In this paper the state estimation problem including hard detection errors are solved using combinatorial optimization algorithms. A novel procedure that combines the classic estate estimation methodology with leverage points theory which are used like sensibility factors and the observability theory are used to penalize the infeasibility in the objective function. The resultant model is solved using several optimization techniques like Tabu Search, Simulated Annealing, Particle Swarm, and the modified genetic algorithm of Chu-Beasley. In order to prove the proposed methodology the 57 nodes IEEE test system is used. The results obtained presents excellent quality.