Dark Mode
Industrial Sensor Faults Time Series
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset is designed to support fault detection and diagnosis in complex mechatronic and industrial automation systems. It addresses the growing need for diagnosis modules in environments where systems are becoming increasingly intricate and operate under uncertain conditions. The primary goal is to help identify abnormal functioning from sensor data, which may be noisy or corrupt, to limit the severe consequences of system failures on human life and assets. It is particularly useful for applying machine learning and deep learning techniques to challenges in industrial diagnostics, enabling better performance, productivity, and system reliability.
Columns
- Timestamp: Represents the time at which a sensor measurement was recorded.
- SensorId: A unique identifier for each sensor. For example,
SensorID = 1corresponds to a PT100 temperature sensor operating in an industrial setting with dust and vibrations. - Value: The measurement taken by the sensor at the specified timestamp. During the measurement series, sensors may be disconnected or in a state of failure.
Distribution
The dataset is provided as a CSV file, named
sensor-fault-detection.csv, with a size of 2.58 MB. It comprises time series measurements and contains 62,629 records across its three columns: Timestamp, SensorId, and Value.Usage
This dataset is ideal for:
- Developing fault detection analytic components.
- Creating models to identify abnormal operation in industrial processes from sensor data.
- Locating and pinpointing the causes of system failures, leading to informed corrective actions or system reconfiguration.
- Applying traditional machine learning algorithms such as Support Vector Machines (SVM), Artificial Neural Networks (ANN), Fuzzy Neural Networks (FNN), Decision Trees (DT), and Bayesian Networks (BN) for diagnosis.
- Exploring novel Deep Learning algorithms, including Denoising Stacked Auto-Encoders, Long Short-Term Memory Networks, and Self-Attentive Convolutional Neural Networks, for advanced fault detection problems.
Coverage
The data originates from an industrial environment, specifically involving a PT100 temperature sensor operating with dust and vibrations. The measurements are time series data collected "over months," reflecting sensor states, including instances where a sensor might be disconnected or failing. The dataset was last modified on March 7, 2018.
License
CC0: Public Domain
Who Can Use It
This dataset is particularly useful for:
- Data scientists and machine learning engineers focusing on predictive maintenance and anomaly detection in industrial settings.
- Researchers in automation, control systems, and industrial IoT seeking real-world sensor data for algorithm development.
- System integrators and engineers aiming to build robust diagnosis modules for complex mechatronic systems.
- Anyone interested in applying advanced analytical methods to improve industrial plant performance and prevent catastrophic failures.
Dataset Name Suggestions
- Industrial Sensor Faults Time Series
- Automation System Anomaly Data
- Mechatronic Sensor Failure Dataset
- Predictive Maintenance Sensor Logs
- Smart Industry Fault Diagnostics
Attributes
Original Data Source: Industrial Sensor Faults Time Series
QUALITY