TY - JOUR

T1 - Predicting CNC Machine Processing Times in Process Chains

T2 - 18th IFAC Workshop on Time Delay Systems, TDS 2024

AU - Denkena, Berend

AU - Friebe, Sven

AU - Nein, Marcus

N1 - Publisher Copyright: © 2024 The Authors.

PY - 2024

Y1 - 2024

N2 - Accurate prediction of machining processing times is a critical factor in optimizing production planning, as deviations in predictions made by CAM software range between 8-74 % for 5-axis simultaneous machining. This prediction error in time can lead to significant deviations between planned and executed production times due to error propagation in process chains. Especially highly utilized machine tools could otherwise be overbooked or underutilized. While previous studies focused either on data-driven approaches like neural networks or analytical models based on machine kinematics, this paper introduces a novel grey box model that combines Artificial Intelligence (AI) and kinematical models. The method uses machine feedback data and the NC code to provide a more comprehensive and interpretable prediction of machining processing times than the current models. The analytical model analyzes the NC code to gain knowledge on planned milling operations and provide a baseline estimate, while the AI model uses machine learning algorithms to refine these estimates based on machine feedback data. The grey box model is validated by milling experiments on a 5-axis CNC machine, achieving an accuracy rate of over 98 % in predicting processing times. The method not only improves machining time estimation but also increases efficiency of production planning for CNC machines with high spindle uptime. This innovation directly contributes to speeding up manufacturing by enabling data-driven, adaptive production planning, thereby optimizing resource allocation and enhancing overall operational efficiency.

AB - Accurate prediction of machining processing times is a critical factor in optimizing production planning, as deviations in predictions made by CAM software range between 8-74 % for 5-axis simultaneous machining. This prediction error in time can lead to significant deviations between planned and executed production times due to error propagation in process chains. Especially highly utilized machine tools could otherwise be overbooked or underutilized. While previous studies focused either on data-driven approaches like neural networks or analytical models based on machine kinematics, this paper introduces a novel grey box model that combines Artificial Intelligence (AI) and kinematical models. The method uses machine feedback data and the NC code to provide a more comprehensive and interpretable prediction of machining processing times than the current models. The analytical model analyzes the NC code to gain knowledge on planned milling operations and provide a baseline estimate, while the AI model uses machine learning algorithms to refine these estimates based on machine feedback data. The grey box model is validated by milling experiments on a 5-axis CNC machine, achieving an accuracy rate of over 98 % in predicting processing times. The method not only improves machining time estimation but also increases efficiency of production planning for CNC machines with high spindle uptime. This innovation directly contributes to speeding up manufacturing by enabling data-driven, adaptive production planning, thereby optimizing resource allocation and enhancing overall operational efficiency.

KW - Machine learning

KW - Milling, Machine tool

KW - Predictive model

UR - http://www.scopus.com/inward/record.url?scp=85213041171&partnerID=8YFLogxK

U2 - 10.1016/j.procir.2024.10.087

DO - 10.1016/j.procir.2024.10.087

M3 - Conference article

AN - SCOPUS:85213041171

VL - 130

SP - 276

EP - 281

JO - Procedia CIRP

JF - Procedia CIRP

SN - 2212-8271

Y2 - 2 October 2023 through 5 October 2023

ER -