Skip to content

Latest commit

 

History

History
197 lines (182 loc) · 11.1 KB

battery.md

File metadata and controls

197 lines (182 loc) · 11.1 KB

battery

open course

SOH & SOC

  • A Neural Network Approach to Absolute State-of-Health Estimation in Electric Vehicles 📖
  • Advanced Machine Learning Approach for Lithium-Ion Battery State Estimation in Electric Vehicles 📖
  • An Online SOC and SOH Estimation Model for Lithium-Ion Batteries 📖
    • 计算 SOC 的三类方法：数据驱动方法，I-t 积分等；适应方法，KF 滤波等；混合方法
    • 计算 SOH，考虑电池的容量衰减和内阻增加
  • State of health estimation for lithium-ion battery by combing incremental capacity analysis with Gaussian process regression 📖
    • SOH 估算需考虑两个因素，容量和内阻
    • SOH 估算的方法有三类
      1. 经验或半经验模型，受不确定因素影响，有局限
      2. 电化学或物理模型，需要积分，计算复杂
      3. 数据驱动方式，有好的非线性，但需要高质量数据

anomaly detection

Overview

  • https://github.com/hoya012/awesome-anomaly-detection
  • https://github.com/rob-med/awesome-TS-anomaly-detection
  • https://github.com/zhuyiche/awesome-anomaly-detection
  • Survey on Anomaly Detection using Data Mining Techniques 📖
  • Novelty Detection in Learning Systems 📖
  • A Survey of Outlier Detection Methodologies 📖
  • DEEP LEARNING FOR ANOMALY DETECTION: A SURVEY 📖
  • Novelty Detection: A Review Part 1: Statistical Approaches 📖
  • Novelty Detection: A Review Part 2: Neural network based approaches 📖
  • A Comprehensive Survey of Data Mining-based Fraud Detection Research 📖
    • summary over 10 year's research
    • 2 criticisim： lack of data and lack of good methods
    • advocate for supervised learning
  • A Comprehensive Survey on Outlier Detection Methods 📖
  • Anomaly Detection : A Survey 📖
    • adding two classes :information theoretic and spectral techniques

AE

  • SAIFE: Unsupervised Wireless Spectrum Anomaly Detection with Interpretable Features
    • use adversarial Autoencoder
    • prefer reconstruction-base over prediction-based
    • localization via plotting x - x_hat, which also helps interpretation
  • Visual Anomaly Detection in Event Sequence Data 📖
    • use LSTM-VAE
    • score calculation method: using new data to calculate LOF(k-nearest neighbour) at latent space
    • use only visual comparison to facilitate interpretation
  • Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network 📖
    • propose OmniAnomaly, a GRU+VAE model
    • use POT to automatically select threshold
    • https://www.youtube.com/watch?v=ERb_itqarsE
    • 对于多维时序，要从整体层面而不是单个维度层面考虑异常检测，之前多维时序异常检测，要么用了 deterministic，要么忽略了 temporal dependency
    • 做了四个模型三个数据集的实验，用 F1 score 证明自己好
    • https://github.com/NetManAIOps/OmniAnomaly
  • Generative Probabilistic Novelty Detection with Adversarial Autoencoders 📖
  • Anomaly Detection with Robust Deep Autoencoders
  • Variational Inference for On-line Anomaly Detection in High-Dimensional Time Series 📖
    • propose RNN+AE
    • train on normal data, test on normal and anomaly
  • A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an LSTM-based Variational Autoencoder 📖
  • Variational Autoencoder based Anomaly Detection using Reconstruction Probability 📖
    • use reconstruction probability instead of reconstruction error
  • DEEP AUTOENCODING GAUSSIAN MIXTURE MODEL FOR UNSUPERVISED ANOMALY DETECTION 📖
  • Multidimensional Time Series Anomaly Detection: A GRU-based Gaussian Mixture Variational Autoencoder Approach 📖
    • propose GRU(to capture correlation)+GMM(as prior)+VAE
    • GMM param k=2, latent dimension=8
  • LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection 📖
    • propose LSTM+AE for multi sensor data
  • DEEP UNSUPERVISED CLUSTERING WITH GAUSSIAN MIXTURE VARIATIONAL AUTOENCODERS 📖
  • Unsupervised Anomaly Detection via Variational Auto-Encoder for Seasonal KPIs in Web Applications 📖
  • GMVAE
  • Adversarial VAE

GAN

Other Models

  • Learning Representations of Ultrahigh-dimensional Data for Random Distance-based Outlier Detection 📖
  • Feedforward Neural Network for Time Series Anomaly Detection :bok:
    • train on labeld raw time-series, supervised learning
    • time-series to vector process
  • Long Short Term Memory Networks for Anomaly Detection in Time Serie 📖
    • use stacked LSTM
    • train on only normal data and make prediction
    • eval precision，recall，F0.1
  • Time-Series Anomaly Detection Service at Microsoft 📖
    • propose Spectral Residual + CNN
  • Deep Anomaly Detection with Deviation Networks 📖
    • propose devNet, and end-to-end framework
    • semi-supervised training, use a few labeled data as prior
  • A Deep Neural Network for Unsupervised Anomaly Detection and Diagnosis in Multivariate Time Series Data 📖
    • propose multi-scale convolutional recurrent encoder-decoder
    • need both inter-sensor correlation and temporal dependency
  • Detecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholding 📖
  • Multivariate Industrial Time Series with Cyber-Attack Simulation: Fault Detection Using an LSTM-based Predictive Data Model 📖
    • Two stacked LSTM layer, MSE as loss, RMSProp as opt
    • use MSE threshold， 0.999 percentile empirical error
  • Detecting Anomalies in Space using Multivariate Convolutional LSTM with Mixtures of Probabilistic PCA 📖
    • on multi-channel/multivariate data
    • use CONV LSTM + MPPCA
  • A Data-Driven Health Monitoring Method for Satellite Housekeeping Data Based on Probabilistic Clustering and Dimensionality Reduction
    • use MPPCA model
    • not considering temporal dependence
    • use percentile to remove anomaly; use previous value to imputate; nomalize 0-1，use 99.9% and 0.1% as threshold
  • Battery Capacity Anomaly Detection and Data Fusion 📖
    • use kalman filter
  • Collective Anomaly Detection based on Long Short Term Memory Recurrent Neural Network 📖
    • use LSTM to train on normal data
    • use prediction error detect collective anomly
  • Adversarially Learned One-Class Classifier for Novelty Detection 📖
  • Online Detection of Unusual Events in Videos via Dynamic Sparse Coding 📖
    • beautifully written
    • use dynamic sparse coding (out-of-date model?)
  • A Symbolic Representation of Time Series, with Implications for Streaming Algorithms 📖
  • Modeling Extreme Events in Time Series Prediction 📖
    • optimized extreme value modeling via deep learning
    • propose extreme value loss and memory network
  • Adaptive-Halting Policy Network for Early Classification 📖
    • a model to handle both earliness and accuracy of classification of time series
    • propose LSTM to generate low-dimensional representation, a controller network
  • Anomaly Detection: Algorithms, Explanations, Applications 📖

Benchmark

  • Systematic Construction of Anomaly Detection Benchmarks from Real Data 📖
    • benchmark 要求：正常数据点来自真实世界；异常数据点来自真实世界而且语义上不同；需要很多数据；需要定义好异常问题，且系统性多样