HPC Predictive Maintenance Using Cleansed Performance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional predictive maintenance systems for High Performance Computing (HPC) systems often take a reactive approach, leading to increased downtime and reduced productivity due to inaccurate data interpretation caused by noise in collected data, which affects the accuracy of fault identification and maintenance requirements.
Innovation Solution
A processor-implemented method and system that performs data abstraction and cleansing by sampling, removing outliers, applying time-series up-sampling, calculating probability distributions, measuring divergence, and adding additional data to reduce noise, followed by data padding and smoothing, to generate cleansed performance data for machine learning-based predictive maintenance predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a proactive approach is used to monitor and automatically identify maintenance scenarios, then maintenance accuracy is improved, but data quality issues cause reduced prediction accuracy
Solution Approach 1:
The patent introduces an intermediary data cleansing module that processes raw performance data before it reaches the predictive maintenance system. This intermediary layer removes noise, handles missing values, and standardizes data formats, thereby improving the reliability of predictions without compromising the proactive monitoring capability. The intermediary acts as a buffer between data collection and analysis, ensuring that only high-quality data is used for fault identification.
Solution Approach 2:
The patent replaces manual data quality assessment and cleaning processes with automated machine learning-based data cleansing algorithms. These algorithms automatically detect and remove noise, identify outliers, and impute missing values without human intervention. This substitution of mechanical/manual processes with intelligent automated systems improves both the speed and accuracy of data preparation, enhancing prediction reliability while maintaining proactive monitoring.
2Device complexity
If reactive maintenance is implemented by triggering maintenance only when component failure alerts are received, then system complexity is reduced, but equipment downtime increases
Solution Approach 1:
The patent implements preliminary action by using predictive analytics to identify potential component failures before they actually occur. The system continuously monitors performance data, detects early signs of degradation, and triggers maintenance alerts in advance of actual failures. This allows maintenance to be scheduled proactively, reducing equipment downtime while maintaining manageable system complexity through automated prediction algorithms.
Solution Approach 2:
The patent incorporates feedback loops where the outcomes of maintenance actions are fed back into the predictive model. When maintenance is performed and component status is updated, this information is used to refine and improve future predictions. The feedback mechanism enables the system to learn from actual maintenance outcomes, improving prediction accuracy over time without significantly increasing system complexity.
Data Source
AI summary
State of the art predictive maintenance systems that generate predictions with respect to maintenance of High Performance Computing (HPC) systems have the disadvantage that they either are reactive, or the predictions are affected due to quality issues associated with the data being collected from the HPC systems. The disclosure herein generally relates to predictive maintenance, and, more particularly, to a method and system for predictive maintenance of High Performance Computing (HPC) systems. The system performs abstraction and cleansing on performance data collected from the HPC systems, and generates a cleansed performance data, on which a Machine Leaning (ML) prediction is applied to generate predictions with respect to maintenance of the HPC systems.