Remote Diagnostics, Machine Learning and Data Collection for Automated High Throughput Mass Spectrometry

Library Number:
PSTR134950987
Author(s):
Emmy M Hoyes, Thomas C. Smallwood, Nicola Johnston, Richard C Chapman, Allen Caswell
Source:
Waters
Content Type:
Posters
Content Subtype:
BMSS
Related Products:
 
 

With the arrival of low cost, high performance mass spectrometers in the hands of non experts, there is now a pressing need for remote monitoring and automated fault detection in order to minimise user intervention and maximise instrument up time.

In recent times, the advent of ubiquitous computing, the internet of things and the machine learning revolution, has meant, it is possible to collect, process and analyse data from many mass spectrometers simultaneously. Using modern machine learning algorithms on this data it is possible to automatically discover and adapt the control system in order to diagnose faults remotely and automatically. This system will ultimately be extended to include predictive monitoring to as much as possible avoid any unscheduled downtime.

Here we present prototype software for the collection, processing and detection of failure modes and degradation using modern machine learning techniques.


Title Format File Size
Download PDF PDF 959.06kB