Business and Industrial Section meeting on Predictive maintenance: Algorithms and engineers

Predictive Maintenance is a hot topic in industry as transport, utilities and other asset-intensive organisations look to take advantage of ‘intelligent infrastructure’ and optimise allocation of their cost-constrained resources. The underlying challenges are statistical, the computational tools easily accessible, but essential insight is still trapped in the in minds of the ageing workforce. This session, organised by the Business and Industrial Section (BIS) set out to see how we can best combine the power of the algorithms and the knowledge of the engineers.

Chairing the session, Neil Spencer and David Smallbone of BIS introduced the audience to the concepts of predictive maintenance. David contrasted the promise with the practicalities of introducing statistics and data science approaches in large, complex organisations where engineering judgement is still king.

They were joined by Matthew King (associate director, KPMG Ireland), Andrew Parnell (Hamilton Professor, Maynooth University and CSO, Prolego Scientific), Adele Marshall (professor of statistics at Queen’s University, Belfast) who shared their own experiences of developing and deploying statistics and machine learning approaches in complex engineering and manufacturing organisations.

Discussion topics raised by the audience included:

  • How can statisticians and analysts build better relationships with their engineering colleagues?
  • Do complex machine learning algorithms actually provide any additional robustness over engineering judgement when they are equally as hard to audit?
  • How do you make sure algorithms are making ‘ethical’ decisions that correctly balance cost, risk and performance?
  • Are organisations getting distracted by the hype around ‘on trend’ ML models when simpler, more auditable statistical models would perform as well or better?

Overall, the discussion recognised the increasing role that statisticians and data scientists can play in working more closely with their engineering colleagues to help them make best use of the available data and drive value from their assets.


Load more