eIQ Time Series Studio Training: Build and Run Time Series ML Models on FRDM-MCXN947

[#1]

As machine learning (ML) continues to move closer to the edge, NXP’s eIQ Time Series Studio (TSS) offers developers a powerful, intuitive environment for building, training, and deploying time series models directly to NXP microcontrollers. 

This two-part hands-on training series walks developers through the complete workflow, from data import to model deployment using the FRDM-MCXN947 development board and the ML Vibro Sens Click sensor board.

---

Lab 1: Classification with Pre-Collected Data

The first part of the training introduces the fundamentals of eIQ Time Series Studio using a fan state classification example. Developers learn how to:

• Explore sample projects in TSS.
• Import and label time series data from a 3-axis accelerometer.
• Train a classification model to detect four fan states: On, Off, Clog, and Friction.
• Evaluate model performance using confusion matrices and emulation.
• Deploy the model to the FRDM-MCXN947 development board using MCUXpresso IDE.

This lab is ideal for getting started quickly with pre-collected data and understanding the end-to-end workflow of model creation and deployment.

---

Lab 2: Anomaly Detection with Real Sensor Data

Building on the foundation from Lab 1, the second part of the training introduces real-time data collection and anomaly detection. Developers use the ML Vibro Sens Click board to collect accelerometer data and train a model to detect abnormal vibrations.

Key steps include:

• Using the Data Logging utility in TSS to collect real sensor data.
• Cleaning and preparing datasets using the Data Operation utility.
• Training an anomaly detection model with custom data.
• Deploying and testing the model on the FRDM-MCXN947 board using a serial terminal interface.

This lab demonstrates how to build models tailored to your own environment and sensor setup, making it ideal for applications like predictive maintenance, vibration monitoring, and industrial anomaly detection.

---

Why eIQ Time Series Studio?

NXP’s eIQ Time Series Studio is part of the broader eIQ Toolkit and is designed specifically for time series data applications. It enables:

• Importing and labeling time series datasets from sensors like accelerometers.
• Training models using classical ML algorithms optimized for embedded systems.
• Testing models with unseen data to validate performance.
• Deploying models seamlessly to NXP microcontrollers using MCUXpresso IDE or VS Code.

Training requirements

• Hardware Setup: Uses the FRDM-MCXN947 board and ML Vibro Sens Click with FXLS8974CF accelerometer.
• Software Tools: eIQ Toolkit 1.15.1, MCUXpresso IDE v24, and MCUXpresso SDK 25.03 for FRDM-MCXN947.

---

Conclusion

This training provides a practical and accessible way to get started with time series ML on the FRDM-MCXN947. Whether you're just getting started or looking to build custom models with real-world data to monitor industrial equipment, detecting anomalies, or building predictive maintenance solutions, eIQ Time Series Studio equips you with the tools and knowledge to bring your ideas to life.