SCANIA Component X dataset: a real-world multivariate time series dataset for predictive maintenance

Area of Science:

• Engineering
• Data Science
• Machine Learning

Background:

• Predictive maintenance relies on comprehensive datasets, which are scarce, especially in time series format.
• Existing real-world data for component failure prediction is limited, hindering research and development.

Purpose of the Study:

• To introduce a unique, real-world multivariate time series dataset for predictive maintenance applications.
• To provide a standardized benchmark for the predictive maintenance field to foster reproducible research.

Main Methods:

• Collection of operational data, repair records, and component specifications from a fleet of SCANIA trucks.
• Anonymization of data from a single engine component (Component X) to ensure confidentiality.
• Inclusion of diverse features such as histograms and numerical counters, alongside temporal information.

Main Results:

• A comprehensive, real-world multivariate time series dataset specifically for Component X.
• The dataset is suitable for various machine learning tasks: classification, regression, survival analysis, and anomaly detection.
• Facilitation of research using real-world data from a major automotive manufacturer.

Conclusions:

• The released dataset addresses the critical need for real-world data in predictive maintenance research.
• This resource will enable researchers to develop and validate advanced machine learning models for improved component failure prediction.
• Establishes a benchmark for future studies in the predictive maintenance domain.