In this paper, we intend to offer a new method based on fuzzy neural networks for finding a real solution of fuzzy equations system. Our proposed fuzzified neural network is a five-layer feed-back neural network that corresponding connection weights to output layer are fuzzy numbers. The proposed architecture of artificial neural network, can get a real input vector and calculates it's corresponding fuzzy output. In order to find the approximate solution of this fuzzy system that supposedly has a real solution, first a cost function is defined for the level sets of fuzzy output and target output. Then a learning algorithm based on the gradient descent method will be introduced that can adjust the crisp input signals. The proposed method is illustrated by several examples with computer simulations.