Microsoft Supports Unilever in Digital Twin

The machines and equipment in the factory are connected so that they can send a mass of data — everything from temperatures to production cycle times — into the model.

The solution offers visibility across all levels of the plant. The collected data is mined for insights and patterns using advanced analytics and machine learning algorithms, which can predict outcomes based on historical data.

The algorithm can reach a level of accuracy where it can be allowed to directly control part of a machine or process. This allows operators to make better-informed decisions and frees them up from repetitive manual tasks for more value-added functions.

The company is streaming data from 15 of its 300 global plants, with plans to connect 70 factories by the end of the year and another 100 or so in 2020.

Unilever has been able to reduce the number of alerts requiring action by 90% per day, ensuring far fewer interruptions and more timely interventions.