Files
myWiki/concepts/non-stationary-time-series.md

31 lines
1.1 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
title: "Non-stationary Time Series非平稳时间序列"
created: 2026-05-11
updated: 2026-05-11
type: concept
tags: [time-series, statistics, machine-learning]
sources: [[liu-koopa-2023]]
---
# Non-stationary Time Series非平稳时间序列
## 定义
非平稳时间序列是指统计特性(均值、方差、协方差)或时间依赖模式随时间变化的时间序列。这是真实世界数据的普遍特征——天气、金融、能耗等几乎都是非平稳的。
## 对深度学习的挑战
- **分布迁移**:训练窗口和推理窗口的数据分布可能截然不同
- **模型泛化困难**:模型学到的模式在分布变化后失效
- **传统应对**:差分、归一化等预处理,但会丢失信息
## Koopa 的处理方式
不将非平稳视为需要消除的"噪声",而是通过 [[fourier-filter-dynamics|Fourier Filter]] 将其**显式解耦**为时变和时不变分量,分别用不同的 [[koopman-predictor|Koopman 预测器]] 处理。
## 相关概念
- [[time-variant-dynamics|时变动力学]]
- [[fourier-filter-dynamics|Fourier Filter]]
- [[liu-koopa-2023|Koopa]]