Niching Estimation of Distribution Algorithm Based on Fuzzy Clustering for Multi-mode Resource-constrained Project Scheduling Problems

: This paper proposes a novel niching estimation of distribution algorithm (EDA) based on fuzzy c-means (FCM) clustering for solving the multi-mode resource-constrained project scheduling problem (MRCPSP). FCM clustering is employed to partition the population into niches to avoid premature convergence of the EDA. Then, the niche capacity is determined by Boltzmann scheme according to the adaptive clearing radius and niche fitness. Besides, a restart up policy is employed based on a new diversity measure. A random walk local search is applied based on delete-then-insert operator (DIRW) to achieve a trade-off between exploration and exploitation abilities. The proposed algorithm is tested and evaluated using benchmark test problems of the project scheduling problem library PSPLIB, and compared with the standard EDA algorithm. Simulation results demonstrate the effectiveness of the proposed algorithm, and outperform the standard EDA.