In semiconductor fabrications, batch processing devices (BPD) are mainly used for oxidation, diffusion, deposition and performance test operations. To optimize scheduling solutions of non-identical semiconductor BPDs where existing multi-products and different due dates, a heuristic scheduling method based on Particle swarm optimization (PSO) and weighted cost rate (WCR) is proposed for minimizing the objective of the earliness/tardiness(E/T) penalties. The two-level heuristic scheduling algorithm is developed. Experiment results indicate that the proposed algorithm is valid and practical. Compared with the previous heuristic scheduling algorithms, it is more efficient in improving the weighted earliness/tardiness performance.