There are numerous riveting points on the large-sized aircraft panel, irregular row of riveting points on delta wing. It is essential to plan the riveting sequence reasonably to improve the efficiency and accuracy of automatic drilling and riveting. Therefore, this article presents a new multi-objective optimization method based on ant colony optimization (ACO). Multi-objective optimization model of automatic drilling and riveting sequence planning is built by expressing the efficiency and accuracy of riveting as functions of the points' coordinates. In order to search the sequences efficiently and improve the quality of the sequences, a new local pheromone updating rule is applied when the ants search sequences. Pareto dominance is incorporated into the proposed ACO to find out the non-dominated sequences. This method is tested on a hyperbolicity panel model of ARJ21 and the result shows its feasibility and superiority compared with particle swarm optimization (PSO) and genetic algorithm (GA).

Purpose: The purpose of this paper is to present an analytical method for throughput analysis of assembly systems with complex structures during transients. Design/methodology/approach: Among the existing studies on the performance evaluation of assembly systems, most focus on the system performance in steady state. Inspired by the transient analysis of serial production lines, the state transition matrix is derived considering the characteristics of merging structure in assembly systems. The system behavior during transients is described by an ergodic Markov chain, with the states being the occupancy of all buffers. The dynamic model for the throughput analysis is solved using the fixed-point theory. Findings: This method can be used to predict and evaluate the throughput performance of assembly systems in both transient and steady state. By comparing the model calculation results with the simulation results, this method is proved to be accurate. Originality/value: This proposed modeling method can depict the throughput performance of assembly systems in both transient and steady state, whereas most exiting methods can be used for only steady-state analysis. In addition, this method shows the potential for the analysis of complex structured assembly systems owing to the low computational complexity.

Fiber-reinforced composites, such as glass fiber-reinforced polymer (GFRP), are widely used across industries but are susceptible to inclusion defects during manufacturing. Detecting and classifying these defects is crucial for ensuring material integrity. This study classifies four common inclusion defects—metal, peel ply, release paper, and PTFE film—in GFRP using terahertz technology and machine learning. Two GFRP sheets with inclusion defects at different depths were fabricated. Terahertz time-domain signals were acquired, and a cross-correlation-based deconvolution algorithm extracted impulse responses. LSTM-RNN, Bi-LSTM RNN, and 1D-CNN models were trained and tested on time-domain, frequency-domain, and impulse response signals. The defect-free region exhibited the highest classification accuracy. Bi-LSTM RNN achieved the best recall and macro F1-score, followed by 1D-CNN, while LSTM-RNN performed worse. Training with impulse response signals improved classification while maintaining accuracy.