Optimization of CFBB by orthogonal analysis based on artificial neural network

To solve the problem of analyzing and optimizing the impact factors which contribute the optimal goal, orthogonal analysis based on artificial neural network (ANN) was put forward. By gathering the data of impact factors and optimal goal from industrial spot, orthogonal experiment analysis was preceded. Because it is hard to get the data of optimal goal which are set on the orthogonal layout, the predictive model based on ANN was established first. According to the model, with appropriate orthogonal layout, the orthogonal experiment could be applied to analyze the contribution of different impact factors. A 135ton/h circulating fluidized bed boiler unit in Fujian was tested in this paper. By using the spot data, the LM_BP-based predictive model was established, and then with six factors, four conditions and the L₃₂(4⁹)orthogonal layout, the orthogonal experiment analysis was used. The test result indicated that the temperature of boiler, the pressure of gas and the ratio of wind and coal are key factors and the result was validated by the industrial optimal experiment. This method will help optimal operation of the Boiler.