Abstract: In order to predict electromechanical equipmentsnonlinear and non-stationary condition effectively, the method of chaos prediction and the prediction theory based on chaos-neural networks are introduced, and the model of chaos-neural networks is set up. Aimed at the industrial smokes and gas turbine, the paper finished the prediction based on the chaos-neural networks and gray predicting method, the two prediction results are compared. The compared result shows that the prediction based on the chaos-neural networks has a higher accuracy and it can forecast the fault more effective.