It's nearly impossible to get these datasets due to their commercially sensitive nature. However NASA has made available some sensor datasets from large civil aircraft with some associated faults.

My approach was to simulate the aircraft and the faults.

also you will find most of the datasets are not maintenance focuses but FDIE type(fault focussed). Thus there is a lack of maintenance history and usage information which means it is almost impossible to generate proper context for the data.

You mind find this URL to be useful:

http://www.reliableplant.com/Read/19701/data-analysis-predictive-maintenance

Hi

You can find these data in the links below:

Nonelectronic Parts Reliability Data

http://www.theriac.org/riacapps/search/?mode=displayresult&id=787

Electronic Parts Reliability Data

http://www.theriac.org/riacapps/search/?mode=displayresult&id=799

But you have to pay to access it.

 Hi Laerte,  Unfortunately both those links are 404 "Not found"

Esma, try the Turbofan Engine Degradation Simulation Data Set, from NASA:

https://ti.arc.nasa.gov/tech/dash/pcoe/prognostic-data-repository/#turbofan

Returning to the interesting Esma question...

Any open data-set to work with "predictive maintenance" tool?

Hi All,

Beside the popular Turbofan engine data (C-MAPSS), there is another predictive maintenance public dataset called Lithium-ion battery aging data (Link: https://c3.nasa.gov/dashlink/resources/133/)

Thank you very much because unfortunately the popular Turbofan engine data (C-MAPSS) is not accesible now :(

Dear Pedro,

The Turbofan engine data (C-MAPSS) is available under the following link: (https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/)

Thank you very much :)

Hi All,

for those of you whom looking for Big Dataset in Predictive Maintenance, you can check a 1.3 GB public data published by MS Azure Blob Storage under the following link: (https://pysparksampledata.blob.core.windows.net/sampledata/sampledata.csv). A short description of the data can be accessed here (https://github.com/Azure/PySpark-Predictive-Maintenance)

I'm looking for some Multivariate Time Series Anomaly Detection Datasets with label (normal or anormal) for supervised anomaly detection task. Please answer here - https://www.researchgate.net/post/Request_Multivariate_Time_Series_Anomaly_Detection_Dataset_with_label

Hey Jean C. Salazar , https://www.kaggle.com/c/predictive-maintenance1/data is not accessible to all. [For me it says - " This is a limited-participation competition. Only invited users may participate. "] Could you share it?

Hi All,

The Telemetry and related datasets used as an example of Predictive Maintenance with Microsoft Azure can be found at this link: https://github.com/Microsoft/AMLWorkshop/tree/master/Data

Complete tutorial is at this location: https://gallery.azure.ai/Notebook/Predictive-Maintenance-Modelling-Guide-R-Notebook-1

It can be combined an produce working example.

Regards

In :

"Doganaksoy, N., Nelson, W., 1991, “A Method and

Computer Program MCFDIFF to Compare Two Samples

of Repair Data,” TIS report 91CRD172, General Electric

Company Research and Development, Schenectady, NY."

there is a set of data of recurrent failure events for locomotive braking grids. This report is not easy to be found. The data though is also reproduced in a research paper of mine (see Appendix B) ---> Article Numerical Estimation of Expected Number of Failures for Repa...

Does somebody know a maintenance data set including comments from operators? I'm interested in text mining applied to maintenance duration estimation. However, I could not find any data set with this feature.

Thank you!

Ball bearing test data for normal and faulty bearings from Case Western Reserve Univ.: https://csegroups.case.edu/bearingdatacenter/pages/download-data-file

Good contribution from all the people here.

