Deep learning-based anomaly detection using one-dimensional convolutional neural networks (1D CNN) in machine centers...

Area of Science:

• Engineering
• Computer Science
• Artificial Intelligence

Background:

• Computer Numerical Control (CNC) and Machine Center (MCT) machines are vital in modern manufacturing.
• Predictive maintenance is crucial for reducing downtime and costs in these automated systems.
• Existing fault detection methods often lack the precision required for early intervention.

Purpose of the Study:

• To develop and evaluate a novel deep learning model for early fault detection in MCT machines.
• To compare the performance of the proposed model against traditional machine learning and other deep learning techniques.
• To demonstrate the model's suitability for automating fault detection in manufacturing environments.

Main Methods:

• Collected sensor data from CNC/MCT machines.
• Applied data preprocessing techniques to prepare the dataset.
• Developed and implemented a 1D Convolutional Neural Network (CNN) model for fault classification.

Main Results:

• The 1D CNN model achieved the highest accuracy (91.57%) among all tested models.
• It outperformed traditional classifiers like Random Forest and Support Vector Machines.
• The model also surpassed other deep learning models, including LSTM, in precision, recall, and F-1 scores.

Conclusions:

• The 1D CNN model is highly effective for early fault detection and classification in MCT machines.
• This approach offers significant advantages for small manufacturing companies seeking to enhance productivity and safety.
• The model's performance supports proactive maintenance strategies, potentially revolutionizing the manufacturing industry.