally time-consuming, lack accuracy, struggle to quantify the uncertainty of losses, fail to meet the needs of near-real-time decision-making, and lack precise handling of the spatialization of fixed assets.

⚗Method used

A two-stage LSTM model architecture is adopted, which combines fixed asset spatialization technology and weighted quantile loss function. The HEC-RAS model is used to simulate the water depth of urban flooding. By combining nighttime light data and water depth-loss rate curves, a training dataset is constructed to achieve accurate prediction of water depth of urban flooding and economic losses.

🎯Benefits of technology

It achieves efficient prediction of water depth in urban flooding, improves computational efficiency by 244 times, takes only 0.054 hours to predict direct economic losses, has a loss prediction error of 15.28%, has near real-time assessment capabilities, and quantifies the uncertainty of losses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
Patent Text Reader

Abstract

This invention discloses a method and system for predicting direct economic losses from urban flooding. To address the problems of low accuracy in predicting urban flooding disaster losses and the lack of a closed-loop assessment system, this method first integrates multi-source data such as historical rainfall, DEM (Dual Earth Scale), land type, and fixed assets. It then simulates time-series flooding depth using HEC-RAS (Hyper-Earth Scale-Range Analyzer) and spatializes fixed assets using a perpetual inventory method and nighttime light data. Next, a two-stage LSTM model is constructed. The first stage predicts water depth, and the second stage integrates parameters such as water depth, asset value, and land type, using a weighted quantile loss function to output the loss prediction interval. Finally, a five-level disaster quantification standard is established, completing grid-level and regional-level assessments. The system achieves a prediction efficiency of 0.063 hours per event, a loss prediction error of 15.28%, and accurate level determination, providing scientific decision-making support for urban emergency management, drainage planning, and disaster prevention and mitigation.
Need to check novelty before this filing date? Find Prior Art