An optimized machine learning model for surface defect detection of heat sink based on hiking optimization algorithm
جاري التحميل...
التاريخ
عنوان الدورية
ردمد الدورية
عنوان المجلد
الناشر
Signal, Image and Video Processing
خلاصة
Identification of surface defects is essential in industrial production since it directly affects the quality of the final product and the efficiency of manufacturing. In this paper, a novel combination between Markov Gibbs Random Field algorithm and Hiking Optimization Algorithm for Random Forest are proposed for detection of gold-plated tungsten-copper alloy heat sink surface defects efficiently. The dataset contains 1000 images (320 × 320 pixels) of gold-plated tungsten-copper alloy heat sink surface defects and their annotation. Firstly, the preprocessing process involved data acquisition, followed by the splitting of data into training and testing sets. Subsequently, feature extraction was applied using the MGRF algorithm to estimate the energy value of each pixel. To improve accuracy, two hyperparameters of the Random Forest model, the number of trees and the depth of each tree must be optimized. The HOA optimizer is an efficient choice for this purpose. As a global optimization method, HOA leverages the search space of the optimization problem, and the prior knowledge of mountain and trail navigation acquired by hikers. After applying the combination of MGRF and HOA optimizer, the accuracy reached 98.73%, thereby validating the efficiency of the proposed model in classifying surface defects.