Predictive model of Mn-Si alloy smelting energy consumption based on genetic neural network

To avoid the BP (Back-Propagation) Network´s disadvantages of low training speed, prone to trapping in a local optimum and poor capability of global search, this paper establishes the model of manual neural network energy prediction system based on generic algorithm with the research on the Mn-Si alloy smelting of a steel company, by optimizing the initialized weights and threshold of neural network with GA. After the test of the program complied by MATLAB language and the comparison with pure BP algorithm, the results show that the methods suggested by this paper improve both the accuracy of predicting and the rate of convergence.