HROlive/Applications-of-AI-for-Predictive-Maintenance ? reverse-engineered prompt

Build me a Jupyter notebook workshop for predictive maintenance with three hands on labs.

I want the notebooks to show how to take time series sensor data, clean and prepare it, then use it in a few different ways to predict equipment problems before they happen. One lab should train an XGBoost model for failure prediction, including a version that uses GPU acceleration if available. Another should build an LSTM model with Keras and TensorFlow to predict failures from sequences. The last one should cover anomaly detection with autoencoders, including both an LSTM autoencoder and a 1D convolutional autoencoder, so I can compare them when there is not much failure data.

Please make it easy to follow like a workshop, with clear explanations, results, and simple evaluation of each model. If you need current library details, look up the latest docs online. Also include a small assignment or practice section at the end so someone can check their understanding.