This paper addresses the multi-objective unrelated parallel machines to minimize makespan, total weighted completion times, and total weighted tardiness simultaneously. To tackle this problem, a hybrid algorithm based on reference-based many objective NSGA-II is proposed that the initial population is generated based on an exact algorithm and multi-directional local search algorithm. Minimizing makespan of the problem using relaxed mathematical model can be solved in an affordable computational time. Then, the multi-directional local search algorithm uses the high quality solution of the exact method to generate the non-dominate solutions by performing greedy search algorithms which is specifically designed for each objective of the addressed problem. The generated high quality solutions are used as the initial solution of the reference-based many objective NSGA-II. The performance of the algorithm is tested based on the well-known benchmark test problems and performance of the proposed algorithm is compared to the-state-of-the-art.