تخمین جداسازی لوبیای چشم بلبلی با میز جداکننده وزنی با استفاده از روش جنگل تصادفی بهینه سازی شده با الگوریتم ژنتیک

نوع مقاله : مقاله پژوهشی

نویسندگان

1 گروه مهندسی مکانیک بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران.

2 استاد گروه مهندسی مکانیک بیوسیستم، دانشکده کشاورزی دانشگاه تهران، پردیس ابوریحان، پاکدشت، ایران.

3 گروه مهندسی بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران

4 گروه مهندسی بیوسیستم ، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران

10.22034/jess.2022.330298.1725

چکیده

جداسازی یک عضو غیرقابل انکار در فرآیند پس از برداشت محصولات فله ای است. دستگاه جداکننده میز وزنی (GTSM) یکی از دستگاه هایی است که برای جداسازی ناخالصی های موجود در توده های دانه استفاده می شود. با توجه به تعداد پارامترهای قابل تغییر در GTSM و تاثیربالای این عوامل بر میزان SP ناخالصی ها در توده لوبیا چشم بلبلی و با توجه به اینکه بررسی تمام مقادیر در این مورد تقریبا غیرممکن به نظر می رسد. استفاده از یادگیری ماشین (ML) برای پیش‌بینی فرآیند SP در برابر تغییرات اعمال شده در این عوامل، استفاده از GTSM برای نخود سیاه چشم (BeP) را تسهیل می‌کند. مطالعه حاضر در مورد پیش بینی عملکرد یک GTSM در جداسازی BeP است. متغیرهای وابسته شامل دانه های تمیز شده (Y1)، وزن دانه های تمیز شده (Y2)، تعداد ناخالص کل (Y3)، وزن ناخالص کل (Y4)، تعداد دانه های پوسیده (Y5)، وزن دانه های پوسیده (Y6)، دانه های شکسته شده بودند. تعداد (Y7) و وزن دانه شکسته (Y8) و متغیرهای مستقل شامل شیب های عرضی جدول (X1)، شیب های طولی جدول (X2)، فرکانس نوسان میز (X3) و سرعت هوای دمنده (X4) می باشد. روش‌های مورد استفاده جنگل تصادفی منفرد (RF) و جنگل تصادفی ترکیبی با الگوریتم ژنتیک (RF-GA) برای بهینه‌سازی پارامترهای RF بودند. نتایج با استفاده از ضریب همبستگی (CC)، شاخص پراکنده (SI) و شاخص ویلموت (WI) مورد ارزیابی قرار گرفت. با توجه به یافته‌ها، روش ترکیبی عملکرد بالاتری نسبت به روش تک ارائه کرد و عملکرد پیش‌بینی را با موفقیت افزایش داد.

کلیدواژه‌ها


عنوان مقاله [English]

Estimation of separation of Black-eyed pea with gravity separator table using Random Forest optimized by Genetic algorithm

نویسندگان [English]

  • Saeed AgaAzizi 1
  • MohammadHosain Kianmehr 2
  • Amir Shayei 3
  • mohammadali Keyhandoust 4
1 Department of Biosystems Engineering, Faculty of Agriculture and Natural Resources, University of Mohaghegh Ardabili, Ardibil, Iran.
2 Professor, Department of Biosystems Engineering, University of Tehran, College of Abouraihan, Pakdasht, Iran
3 Department of Biosystems Engineering, Faculty of Agricultural Technology and Natural Resources, University of Mohaghegh Ardabili, Ardibil, Iran
4 Department of Biosystems Engineering, Faculty of Agricultural Technology and Natural Resources, University of Mohaghegh Ardabili, Ardibil, Iran
چکیده [English]

