UQPyL

Python 中的不确定性量化、校准与优化工具

  • 用统一的 Problem / ModelProblem 定义不确定性量化、校准与优化任务
  • 运行试验设计、敏感性分析、贝叶斯推断等方法
  • 为高成本仿真构建代理模型,并支持代理辅助优化
  • 使用基准问题测试和比较算法表现

Problem and Workflow

UQPyL 从可复用的问题定义开始,把采样、分析、推断、代理建模、校准和优化连接到同一套接口上。

Problem First
  1. 01

    定义 Problem

    把变量、边界、目标、约束或仿真模型封装成统一入口。

    Problem/ModelProblem
  2. 02

    运行方法

    把同一个 problem 传给 DOE、分析、优化、推断、校准或代理模型模块。

    DOE/Analysis/Optimization/Inference/Calibration/Surrogate
  3. 03

    查看结果

    读取标准结果对象,或查看保存下来的运行记录。

    verbose/log/save
Module MapUQPyL architecture map

六大功能模块

每个模块负责不确定性量化、校准与优化工作流中的一个核心环节。

Design of Experiment

01

Generate samples for model exploration, sensitivity analysis, inference, and surrogate model training.

Design of Experiment
LHSSobolSaltelli
查看

Surrogate Model

02

Approximate expensive models and accelerate analysis or optimization workflows.

Surrogate Model
KrigingGPRBF
查看

Sensitivity Analysis

03

Quantify how input uncertainties influence outputs and identify important variables.

Sensitivity Analysis
SobolFASTMorris
查看

Inference

04

Estimate uncertain parameters and posterior distributions with Bayesian sampling methods.

Inference
MHDEMCDREAM-ZS
查看

Calibration

05

Calibrate model parameters and support hydrological or external-model workflows.

Calibration
GLUESUFI2IES
查看

Optimization

06

Search for optimal decisions with single-objective, multi-objective, and surrogate-assisted algorithms.

Optimization
SCE-UANSGA-IIASMO
查看

应用案例

覆盖水文校准、代理辅助优化和完整不确定性工作流的代表性案例。

Dongjiang River Basin

Hydrological model calibration with sensitivity analysis and uncertainty evaluation.

Dongjiang River Basin

Surrogate-Assisted Optimization

Optimization under expensive simulation settings with adaptive surrogate modelling.

Surrogate-Assisted Optimization

Workflow Overview

A compact end-to-end view of how problems, modules, and workflows connect in practice.

Workflow Overview

从这里开始

把首页作为进入 UQPyL 文档的主入口。