Dark Mode
Industrial E-paint Coating Process Data
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset originates from an industrial case study involving an electrophoresis painting plant. It focuses on E-paint coating processes, including the degradation progressions of multiple filters. The data was collected following the deployment of an Industrial Internet of Things (IIoT) system. The dataset's primary purpose is to facilitate the development and application of advanced data analytics methods, such as Deep Learning, for addressing manufacturing problems. It also aims to improve understanding of manufacturing production processes, particularly in areas like maintenance scheduling and continuous processing. This forms an initial step towards predictive maintenance. The dataset presents a challenge by combining two disparate data sources—manual inspection records spanning seven years and high-resolution IIoT sensor readings collected over fifteen days—to demonstrate how machine learning models can predict process conditions even with limited resources, addressing common issues of data quality and quantity in real-world manufacturing environments.
Columns
The dataset comprises nine columns, primarily focusing on sensor readings related to the ultrafiltration and circulation subsystems within the electrophoresis painting plant. All sensor readings, specifically pressure and temperature, have been normalised to a range of (0,1) to protect sensitive business information while maintaining data usability. It is important to note that all pressure sensors were normalised collectively.
- TIME: The timestamp of the recorded data, indicating the exact date and time.
- FM1: Flow meter 1 readings for the ultrafiltration subsystem.
- PE1: Pressure 1, representing the input pressure for the ultrafiltration subsystem.
- PE2: Pressure 2, indicating the output pressure for the ultrafiltration subsystem.
- PE3: Pressure 3, detailing the input pressure for the circulation subsystem.
- PE4: Pressure 4, showing the output pressure for the circulation subsystem.
- TP1: Temperature 1, measured at the paint tank.
- TP2: Temperature 2, measured at the radiator of the circulation subsystem. This feature was introduced with the IIoT system and is not available in manual inspection records.
- EPOCH: The epoch timestamp, providing a numerical representation of the timestamp.
Distribution
The dataset is typically provided in a CSV file format. The example sample,
iiot_30min_norm.csv, is approximately 79.85 kB in size and contains 9 columns and 720 records. This particular dataset is a unified collection of IIoT sensor readings, originally sampled every 10 seconds over 15 days, which have been down-sampled to a consistent rate of every 30 minutes per sample. A related dataset of manual inspection records, originally collected every 8 hours over seven years, was also up-sampled to this 30-minute interval. The specific time range for the IIoT sensor data provided in the sample spans from 6th July 2020 to 21st July 2020.Usage
This dataset is ideal for:
- Developing, testing, and applying advanced data analytics, including Deep Learning methods, to solve manufacturing challenges.
- Improving understanding of manufacturing production processes, particularly regarding maintenance scheduling and continuous operational efficiency.
- Predicting process conditions as a foundational step for implementing predictive maintenance strategies.
- Applying time-series forecasting techniques, such as Long Short-Term Memory (LSTM) networks, to manufacturing process data.
- Investigating methods like Transfer Learning to overcome typical real-world manufacturing dataset challenges, such as issues with data quality and quantity.
- Training machine learning models to predict process conditions effectively, especially when faced with limited project resources or insufficient historical data.
Coverage
The data pertains to an electrophoresis painting plant, which was established two decades ago. The time scope for the IIoT sensor readings dataset spans 15 days, specifically from 6th July 2020 to 21st July 2020, with readings unified to a 30-minute interval. For context, there is also a manual inspection record dataset covering a seven-year period. It is important to note that the high-resolution IIoT data covers a short period, potentially limiting the observation of equipment failures or maintenance activities. Conversely, the manual inspection records, while covering a longer duration, have lower quality due to infrequent measurements (maximum three times daily, excluding weekends and holidays) before up-sampling. The feature 'TP2' (Temperature 2) is only available from the IIoT system and not from the older manual inspection records.
License
CC BY-NC-SA 4.0
Who Can Use It
- Data scientists and machine learning engineers: To build and validate advanced analytical models for industrial applications, focusing on predictive maintenance and process optimisation.
- Manufacturing experts and process engineers: To gain deeper insights into their production lines, identify potential issues, and enhance maintenance scheduling.
- Researchers and academics: For studies into industrial IoT, Deep Learning applications in manufacturing, time-series analysis, and addressing data challenges in real-world settings.
- Students: As a practical case study for learning about industrial data analytics, sensor data processing, and machine learning model development for manufacturing.
Dataset Name Suggestions
- E-coating Plant Predictive Maintenance Data
- IIoT Electrophoresis Manufacturing Process Dataset
- Ultrafiltration System Sensor Readings for Analytics
- Industrial E-paint Coating Process Data
- Manufacturing Process Condition Prediction Dataset
Attributes
Original Data Source: Industrial E-paint Coating Process Data
QUALITY