Artificial Intelligence in Support of Industrial Processes

Euroopan unionin osarahoittama 2023
Keski-Pohjanmaan liitto
Kokkolan kaupunki

The project promotes the adoption of artificial intelligence in industrial companies in Kokkola by producing practical, hands-on knowledge through pilot implementations carried out in collaboration with companies. In these pilots, process data measured from industrial production equipment is analyzed using AI, with a particular focus on two key application areas: predictive maintenance and quality monitoring.

The project is co-financed by the European Union through the Kokkola Ecosystem Agreement. The project has also received funding from the City of Kokkola.

Motivation

Many industrial companies have been storing measurement data collected from their production equipment for years, often without a clear plan for how the data could be utilized. In the development of various AI applications, such long-term datasets represent valuable training material. Stored data can be used to train AI models, for example, to predict equipment maintenance needs and product quality.

However, building, training, and deploying AI models require specialized expertise that is not available in all industrial companies. Furthermore, industrial actors may not be fully aware of the opportunities AI offers for process data analysis. The Artificial Intelligence in Support of Industrial Processes project seeks to address these challenges by collecting and compiling practical knowledge on developing AI models suitable for process data.

Activities

The project develops pilot implementations of AI applications for process data and gathers experiences from their development and use. In the pilots conducted in collaboration with companies, particular attention is given to two themes: anomaly detection and soft sensors.

An important part of the project is also to compile the experience gained from these pilots into practical guidance that companies can continue to use after the project has ended.

Anomaly Detection

Anomaly detection can be used to enable AI applications to identify maintenance needs in production equipment. The idea is to train an AI application to recognize the characteristics of data that correspond to normal equipment operation and to trigger an alert whenever it detects data that deviates from normal patterns.

This makes it possible to identify early signs of potential equipment failures, long before the equipment breaks down and becomes unusable.

Soft Sensors

Soft sensors refer to virtual sensors that combine data from multiple sources to generate new information. In an industrial context, soft sensors can be used, for example, for quality monitoring by predicting the properties of the final product from production sensor data.

The outputs produced by such soft sensors can be utilized for process guidance, control, and optimization.

Dissemination of Results

During the project, practical documents and guidelines will be prepared to support the development of industrial AI applications. These recommendations, based on experiences gained from the pilot implementations, can later be used by companies to support decision-making, for example when considering whether to develop similar AI applications themselves or purchase them as consulting services.

Project manager

Leevi Lind, postdoctoral researcher

Information technology, teaching, research and projects

Email

Executive Director of the project

Ismo Hakala, professor

Head of Information Technology Unit

Email
Phone number
Room
418

Jari Luomala, project researcher

Information technology, research and projects

Email
Phone number
Room
432B
Euroopan unionin osarahoittama 2023
Keski-Pohjanmaan liitto
Kokkolan kaupunki

Active projects