The AIFD MotorFault GUI is a graphical user interface designed to aid in the intelligent diagnosis of stator faults in induction motors. This project leverages vibration signals and machine learning to provide an end-to-end solution for motor fault detection and severity prediction. The GUI allows users to upload motor data and receive detailed fault diagnosis results without requiring any coding knowledge. App link: AIFD MotorFault GUI Paper link : Research Paper
- User-Friendly Interface: Built with Streamlit, the GUI enables easy dataset uploads, exploratory data analysis (EDA) visualization, and interaction with prediction results.
- Automated Model Selection: Utilizes AutoKeras to automatically select and optimize the best deep learning model for the given dataset.
- Explainable AI: Incorporates Explainable AI techniques to improve model transparency and trustworthiness.
- High Accuracy: Achieves 99.81% accuracy in stator fault detection and severity prediction.
- Python 3.10
-
Clone the repository:
git clone [email protected]:patelharsh21/stator-motor-deploy-streamlit.git
-
Install the required dependencies:
pip install -r requirements.txt
-
Run the Streamlit application:
streamlit run app.py
-
Open your web browser and go to
http://localhost:8501
. -
Upload your motor vibration data through the provided interface. A sample vibration data file is provided in the repo.
-
View the EDA results and interact with the fault diagnosis predictions.
- Data Collection: Vibration signals from the stator motor are collected using a Brüel & Kjær (B&K) accelerometer.
- Preprocessing: The data undergoes preprocessing, including invalid sample removal and normalization.
- Feature Extraction: Various time-domain and frequency-domain features are extracted from the vibration data.
- Model Training: An AutoKeras model is trained on the processed data to identify the optimal model for fault diagnosis.
- Prediction and Explanation: The trained model predicts stator faults and their severity, with Explainable AI providing insights into the model's decision-making process.
- Jitendra Kumar Dewangan (IIIT Naya Raipur)
- Harsh Patel (IIIT Naya Raipur)
- Aparna Sinha (IIIT Naya Raipur)
- Debanjan Das (IIT Kharagpur)