Category: Kristina’s Status Reports

Recollected Normal Condition Data

• Ambient temperature shifted noticeably from previous data collection runs
• Repeated full normal data collection loop to collect data at current ambient temperature

Full Flush of Cooling System

• Coolant turned yellow and contained some kind of brown particles, likely residue from radiator
• Fully drained the loop, flushed several times, and replaced water

Retrained Model with New Dataset

• Updated preprocessing steps, retrained regression and autoencoder models using new data
• Updated model architecture to use only delta temperature features
  • Reduce model sensitivity to ambient drift
• Began integrating updated models into live inference loop

Schedule & Progress

• On schedule for final demo

Next Steps

• Finish testing live inference loop
• Complete final report and final video

Data Collection for Fault Cases

• Completed power fault synthetic test patterns
• Started reduced flow synthetic test patterns data collection loop

Regression Model for CPU Power Prediction

• Finalized feature set and preprocessing for regression model
• Validated regression model performance on synthetic traces and exported model for use in autoencoder

Autoencoder Anomaly Detection Model

• Implemented sliding window AE on RF output
• Tuned initial architecture (window size, latent dimension, dropout, batch size)
• Ran verification sweep on fault windows
• Found strong overlap between normal and fault windows causing inconsistent fault detection
• Developing event based detection: detect at least once per fault event rather than per-window

Final Presentation

• Started drafting final presentation slides

Schedule & Progress

• On schedule: working on tuning ML model

Next Steps

• AE tuning + final presentation slides

Learning Strategies

• ML autoencoder: Learned to build autoencoder for anomaly detection
  • Used Medium blogs, GitHub examples, and other people’s AE projects to understand typical structures and tuning strategies
  • Referenced scikit-learn documentation and online tutorials for scaling, windowing, PCA plotting, and general debugging
  • Already had background from ML classes but needed to apply to actual sensor data
• Testbed Assembly: Had previous experience with similar setup with Raspberry Pi 5 and sensors so integrating was familiar
  • Had to learn how to run long data collection loops
  • Communicate between Pi and Pico over UART
  • Learn to flash code and debug UART on Pico with online tutorials and datasheets
• Practical Skills:
  • Crimping connectors and assembling hardware setup
  • Used YouTube videos, forums, and online guides
• Overall, most new knowledge came from informal learning
  • Videos, online blog posts, GitHub repositories, datasheets, and example projects

Hardware Updates & Repairs 

• Replaced previous PSU with a higher-wattage supply to start full-load heater tests
• Fixed issue where screw heads completely de-threaded
  • Used pliers to remove and mount new PSU
• Verified stable operation of power output with new PSU

Data Collection & Testing

• Began running automated synthetic test patterns, including:
  • Ramp up and ramp down CPU power profiles
  • Step spikes
  • Holds
• Debugged issue where tests suddenly stopped mid-run
• Finished a complete sweep of synthetic data collection
• Collected real CPU power traces using HWiNFO on my laptop to simulate realistic user workloads
  • Imported traces into power traces and successfully began running them on testbed

Schedule & Progress

• On schedule

Next Steps

• Begin processing data
• Start building regression model to predict power output
• Start fault-condition runs to collect fault data for anomaly detection

Hardware Integration & Testing

• Finished full testbed assembly
• Started testing on actual hardware setup (pump, fan, heaters, sensors all active)
• Verified operation of all components under load

SSR Calibration

• Conducted oscilloscope testing to measure duty cycle vs. output power
• Derived a linear equation relation PWM duty value to wattage output
• Updated control code to use calculated duty values

Integration

• Finished integrating UART communication between Raspberry Pi 5 and Pico
• Verified that heartbeat and command messages correctly sent and received

Schedule & Progress

• On schedule
  • Debugging hardware issues and calibration tasks from last week have been resolved
• System now ready for automated data collection testing

Next Steps

• Write and test automated data collection routines on the Pi 5
• Begin extended test runs to log data continuously
• Start ML code development

Testbed Assembly

• Assembled the basic testbed and power routing
• Brainstormed alternative ways to mount pump

Schedule & Progress

• Slightly behind schedule due to fit issues with the pump mount
  • Mounting holes did not align
• Exploring alternative mounting solutions to get back on schedule

Next Steps

• Complete water loop assembly with tubing
• Begin SSR PWM control testing with Pico once loop is complete
• Add resistor and water block assembly for load simulation

UART Pi5 to Pico Communication Code

• Successfully set up UART communication between Pi5 and Pico with wiring and code
• Verified with test messages from Pi5 to Pico

