Dark Mode
Filter Remaining Useful Life Dataset
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset focuses on the degradation process of filters, specifically the clogging that occurs when separating solid particles from gas. Its primary purpose is to enable the transition from preventive to predictive maintenance by allowing for the precise prediction of the remaining useful life (RUL) of these filters. The data originates from practice-relevant degradation processes, ideal for Prognostics and Health Management (PHM) applications. A key challenge addressed by this dataset is how to effectively utilise right-censored life data, which arises because filters are typically replaced after a fixed period regardless of their actual condition, meaning the end of their life is often unobserved in training records.
Columns
- Data_No: An identifier for individual test runs or life cycles.
- Differential_pressure: The measured pressure difference across the filter, a critical indicator of its condition. Filter failure is defined when this pressure exceeds 600 Pa.
- Flow_rate: The rate at which gas flows through the filter during testing.
- Time: The elapsed time during a specific test run.
- Dust_feed: The constant amount of dust fed into the system per unit time during a test run. This influences the maintenance interval and censoring time.
- Dust: The type of dust used in the test, for example, 'ISO 12103-1, A3 Medium Test Dust'.
- RUL: The Remaining Useful Life of the filter, which is the primary prediction target.
Distribution
The dataset is provided in various formats, including CSV files (
Test_Data_CSV.csv, Train_Data_CSV.csv) and MATLAB files (Data.mat, Train_Data_Uncensored.mat, and files for different particle size distributions). The Data.mat file is approximately 748.95 kB, and the overall version 8 of the data is about 9.36 MB. The dataset is structured into training and test data, with 50 life tests provided for each. The training data includes runs up to a periodic replacement interval, making many records right-censored, while the test data includes randomly right-censored run-to-failure measurements with the corresponding RUL as ground truth. A detailed description file, Preventive to Predictive Maintenance dataset.pdf, is also included.Usage
This dataset is ideally suited for:
- Developing and testing machine learning models for RUL prognosis in Prognostics and Health Management (PHM).
- Facilitating the transition from preventive to predictive maintenance strategies for replaceable parts like filters.
- Researching methods to effectively handle and make use of right-censored life data.
- Exploring theory-guided data science or informed machine learning approaches by integrating physical knowledge and models with data-driven methods.
Coverage
The data originates from a test bench performing automated life testing of filter media, specifically the clogging of filters when separating solid particles from gas. It employs dust complying with ISO standard 12103-1. The filter media used exclusively for these measurement samples is CC 600 G. The variations in filter lifetimes are primarily influenced by the type of dust, the flow rate, and manufacturing tolerances. The data situations represent practice-relevant scenarios, with various configurations of filter tests used to avoid carryover between different datasets uploaded by the creator.
License
Attribution 4.0 International (CC BY 4.0)
Who Can Use It
This dataset is valuable for:
- Machine learning engineers and data scientists focused on predictive analytics and remaining useful life estimation.
- Researchers and academics in the fields of Prognostics and Health Management (PHM), reliability engineering, and industrial data science.
- Maintenance managers and engineers looking to implement or improve predictive maintenance strategies in industrial settings.
- Students studying time series analysis, regression, and the application of machine learning to real-world industrial problems.
Dataset Name Suggestions
- Filter Remaining Useful Life Dataset
- Predictive Maintenance Filter Clogging Data
- Industrial Filter Prognostics Data
- Censored Filter Life Cycle Dataset
- PHM Filter Degradation Data
Attributes
Original Data Source: Filter Remaining Useful Life Dataset
QUALITY