I would like to add another one:

Backblaze Hardrive Data (https://www.backblaze.com/b2/hard-drive-test-data.html)

Nevertheless, most of the datasets cited above are well known in the academic literature. I don't know up to what extend they can be considered as "real use cases". "Real" datasets from industry are by far so complete. Most of the time it's difficult (if not imposible) to have failure records from machines because the are not allowed to run to failure in real conditions. On top of that we have to work with a lot of noise from regular maintenance activities and imprecise maintenance WO imputations ... real life is tough

http://data-acoustics.com/

https://github.com/nagdevAmruthnath/Predictive-Maintenance

Hi everybody,

you should also consider Google's new service "Dataset Search" (https://toolbox.google.com/datasetsearch) as it proves to be a useful starting point for spotting appropriate datasets for your individual research demands.

Kind regards,

Patrick

Dear Ravi Pandit

You can find Wind Turbine High-Speed Bearing Prognosis Data and its example for predictive maintenance on the following URL:

https://github.com/mathworks/WindTurbineHighSpeedBearingPrognosis-Data

You can see the dataset from NASA Data Repository https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/

Hi,

here is a dataset that we generate for unbalance loads through vibration signals:

Conference Paper A Case Study for Problem-based Learning Education in Fault D...

Recently published two online articles about Predictive Maintenance, which include datasets as well complete implementation of:

1. data transformation (part I) and ( https://www.codeproject.com/Articles/5252988/Predictive-Maintenance-on-NET-Platform )

2. machine learning (part II) https://www.codeproject.com/Articles/5260939/Building-Predictive-Maintenance-Model-Using-ML-NET .

You can also check IEEE PHM Data Challenge http://eng.fclab.fr/ieee-phm-2014-data-challenge/

Another source: PHM Society Data Challenge https://www.phmsociety.org/events/conference/phm/18/data-challenge

Bosch also has its data challenge https://www.kaggle.com/c/bosch-production-line-performance

Dear Esma,

The reality is pretty much NO, except for internal company data that may or may not be released or shared. Even then, they serve primarily as examples. That may be fine for some of the predictive algorithm exercises already provided here. But that is far from the industrial objective of improved performance! ONLY the careful and rigorous collection of data from your own machines and processes can really give you the data that you need to develop a realistic Predictive Maintenance process. There are no "off the shelf" databases that will define the operating characteristics of any machine in YOUR situation. At best (and these are often available from manufacturers) you may find averaged, normalized or otherwise aggregated wear data and nominal specifications for any machine part or operation.

The recommendation by Laerte Oliveira had two 404 Links. However, the web archive project has our backs.

https://web.archive.org/web/20110703122943/http://theriac.org/productsandservices/products/ReliabilityData.php

Although the old page was found by Daniel Roßburg , I noticed that the actual publications (IF available) date back to 1995-97. That may be sufficient for general modeling needs, but it will not include any current information (including improvements). Use it with caution. You may use the sets (payment required) to develop your own models into which you can put data from current operations.

Completely agree with mr Cooke . No mather what sales engineer says about predictive maintenance softwares and algorythms main task is to provide Your own database of values for training and reference . In most cases you have to give some own assumed relations to have usefull and good prediction .

Hi,

you can check https://biaowang.tech/xjtu-sy-bearing-datasets/, a dataset containing 15 run-to-failure bearing degradation signals is provided.

can be useful for you :

https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/

A new Data Set is available for Data Challenge from Foxconn Industrial AI. https://iaiinstitute.com/competitions/17/eventData and the relevant link https://www.iaiinstitute.com/competitions/17/eventIntroduction

Access denied.

Dear Esma for predictive analytics the most important is know your data. Searching for some data set is not a good strategy. Pilot projects with Industry is my recommendation.

Check out this research:

https://www.researchgate.net/publication/344136404_Real-time_Data_Analytics_Edge_Computing_Application_for_Industry_40_The_Mahalanobis-Taguchi_Approach

Hamzeh Soltanali sorry for my late reply. What I mean by "comments from operators" is a dataset that would be used for Natural Language Processing: it would contain the description of the symptoms of a failure (given by the operator) and such description would have some corresponding variables that would be used as outputs (causes, criticality, workload required to solve the issue, etc.)

If somebody knows something like this, I'd be glad to know!

Thank you,

J.