TECH & AI

compressor that has run for decades gives few warnings before it fails. By the time a vibration or a temperature spike is visible, the damage is often already done and the production line pays the price. That gap, between a problem forming and a problem showing, is where a new kind of maintenance is being put to use. Anindya Chatterjee, Head of Digital Industry Value Engineering and Data Science for Industrial AI at ABB, has spent recent years closing it.

Catching failure before it happens Anindya says machine learning is changing the maintenance sequence entirely, continuously analysing operational data to surface subtle patterns long before a human would notice anything wrong.

In customer deployments, ABB has used multivariate anomaly detection models to identify potential failures, Anindya says,“ up to six to 10 days before they occur”. While nine a few days may not sound dramatic, this can be the difference between an orderly repair and an emergency shutdown.

The same idea can also be applied to equipment nearing the end of its life. Instead of servicing assets on a fixed schedule, ABB’ s models predict remaining useful life based on actual condition, allowing for work to be planned around real risk and saving effort where there may not be a problem.

Anindya says that most industrial maintenance budgets are still consumed by corrective repairs carried out after something has already failed. What machine learning offers, he argues, is a route out of that cycle, towards greater reliability and far less disruption to production.

Data trust and data volume Ask most manufacturers whether they have enough data, and the answer is invariably yes. Sensors, control systems and connected equipment already generate vast quantities of it. The problem, Anindya says, is having data you can trust.

Maintenance histories, failure records and reliability data are often much less structured than operational readings and it is here, he says, that“ many organisations face the greatest

TITLE: HEAD OF DIGITAL INDUSTRY VALUE ENGINEERING AND DATA SCIENCE FOR INDUSTRIAL AI

COMPANY: ABB

Anindya has more than 30 years of experience across refining, petrochemicals, operations, maintenance, reliability and digital transformation. Today, he leads teams responsible for value engineering, industrial AI and solution delivery.

74 August 2026