Aiming at algorithm execution time short but can not get good result problem of heuristic algorithm at solving scheduling, put forward a constructive-optimizer neural network (NN) algorithm based on object increment. Firstly, a dummy job with 0 processing time at any machine was introduced, a matrix denoting minimum finish times of adjacent jobs was created and the scheduling problem can then be transferred into a TSP problem, so the scheduling problem was mapped onto a constructive-optimizer neural network (NN). The simulation results based on the well known benchmarks show that the proposal outperforms the best know methods: SA2, RAJ, GR and objective increment method in most cases especially for large scale problems.