System und Verfahren für Fundamentmodellfunktionalitäten für Zeitreihendaten von Betriebsprozessen mit Industriekontextinformationen für Industrielle Anwendungen

Details

Veröffentlichungs-Nr.

EP4783069

Anmeldetag

24. Januar 2025

Veröffentlichung

29. Juli 2026

Rechtsraum

EP

Abstract

It is disclosed a computer-implemented method for obtaining, for industrial applications, foundation model functionalities for time series data of operational processes with industrial context information, the method comprising: transforming the time series data of operational processes associated with at least a first industrial plant into a sequence of time units associated with respective pieces of the operational processes; attaching time units of the sequence of time units with the industrial context information, wherein pieces of the industrial context information are associated with the pieces of the operational processes and wherein the time units are attached with concurrent pieces of the industrial context information; obtaining by means of a tokenization method a sequence of context tokens from the attaching, wherein one context token represents one time unit to which concurrent pieces of the industrial context information are attached; training a generative AI model on the sequence of context tokens using a loss function; and using the trained generative AI model having the foundation model functionalities for at least one of: i) finetuning the trained generative AI model for a second industrial plant different from the first industrial plant and ii) forecasting a further progression of an operational process for the second industrial plant.

Anmelder

Firma

ABB SCHWEIZ AG

Land

🇨🇭 Schweiz