Server Acoustic and Vibration Detection for Nonintrusive Fault Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Detecting faults in server components installed in restricted locations without interrupting operation is challenging due to limited physical access and the need for continuous monitoring, making it difficult to identify and isolate the root cause of failures.
Innovation Solution
Implementing acoustic sensors and vibration sensors within a noise-controlled environment, combined with machine learning models, to analyze acoustic and vibration data for early detection of mechanical and non-mechanical failures, and pinpointing the location of defective components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If continuous monitoring or periodic inspections by technicians are implemented to detect server failures, then detection capability is improved, but server operation is interrupted and maintenance complexity increases
Solution Approach 1:
The server system performs self-diagnosis through acoustic monitoring. Acoustic sensors continuously capture sounds from server components, and machine learning models automatically analyze these sounds to detect anomalies and predict failures. This eliminates the need for manual technician inspections, allowing continuous operation without interruption while maintaining high detection capability.
Solution Approach 2:
Manual inspection by technicians is replaced with an automated acoustic monitoring system. Acoustic sensors capture mechanical sounds from server components, and machine learning algorithms process these acoustic signals to identify failures. This substitution of mechanical human inspection with automated acoustic analysis enables continuous monitoring without server interruption.
2Measurement precision
If physical inspection or testing of server components is performed to identify root cause of failures, then diagnostic accuracy is improved, but server operation is interrupted and access difficulty increases
Solution Approach 1:
Physical inspection and testing of server components is replaced with non-contact acoustic sensing. Acoustic sensors mounted on the server chassis capture sounds from internal components through the enclosure. Machine learning models analyze these acoustic signals to precisely identify the root cause of failures, achieving high diagnostic accuracy without requiring physical access to components or interruption of server operation.
Solution Approach 2:
The server enclosure acts as an intermediary that transmits acoustic signals from internal components to external sensors. Acoustic sensors mounted on the chassis receive sounds generated by internal components through the enclosure structure, enabling non-intrusive detection of component failures without physical access to the components themselves.
3Measurement precision
If acoustic sensors and vibration sensors are deployed to enable early detection of failures, then detection precision is improved, but device complexity and noise sensitivity increase
Solution Approach 1:
Acoustic sensing and vibration sensing capabilities are merged into a single integrated system. The server employs acoustic sensors that capture both acoustic signals and vibration signals from components. Machine learning models process these combined signals to detect anomalies, achieving high detection precision while reducing overall system complexity compared to separate acoustic and vibration sensing systems.
Solution Approach 2:
Acoustic sensors are designed to perform multiple functions: detecting acoustic emissions from components and simultaneously capturing vibration signals. This multi-functionality reduces the number of separate sensors needed, simplifying the device while maintaining high anomaly detection precision through comprehensive signal capture.
Data Source
AI summary
Examples described herein relate to circuitry to receive data associated with a device and indicate whether the device is potentially malfunctioning based on anomalous sounds in an operational server and based on an activity indicator of the server. In some examples, the device includes one or more of: a processor, a memory device, a thermal manager device, or a circuit board. In some examples, the data comprises a temperature signal and a sound signal and the circuitry is to: based on a first level of the temperature signal and a first characteristic of the sound signal, determine that the device of the server is potentially malfunctioning and based on a second level of the temperature signal and a second characteristic of the sound signal, determine that the device of the server is not potentially malfunctioning.