Abstract: To tackle the real-time scheduling problem of multi-load carriers in automobile assembly lines, a real-time scheduling approach was proposed based on Support Vector Machine (SVM). First, the real-time scheduling problem of multi-load carriers is formally described, and an objective function of assembly line output and the delivery distance is set up. Then, samples are generated by the simulation of the assembly line with material handling system to train an SVM model offline. Finally, the trained SVM model is used in the real-time scheduling process to make “wait” or “deliver” decisions. Experimental results indicate that the proposed approach outperforms the Minimum Batch Size (MBS) approach. It runs fast, performs well and to some degree adapts to the changes in the dynamic environment. It can be effectively used to improve the real-time scheduling of multi-load carries.

Key words: artificial intelligence, multiple-load carriers, support vector machine(SVM), real-time scheduling, material handling