ML Surrogate Model for Rapid Flood Map Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for predicting water flow and establishing flood maps are computationally intensive and require lengthy simulations, making them inefficient and time-consuming.
Innovation Solution
The development of a machine learning (ML) deluge service that uses a surrogate model to predict water flow and generate flood maps without solving equations, leveraging a trained ML algorithm on circa 10,000 simulations to achieve speeds 16× to 25× faster than traditional methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional deluge simulations are used to solve equations for each mesh element, then flood map accuracy is improved, but simulation time and computational cost increase significantly
Solution Approach 1:
The patent creates a surrogate model that copies the essential behavior of traditional deluge simulations without actually solving the governing equations. The surrogate model is trained on simulation data and can predict flood maps much faster while maintaining acceptable accuracy, directly addressing the contradiction between simulation speed and accuracy.
Solution Approach 2:
The patent replaces the mechanical/mathematical simulation process (solving differential equations numerically) with a machine learning-based prediction system. The ML model substitutes for the complex computational hydraulics, enabling rapid flood map generation without the heavy computational burden of traditional methods.
2Reliability
If traditional deluge simulations solve equations for each mesh element and time step, then flood prediction reliability is improved, but computational intensity and processing time worsen
Solution Approach 1:
The surrogate model copies the input-output relationship of traditional simulations without replicating their computational intensity. By training on pre-computed simulation data, the model captures the essential physics and geometry relationships, providing reliable predictions with minimal computational power required during actual use.
Solution Approach 2:
The patent performs the computationally intensive equation solving during the training phase rather than during actual flood prediction. The surrogate model is pre-trained on comprehensive simulation data, so that when real flood predictions are needed, only lightweight inference is required, dramatically reducing operational computational intensity.
3Measurement precision
If meshing ground models and solving equations for each element is performed, then flood map accuracy is improved, but device complexity and system requirements worsen
Solution Approach 1:
The patent simplifies the system by replacing complex equation-solving infrastructure with a machine learning inference system. The surrogate model encapsulates the complexity of traditional deluge simulations in its trained parameters, allowing users to obtain accurate flood maps without needing sophisticated computational hydraulics software or expertise.
Data Source
AI summary
A method and system provide for generating a stormwater overland flow map. Simulation inputs (ground surface data) and simulation outputs (stormwater overland flow maps) are obtained from a deluge simulation model that simulates where water will channel and accumulate on a surface. A convolutional neural network (CNN) is trained to approximate the simulation outputs of the deluge simulation model. The CNN is a sequence of CNN models that each represent a time step and each CNN model in the sequence takes CNN output from a previous CNN model as its input. The CNN output is a video output and a visual representation of stormwater overland flow over time. A new input is a first format is obtained. A collection of custom objects representing points of a grid are stored and includes z-values of points on the surface and stormwater controls. The grid is populated and then processed in/by the CNN.