Testbed Assembly

• Assembled fan and radiator subassembly to prepare for full assembly next week
• Mounted Pico and SSRs to 3d printed component mounts

Schedule & Progress

• Slightly behind schedule for full system assembly since we are waiting on the custom cut boards expected to arrive Monday
• Once they arrive, plan to finish assembling testbed early next week to catch up to schedule

Next Steps

• Complete final assembly of testbed and water loop
• Begin SSR testing with Pico

UART Pi5 to Pico Communication Code

• Began writing code for UART communication between Pi5 and Pico

Design Report and Research

• Focused on finishing the Design Report and polishing all sections for submission
• Strengthened justifications for system and model design tradeoffs

Schedule & Progress

• On schedule:
  • All off-the-shelf parts have arrived
  • Custom part dimensions verified and adjusted to match real measurements
  • Plan to send out custom cut parts after Fall Break
• Behind schedule:
  • UART Pi5-Pico communication delayed due to debugging
  • Minor delays from Pi5 setup and network configuration issues

Next Steps

• Begin setup and partial assembly of testbed
• Continue debugging UART communication
• Assist with sensor integration and PWM control code

Raspberry Pi Setup

• Set up Raspberry Pi 5 for future UART communication with Pi Pico code development
• Troubleshot several setup issues:
  • Broken mini-HDMI cable causing monitor display issues
  • Missing microSD reader, preventing Pi OS installation through the Pi Imager
  • SSH connection issues due to the wired ethernet network being different from CMU Secure wireless
• Successfully imaged Pi OS onto the microSD card and completed initial configuration at home before bringing the Pi back to campus, which resolved network issues

Parts Procurement

• Ordered off the shelf components listed in the BOM
• Currently waiting for parts to arrive before verifying part dimensions and fabricating laser cut mounting plates

Design Report and Research

• Worked on Design Report sections covering use case and requirement justifications
• Found IEEE Xplore papers related to anomaly detection to reference for style and technical wording in the Design Report

Schedule & Progress

• Currently slightly behind schedule due to:
  • Waiting for hardware parts to arrive before starting laser cut fabrication
  • Delays in Pi 5 setup and network access during initial configuration

Next Steps

• Begin laser cut fabrication once parts (specifically the pump) arrive
• Continue 3D printing remaining brackets and verify fits
• Start Pi 5 to Pico UART code development
• Finish Design Report as soon as possible to get feedback

Completed CAD Model

• Completed CAD model with water loop and other component models downloaded from online (e.g. SSRs, DC-DCs, etc.)
• Adjusted radiator + fan position to reduce board size
• Planned out and added power planes (24V power supply stepped down to 12V, 5V for fan/pump, servo)
• Began 3d printing some parts from CAD design

  

  

  

  

  

  

Block Diagram

• Finalized block diagram to add power supply and conversion
• Adjusted coolant temp sensor connection to separate sensor data collection from controls (fan/pump, valve, heater)
• Added software

  

BOM Draft

• Completed BOM, added case components like linear motion rods and connectors
• Added power components like DC-DC converter, power supply

Schedule & Progress

• Updated schedule to account for part lead times
• Need to order parts ASAP since fabrication for laser cut parts depend on dimensions of actual mounting holes

Next Steps

• Begin ordering off the shelf components from BOM
• Work on design report

Block Diagram Draft

• Made first draft of block diagram showing connections between sensors, Raspberry Pi 5, Pi Pico, SSRs, power resistors, servo valves, and cooling components

BOM Draft

• Made first draft of BOM listing major components based on block diagram and physical assembly, to be updated

CAD Models of Liquid Cooling Components

• Found online CAD models of radiator, fan, and water block
• Altered radiator model to better match actual part
• Created CAD models for pump and reservoir based on product photos and references

Initial CAD Assembly of Cooling Loop

• Built an initial CAD assembly of liquid cooling loop with component models
• Used linear motion rods and connector system for structural frame

Revised CAD Assembly of Cooling Loop

• Updated initial assembly to accommodate sensor placements, including radiator intake/exhaust temperature sensors and mounting space for Raspberry Pi 5 and supporting components

Schedule & Progress

• Currently on schedule
• CAD modeling, assembly, block diagram, and BOM align with Gantt chart

Next Steps

• Iterate on block diagram
• Expand BOM to include linear motion rod and connectors
• Start purchasing off the shelf parts in order to update CAD assembly with corrected dimensions and order 3D print/laser cut parts