Hybrid LSTM-GA and HS-CS Algorithms for Automation-Assisted Defect Identification in CNC Milling Robotics: Enhancing Global Search and Optimization

This paper introduces a hybrid algorithmic framework that merges Long Short-Term Memory (LSTM) networks, Genetic Algorithms (GA), Harmony Search (HS), and Cuckoo Search (CS) for detecting flaws in Computer Numerical Control (CNC) milling automatically. LSTM networks are used for processing time-series data, while GA optimizes hyperparameters to enhance accuracy. Efficient feature extraction and overall optimization search benefit from the use of HS and CS. The hybrid approach addresses challenges in real-time defect detection by surpassing individual models in metrics like accuracy, precision, recall, and mean squared error (MSE). In CNC milling, this minimizes interruptions and enhances production excellence. The study aims to optimize real-time flaw detection in CNC milling by utilizing a combination of LSTM-GA and HS-CS algorithms. The project's goal is to achieve accurate and efficient defect classification by utilizing HS-CS for global search and feature selection, with minimal human involvement and production delay. The goal is to demonstrate how hybrid algorithms improve defect detection efficiency and reduce error rates, leading to precise and automated CNC milling solutions. HS and CS are used to select features and perform a global search, while LSTM is integrated into the model to analyze time-series data, and GA is employed to optimize hyperparameters. Accuracy, precision, recall, F1 score, and MSE are metrics utilized for evaluating performance. The hybrid LSTM-GA and HS-CS model outperformed individual approaches with 97.5% accuracy, 96.3% precision, 94.5% recall, and an F1 score of 0.954, which is equal to its MSE. This combination method is suitable for real-time industrial automation tasks as it enhances fault detection in CNC milling, reducing errors and enhancing precision.