AI-Driven Process Optimization for Electrical Product Manufacturing
Tilander, Roope (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025112830743
https://urn.fi/URN:NBN:fi:amk-2025112830743
Tiivistelmä
This thesis examines AI-driven process optimization in electrical product manufacturing with a focus on predictive maintenance. The research shows how artificial intelligence technologies can optimize modern manufacturing processes through increased data availability. The research utilizes a mixed-methods approach, integrating a systematic literature review of 32 sources with a qualitative case study conducted at an electrical manufacturing facility. Through semi-structured interviews and analysis, the research investigates current implementations, identifies challenges, and assesses potential AI implementations.
Key findings indicate that predictive maintenance prolongs equipment lifespan by as much as 25% and diminishes unplanned downtime by 30%, whereas automated quality control can facilitate up to 100% inspection rates. Though these findings are subjective and will differ from other implementations. The case study indicates that effective implementation relies on organizational readiness, specifically data maturity and cultural readiness. Primary obstacles for implementation include return on investment timelines, extensive product diversity, inadequate real-time data infrastructure, and challenges in quantifying advantages. Cost-effective solutions were proposed, including simulation-based material flow optimization and computer vision systems for quality assurance, both promising a relatively short return on investment window. The study suggests that effective AI implementation in manufacturing requires a thorough evaluation of organizational readiness, incremental deployment, clear communication with the workforce, and strong data management standards.
Key findings indicate that predictive maintenance prolongs equipment lifespan by as much as 25% and diminishes unplanned downtime by 30%, whereas automated quality control can facilitate up to 100% inspection rates. Though these findings are subjective and will differ from other implementations. The case study indicates that effective implementation relies on organizational readiness, specifically data maturity and cultural readiness. Primary obstacles for implementation include return on investment timelines, extensive product diversity, inadequate real-time data infrastructure, and challenges in quantifying advantages. Cost-effective solutions were proposed, including simulation-based material flow optimization and computer vision systems for quality assurance, both promising a relatively short return on investment window. The study suggests that effective AI implementation in manufacturing requires a thorough evaluation of organizational readiness, incremental deployment, clear communication with the workforce, and strong data management standards.