Edge Based Predictive Maintenance Using a TinyML Temporal Convolutional Transformer for NASA Turbofan Remaining Useful Life Estimation

Keyphrases

• Predictive Maintenance 100%
• Turbofan 100%
• Remaining Life Prediction 100%
• Edge-aware 100%
• TinyML 100%
• Temporal Convolutional Transformer 100%
• Microcontroller 75%
• STM32 50%
• Maintenance Scheduling 25%
• Prediction Accuracy 25%
• Deep Learning Model 25%
• Turbofan Engine 25%
• Benchmark Dataset 25%
• Computational Requirements 25%
• Equipment Condition 25%
• Resource-constrained 25%
• Financial Loss 25%
• Transformer Architecture 25%
• Memory Footprint 25%
• Memory Constraints 25%
• Embedded Platform 25%
• Deployable 25%
• Sequence Modeling 25%
• Engine Degradation 25%
• Pattern Extraction 25%
• Extreme Edge 25%
• TensorFlow Lite 25%
• Remaining Useful Life 25%
• Inference Latency 25%
• Latency Constraints 25%
• Temporal Convolutional Network 25%
• Maintenance Framework 25%
• Self-attention Mechanism 25%
• Modern Industrial System 25%
• Lightweight Self-attention 25%
• Operational Losses 25%
• Cloud Based Computation 25%
• Unplanned Downtime 25%
• Long-range Dependency Modeling 25%
• Quantized Model 25%
• Degradation Pattern 25%
• Transformer-based Methods 25%
• Service Interval 25%
• Competitive Prediction 25%
• Local Degradation 25%

Engineering

• NASA 100%
• Predictive Maintenance 100%
• Remaining Useful Life 100%
• Tiny Machine Learning 100%
• Microcontroller 75%
• Turbofan Engine 25%
• Industrial System 25%
• Recurrent 25%
• Self-Attention Mechanism 25%
• Constrainedness 25%
• Critical Component 25%
• Deep Learning Model 25%
• Memory Footprint 25%
• Prediction Accuracy 25%

Computer Science

• Microcontroller 100%
• Deep Learning Model 33%
• Prediction Accuracy 33%
• Industrial System 33%
• Range Dependency 33%
• Memory Footprint 33%
• Self-Attention Mechanism 33%
• Embedded Platform 33%
• Extraction Pattern 33%
• TensorFlow 33%
• Convolution 33%
• Unplanned Downtime 33%