Parallel Nested Optimization Algorithm On Pick-up Scheduling Of Steel Plate In A Shipyard

In shipbuilding firms, making schedule for steel plate pick-up operation is the basis of the steel plate stacking yard pick-up operation, and the beginning of the production. The schedule should be up to some operation goals-how to minimize pick-up operation time, for example. Moreover, because pick-up operation takes place every day, the plans need to be worked out in the prescribed time, which can be described as timeliness requirements. In fact, the problems have very large scale; regular serial program cannot fulfill timeliness requirements because of its low execution efficiency. With the population of multiprocessor computers and the parallel computing theory and application getting more and more deepening, parallel optimization algorithm can be used to improve execution efficiency of multi-level nested genetic algorithm. So timeliness requirements can be achieved.After a deep analysis on mathematical model of the pick-up operation schedule, and testing the existing multi-level nested genetic algorithm which has been used to realize the mathematical model, a parallel optimization model which applies parallel computing method is constructed. Master-slave parallel model is adopted for parallelizing outer individual fitness evaluation of multi-level nested genetic algorithm. Load balance and reading and writing of shared data are designed. The serial program is parallelized and optimized using Thread Building Blocks which is brought out by Intel. In the implementation process, parallel grain size and load balance are realized, scalable memory allocator and task scheduler are respectively used to allocate the memory and schedule parallel tasks.At the last, according to the actual operation situation, steel plate stacking data is simulated and testing data is designed. These data are used to test the efficiency of serial program and parallel program. By comparing the results, the efficiency of making pick-up operation schedule is improved effectively by parallel program. The timeliness requirements of shipbuilding farms are satisfied. A useful solution is provided to improve the efficiency of making pick-up scheduling of steel plate.