Separation (SP) is an undeniable member in the process set after harvesting of bulk products. The gravity tableseparator machine (GTSM) is one of the devices used to separate modal impurities in grain masses. Due to the continuity in the range of changes in the parameters of the GTSM and the high number of these factors affecting the rate of SP of impurities in the mass of cowpea beans, and considering that it seems almost impossible to examine all the values in this range. The use of machine learning (ML) to predict the process of SP against the changes applied to these factors facilitates the use of the GTSM for black-eyed pea (BeP). The present study is about predicting the performance of a GTSM in separating the BeP. The dependent variables included the cleaned seeds (Y1), weight of the cleaned seeds (Y2), total gross number (Y3), total gross weight (Y4), rotten seeds number (Y5), rotten seeds weight (Y6), broken seeds number (Y7) and broken seeds weight (Y8) and the independent variables included transverse slopes of the table (X1), longitudinal slopes of the table (X2), frequencies of table oscillation (X3) and blower air speeds (X4). The employed methods were single Random forest (RF) and a hybrid Random forest integrated by Genetic algorithm (RF-GA) for optimization of RF parameters. Results were evaluated using correlation coefficient (CC), Scattered Index (SI) and Willmott’s Index (WI). According to findings, hybrid method provided higher performance compared with that for the single method and increased the prediction performance, successfully.
Separation (SP) is an undeniable member in the process set after harvesting of bulk products. The gravity tableseparator machine (GTSM) is one of the devices used to separate modal impurities in grain masses. Due to the continuity in the range of changes in the parameters of the GTSM and the high number of these factors affecting the rate of SP of impurities in the mass of cowpea beans, and considering that it seems almost impossible to examine all the values in this range. The use of machine learning (ML) to predict the process of SP against the changes applied to these factors facilitates the use of the GTSM for black-eyed pea (BeP). The present study is about predicting the performance of a GTSM in separating the BeP. The dependent variables included the cleaned seeds (Y1), weight of the cleaned seeds (Y2), total gross number (Y3), total gross weight (Y4), rotten seeds number (Y5), rotten seeds weight (Y6), broken seeds number (Y7) and broken seeds weight (Y8) and the independent variables included transverse slopes of the table (X1), longitudinal slopes of the table (X2), frequencies of table oscillation (X3) and blower air speeds (X4). The employed methods were single Random forest (RF) and a hybrid Random forest integrated by Genetic algorithm (RF-GA) for optimization of RF parameters. Results were evaluated using correlation coefficient (CC), Scattered Index (SI) and Willmott’s Index (WI). According to findings, hybrid method provided higher performance compared with that for the single method and increased the prediction performance, successfully.
Separation (SP) is an undeniable member in the process set after harvesting of bulk products. The gravity tableseparator machine (GTSM) is one of the devices used to separate modal impurities in grain masses. Due to the continuity in the range of changes in the parameters of the GTSM and the high number of these factors affecting the rate of SP of impurities in the mass of cowpea beans, and considering that it seems almost impossible to examine all the values in this range. The use of machine learning (ML) to predict the process of SP against the changes applied to these factors facilitates the use of the GTSM for black-eyed pea (BeP). The present study is about predicting the performance of a GTSM in separating the BeP. The dependent variables included the cleaned seeds (Y1), weight of the cleaned seeds (Y2), total gross number (Y3), total gross weight (Y4), rotten seeds number (Y5), rotten seeds weight (Y6), broken seeds number (Y7) and broken seeds weight (Y8) and the independent variables included transverse slopes of the table (X1), longitudinal slopes of the table (X2), frequencies of table oscillation (X3) and blower air speeds (X4). The employed methods were single Random forest (RF) and a hybrid Random forest integrated by Genetic algorithm (RF-GA) for optimization of RF parameters. Results were evaluated using correlation coefficient (CC), Scattered Index (SI) and Willmott’s Index (WI). According to findings, hybrid method provided higher performance compared with that for the single method and increased the prediction performance, successfully.
Separation (SP) is an undeniable member in the process set after harvesting of bulk products. The gravity tableseparator machine (GTSM) is one of the devices used to separate modal impurities in grain masses. Due to the continuity in the range of changes in the parameters of the GTSM and the high number of these factors affecting the rate of SP of impurities in the mass of cowpea beans, and considering that it seems almost impossible to examine all the values in this range. The use of machine learning (ML) to predict the process of SP against the changes applied to these factors facilitates the use of the GTSM for black-eyed pea (BeP). The present study is about predicting the performance of a GTSM in separating the BeP. The dependent variables included the cleaned seeds (Y1), weight of the cleaned seeds (Y2), total gross number (Y3), total gross weight (Y4), rotten seeds number (Y5), rotten seeds weight (Y6), broken seeds number (Y7) and broken seeds weight (Y8) and the independent variables included transverse slopes of the table (X1), longitudinal slopes of the table (X2), frequencies of table oscillation (X3) and blower air speeds (X4). The employed methods were single Random forest (RF) and a hybrid Random forest integrated by Genetic algorithm (RF-GA) for optimization of RF parameters. Results were evaluated using correlation coefficient (CC), Scattered Index (SI) and Willmott’s Index (WI). According to findings, hybrid method provided higher performance compared with that for the single method and increased the prediction performance, successfully.

کلیدواژه‌ها [English]

  • Black-eyed pea
  • machine learning
  • separation
  • gravity table