Compact estimation of distribution algorithm for semiconductor final testing scheduling problem

The semiconductor final testing scheduling problem (SFTSP) is crucial to the efficiency of the semiconductor manufacturing process. However, it is difficult to solve the problem in a short time due to its complexity. This paper is an attempt to present a simple and effective method called the compact estimation of distribution algorithm (cEDA) for solving the SFTSP. According to the characteristics of the problem, the cEDA employs the permutation based encoding and decoding schemes to obtain a solution of the SFTSP as well as its makespan value. In addition, it builds a well-designed probability model for describing the distribution of the solution space. In each generation, the probability model is sampled to generate two new solutions and updated with the superior one. With the low complexity and comparatively fewer parameters to set, the cEDA is simple and efficient. Simulation results based on benchmark instances and comparisons with some existing algorithms demonstrate the effectiveness of the proposed algorithm.