A Self-Adaptive Differential Evolution Algorithm Based on Model Transformation for Flexible Job-Shop Scheduling Problem with Lot Streaming

As the globalization continues to advance, the econ-omy of countries all over the world is greatly influenced. At the same time, the increasing level of customization leads to smaller production batches, more frequent changes, and higher material losses in manufacturing industry. As a result, lot streaming is widely used in production and manufacture. This article address-es the flexible job-shop scheduling problem with lot streaming (FJSP-LS). A self-adaptive differential evolution algorithm based on model transformation (SDEA-MT) is presented. First, in order to generate diverse population with high quality, two heuristics are employed cooperatively for hybrid initialization. Second, the mathematical model is converted into continuous mode based on a specially designed transformation scheme. Third, a probability-based mutation method and a problem-specific crossover strategy are designed cooperatively to generate better solutions. Forth, a local search method is implemented to balance the exploration and exploitation. The effects of parameter setting is investigated through extensive computational tests. The competitive results demonstrate the effectiveness of every special design and the efficiency of SDEA-MT.