Dark Mode
Multi-Machine Manufacturing Anomaly and Throughput Data
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
Operational data from five industrial packaging machines provides a detailed foundation for identifying anomalies and performance losses in manufacturing processes. This collection, known as the Packaging Industry Anomaly DEtection (PIADE) dataset, tracks production intervals, alarm codes, and throughput metrics across 133 different types of alerts. By recording machine states such as downtime, performance loss, and idle time, the material offers a thorough look into equipment reliability and production efficiency over a two-year period.
Columns
- interval_start: The timestamp marking the beginning of the recorded production interval.
- equipment_ID: A unique identifier for the specific industrial machine, ranging from s_1 to s_5.
- alarm: An alarm code representing the active reason for a machine stop or alert.
- type: The operational state of the machine, categorised as production, performance loss, downtime, idle, or scheduled downtime.
- start: The precise starting time of the production segment in a numerical date-time format.
- end: The precise completion time of the production segment in a numerical date-time format.
- elapsed: The total duration of the production interval measured in seconds.
- pi: The count of input packages processed during the specific interval.
- po: The count of output packages successfully produced during the interval.
- speed: The operational speed of the machine, quantified as packages per hour.
Distribution
The information is delivered as a tabular CSV file titled
raw_data.csv with a file size of 49.34 MB. It consists of 429,394 valid records across 10 primary columns. The data exhibits high integrity with 100% validity and no missing or mismatched entries. This is a static resource with an expected update frequency of never and a usability rating of 10.00.Usage
This material is ideal for training anomaly detection algorithms to identify irregular machine behaviour in industrial settings. It supports predictive maintenance research, allowing users to model correlations between specific alarm codes and machine downtime. Additionally, the data can be applied to manufacturing analytics to calculate Overall Equipment Effectiveness (OEE) and identify bottlenecks in automated packaging lines.
Coverage
The scope focuses on five industrial packaging machines (s_1 to s_5) over a timeframe spanning 1st January 2020 to 2nd January 2022. Each machine has a specific recording window, with some starting in early 2020 and others in mid-to-late 2020. The records provide a high-resolution view of 133 distinct alert types and various operational states across hundreds of thousands of production intervals.
License
Attribution 4.0 International (CC BY 4.0)
Who Can Use It
Industrial data scientists can use these records to develop and benchmark predictive maintenance models for heavy machinery. Manufacturing engineers and plant managers may apply the throughput and speed data to optimise production workflows. Additionally, students and beginners in data science can utilise the dataset for practicing data cleaning and storytelling within a real-world industrial context.
Dataset Name Suggestions
- PIADE: Packaging Industry Anomaly DEtection Dataset
- Industrial Packaging Machine Performance and Alarm Records
- Multi-Machine Manufacturing Anomaly and Throughput Data
- Packaging Line Operational State and Anomaly Repository
Attributes
Original Data Source: Multi-Machine Manufacturing Anomaly and Throughput Data
QUALITYLoading...