Daikin and Hitachi trial AI for fault diagnosis

JAPAN: Daikin and Hitachi are to begin a collaborative trial to commercialise a generative AI agent that supports equipment fault diagnosis in factories.

The project, based at Daikin’s Rinkai commercial air conditioner factory in Sakai, Osaka, aims to combine the operational technology knowledge that Daikin has accumulated over many years with Hitachi’s advanced IT to improve productivity.

In 2017, Daikin began a collaboration with Hitachi to establish a next-generation production model utilising Hitachi’s Lumada IoT platform to enable digitalisation, comparison and analysis of the skills of workers on the brazing line at its residential air conditioning factory in Shiga. 

The intention was to use this data analytics and machine learning to support predictive maintenance and the optimisation of manufacturing processes.

In the new project, when a maintenance engineer using a tablet device finds a fault in a pump or valve while inspecting production equipment, the equipment fault diagnosis AI agent presents the cause and countermeasures to the maintenance engineer. 

To achieve that, Daikin first converts the factory equipment drawings into a knowledge graph that can be read by the generative AI. Then, the generative AI is trained on the knowledge graph, maintenance records, and other operational technology data to achieve fault diagnosis at the same level as or better than that of a typical Daikin maintenance engineer. 

In a previous proof-of-concept experiment, it was confirmed that the equipment fault diagnosis AI agent can answer the cause and countermeasures of equipment faults within 10 seconds and with an accuracy of over 90%.

Daikin and Hitachi plan to complete the trial operation by September and put the equipment fault diagnosis AI agent into practical use in Daikin factories worldwide.