Applying artificial intelligence in ERP systems to improve production planning and work scheduling in the machining industry
Aho, Aaro (2026)
2026
Tuotantotalouden DI-ohjelma - Master's Programme in Industrial Engineering and Management
Johtamisen ja talouden tiedekunta - Faculty of Management and Business
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2026-05-22
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202605216088
https://urn.fi/URN:NBN:fi:tuni-202605216088
Tiivistelmä
This study examines how artificial intelligence (AI)-supported solutions can be implemented within an enterprise resource planning (ERP) environment to improve production planning and control in a manufacturing small and medium-sized enterprise (SME). The research was conducted as a qualitative-dominant embedded case study at the case company, a make-to-order machine shop specialising in the maintenance and refurbishment of paper machine rolls. The empirical data consisted of semi-structured interviews, direct observations, and quantitative analysis of historical production data, enabling a triangulated understanding of the research problem.
The findings show that the implementation of AI-supported production control in an SME context consists of interdependent process stages, where ERP modernisation, data quality, and organisational readiness form critical prerequisites. AI does not create value as a standalone technology but as an extension of the ERP data infrastructure, utilising structured and time-stamped operational data to support human decision-making.
The study identifies key limitations in the current production control system, including fragmented data management, delayed reporting, lack of real-time visibility, and strong dependence on tacit knowledge. These challenges constrain both operational efficiency and the systematic utilisation of historical data. The results further demonstrate that ERP selection and implementation choices directly influence the feasibility of AI utilisation by shaping data structures, integration capabilities, and overall system usability.
A practical AI pilot was developed to evaluate feasible use cases in production planning. The pilot applied a similarity-based retrieval approach to estimate work durations based on historical production data. The results indicate that such approaches can provide transparent and actionable decision support without requiring complex machine learning models, making them particularly suitable for SME environments with limited data maturity.
The study contributes to the understanding of AI adoption in manufacturing SMEs by demonstrating that data quality, system integration, and socio-technical factors are more critical to success than algorithmic sophistication. The findings emphasise that ERP modernisation should be treated as a foundational step toward AI-enabled production control, and that incremental, data-driven development offers a practical pathway for integrating AI into daily operations.
The findings show that the implementation of AI-supported production control in an SME context consists of interdependent process stages, where ERP modernisation, data quality, and organisational readiness form critical prerequisites. AI does not create value as a standalone technology but as an extension of the ERP data infrastructure, utilising structured and time-stamped operational data to support human decision-making.
The study identifies key limitations in the current production control system, including fragmented data management, delayed reporting, lack of real-time visibility, and strong dependence on tacit knowledge. These challenges constrain both operational efficiency and the systematic utilisation of historical data. The results further demonstrate that ERP selection and implementation choices directly influence the feasibility of AI utilisation by shaping data structures, integration capabilities, and overall system usability.
A practical AI pilot was developed to evaluate feasible use cases in production planning. The pilot applied a similarity-based retrieval approach to estimate work durations based on historical production data. The results indicate that such approaches can provide transparent and actionable decision support without requiring complex machine learning models, making them particularly suitable for SME environments with limited data maturity.
The study contributes to the understanding of AI adoption in manufacturing SMEs by demonstrating that data quality, system integration, and socio-technical factors are more critical to success than algorithmic sophistication. The findings emphasise that ERP modernisation should be treated as a foundational step toward AI-enabled production control, and that incremental, data-driven development offers a practical pathway for integrating AI into daily operations.