Abstract
Owing to large production capacity and high efficiency, multibridge machining systems (MBMSs) have gained increasing attention in industry. Their multiple bridge machines work concurrently in their serially arranged and partially overlapping workspaces. To solve totally the job scheduling and collision resolution problems of MBMS, our prior work proposes a serial-colored traveling salesman problem (S-CTSP)-based method. Each salesman in S-CTSP visits his exclusive cities and some cities shared with his neighbor(s). To endow MBMS with fault-tolerance ability, this paper presents a two-stage method for both static path planning and dynamic collision avoidance of multiple machines. At the first stage, the path planning problem abstracted as an S-CTSP is solved by a population-based incremental learning (PBIL) algorithm. The PBIL introduces a local search operation, two types of possibility vectors, and a selection strategy of exclusive and shared cities. The second stage uses a Petri net supervisor to dynamically avoid any emerging collision when performing the scheduled work. This work also provides a novel priority net structure to prevent mutual waiting of two neighboring machines ready to enter their overlapping workspace. Then, we apply the presented method to a large tribridge waterjet cutting case to show its performance.