基于增量学习的FD

[1]A New Incremental Learning for Bearing Fault Diagnosis Under Noisy Conditions Using Classification and Feature-Level Information

问题域

数据增量

思路方法

将current phase 和previous phase看做源域和目标域，引入迁移学习的思想

[2]Cross-Domain Class Incremental Broad Network for Continuous Diagnosis of Rotating Machinery Faults Under Variable Operating Conditions

问题域

类别增量

思路方案

（1）A novel domain-adaptation learning loss function is designed based on the similarity between the source and target domains, which enables the traditional broad network to handle the task of class increment with different distributions well
（2）A cross-domain CIL mechanism is devised, which effectively preserves the knowledge of previous classes without requiring replaying their data when learning new classes.

[3]A new feature boosting based continual learning method for bearing fault diagnosis with incremental fault types

问题域

类别增量

方法

（1）gradient boosting is employed to construct an initial fault diagnostic model
（2）new modules are continuously extended dynamically for the initial diagnostic model to fit the residuals between the actual label and the output of the initial diagnostic model
（3）to maintain the backbone of the fault diagnostic model as a single one, redundant parameters and feature dimensions are removed using an effective distillation strategy

[4] Imbalanced class incremental learning system: A task incremental diagnosis method for imbalanced industrial streaming data

问题域

class incremental learning without replaying old data， class imblance

思路方法

（1） a novel graph convolutional sparse autoencoder is firstly designed for the imbalanced dataset to extract feature information with large inter-class scatter
（2） a classification loss function is designed to enhance the classification decision boundary between majority and minority classes by utilizing the prior distribution information of the imbalanced data
（3） a novel imbalanced class incremental learning rule is derived to realize new class learning without replaying old class data

[5]Class-incremental continual learning model for plunger pump faults based on weight space meta-representation

问题域

故障类别增量

思路方法

（1） A continual learning base model based on WaveletKernelNet (WKN) with attention mechanism is proposed to extract representative features from multi-modal signals
（2） Implement a variational autoencoder-like (VAE-like) hypernetwork that learns WSMR from base model parameter spaces corresponding to various faults, conserving prior knowledge of previous tasks at a lower cost to prevent catastrophic forgetting. Hypernetwork allows model to learn multiple fault types simultaneously in a single task.
（3）VAE generates inference models for different health states by
sampling meta-representation (MR) distribution through decoders and completes fault continual learning after rapid knowledge
integration