Deep Learning for the Behavior of Linear Compressor Pistons
Deep learning–based sensorless control of linear compressor pistons with over 90% performance improvement.
2022.12 – 2023.12
·Motivation
- 80% of a refrigerator's total power consumption is consumed by the compressor.
- Cooling is determined by the distance and position of piston movement.
·Goal
Precision control of the piston and sensorless technology are required.
·Data
Build an automated data-collection system measuring current, voltage, and stroke.
·Methodology
- A four-layer Multi-Layer Perceptron (MLP) with drop-out.
- Contribution: more than 90% performance improvement.
