دانشگاه گیلان
معاونت پژوهش و فناوری دانشگاه گیلان
انتشارات دانشگاه گیلان

 آمار

تعداد نشریات33
تعداد شماره‌ها886
تعداد مقالات8,690
تعداد مشاهده مقاله55,110,949
تعداد دریافت فایل اصل مقاله10,383,583
Ghafouri, Alireza, Arghand, Hesamaddin. (1405). Feedforward Neural Network for Rotor Unbalance Identification: Influence of Number of Measurement Runs. فیزیولوژی و بیوتکنولوژی آبزیان, (), -. doi: 10.22124/cse.2026.33575.1185
Alireza Ghafouri; Hesamaddin Arghand. "Feedforward Neural Network for Rotor Unbalance Identification: Influence of Number of Measurement Runs". فیزیولوژی و بیوتکنولوژی آبزیان, , , 1405, -. doi: 10.22124/cse.2026.33575.1185
Ghafouri, Alireza, Arghand, Hesamaddin. (1405). 'Feedforward Neural Network for Rotor Unbalance Identification: Influence of Number of Measurement Runs', فیزیولوژی و بیوتکنولوژی آبزیان, (), pp. -. doi: 10.22124/cse.2026.33575.1185
Ghafouri, Alireza, Arghand, Hesamaddin. Feedforward Neural Network for Rotor Unbalance Identification: Influence of Number of Measurement Runs. فیزیولوژی و بیوتکنولوژی آبزیان, 1405; (): -. doi: 10.22124/cse.2026.33575.1185

Feedforward Neural Network for Rotor Unbalance Identification: Influence of Number of Measurement Runs

Computational Sciences and Engineering
مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 07 آبان 1405
نوع مقاله: Original Article
شناسه دیجیتال (DOI): 10.22124/cse.2026.33575.1185
نویسندگان
Alireza Ghafouri1؛ Hesamaddin Arghand* 2
1M.Sc. in Mechanical Engineering, Dept. of Mech. Eng., Faculty of Eng., University of Zanjan, Zanjan 45171-38791, Iran
2Assistant Professor, Dept. of Mech. Eng., Faculty of Eng., University of Zanjan, Zanjan 45171-38791, Iran
چکیده
Numerous methods exist for balancing rotating machinery, each possessing various advantages and limitations. This study presents a rotor balancing method utilizing machine learning (ML). To this end, the governing equations of a rigid rotor with 4 degrees of freedom (4-DOFs) under unbalance excitation are derived, and a simulator is developed to study its dynamic behavior. Subsequently, by utilizing this simulator across 2,000 simulation runs, the vibration response at four points resulting from applying random initial unbalances, followed by the addition of trial weights to the same rotor, was extracted to construct a dataset. A feedforward neural network (FFNN) is then employed to investigate the feasibility of identifying the initial unbalance based on vibration data. The FFNN performance was evaluated across various input cases (including 4-runs, 3-runs, and 2-runs). The results demonstrate that using 4-run and 3-run data, the model can accurately estimate both unbalance magnitude and position. However, with 2-run data, the model can only estimate the unbalance magnitude and fails to predict its position.
کلیدواژه‌ها
Rigid rotor؛ Vibration؛ Balancing؛ Machine learning؛ Feed forward neural network
آمار
تعداد مشاهده مقاله: 2


سامانه مدیریت نشریات علمی. قدرت گرفته از سیناوب