ML-Based Power Supply Unit Control for Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Controlling Power Supply Units (PSUs) in communications networks is labor-intensive and requires significant human input, with existing methods not optimizing individual PSU efficiency effectively, leading to potential performance degradation and maintenance challenges.
Innovation Solution
Implementing a system that measures PSU properties and uses a machine learning (ML) agent to process these measurements, predicting the effects of suggested actions and transmitting only a subset of actions to improve PSU efficiency and reduce degradation, thereby minimizing human input and stabilizing the PSU.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human experts manually control and monitor PSU performance, then PSU operation can be optimized, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The system enables self-service through automated ML-based control. The ML agent continuously monitors PSU properties, predicts optimal control actions, and executes adjustments without human intervention. The PSU effectively controls itself by receiving automated instructions to adjust parameters like switching frequency and input current, eliminating the need for manual expert monitoring while maintaining optimal performance.
Solution Approach 2:
The patent replaces manual mechanical control with an automated digital system. Human expert knowledge is encoded into trained ML models that process PSU measurements and generate control instructions automatically. This substitution transforms the manual control process into an automated information-processing system that continuously optimizes PSU operation without human time investment.
2Reliability
If comprehensive PSU control adjustments are implemented, then efficiency and lifetime can be improved, but the complexity of control increases
Solution Approach 1:
The trained ML model serves as an intermediary between raw PSU measurements and control actions. Instead of complex rule-based control logic, the ML agent processes measurements, predicts optimal actions, and generates instructions. This intermediary approach simplifies the control system by using a single trained model to handle multiple control parameters rather than implementing complex manual control logic for each parameter.
Solution Approach 2:
The system optimizes PSU performance by automatically adjusting key parameters such as switching frequency and input current based on ML predictions. Rather than controlling every possible parameter manually, the system focuses on changing the most impactful parameters that influence efficiency and lifetime, thereby reducing control complexity while maintaining effectiveness.
3Reliability
If all suggested actions are transmitted to the PSU, then comprehensive optimization is achieved, but data transmission volume increases and PSU stability may be compromised
Solution Approach 1:
The system applies partial action by selecting and transmitting only the most relevant suggested actions to the PSU. The ML agent filters and prioritizes actions based on their predicted impact on PSU properties, transmitting only the essential control parameters rather than all possible adjustments. This selective transmission reduces data volume while maintaining optimization effectiveness.
Solution Approach 2:
The patent extracts only the critical control information from the full set of suggested actions. The ML model identifies and extracts the most impactful actions that will significantly improve PSU efficiency and stability, separating these from less important adjustments. This extraction approach reduces transmission volume by focusing only on essential control data.
Data Source
AI summary
Methods and systems for power supply unit (PSU) control. A method includes measuring one or more properties of the PSU to obtain property measurements, and initiating transmission of the property measurements to a machine learning (ML) agent hosting a trained ML model. The method further includes receiving the property measurements at the ML agent, and processing the received property measurements using the trained ML model to generate suggested actions to be taken by the PSU. The method further includes predicting the effect of each of the suggested actions on the measured PSU properties, and selecting a subset of the suggested actions predicted to have a significant impact on the measured PSU properties. The method further includes initiating transmission of the selected subset of suggested actions to the PSU, and performing, at the PSU, the selected subset of suggested actions.